1 systemd System and Service Manager
3 CHANGES WITH 256 in spe:
5 Announcements of Future Feature Removals and Incompatible Changes:
7 * Support for automatic flushing of the nscd user/group database caches
8 will be dropped in a future release.
10 * Support for cgroup v1 ('legacy' and 'hybrid' hierarchies) is now
11 considered obsolete and systemd by default will refuse to boot under
12 it. To forcibly reenable cgroup v1 support,
13 SYSTEMD_CGROUP_ENABLE_LEGACY_FORCE=1 must be set on kernel command
14 line. The meson option 'default-hierarchy=' is also deprecated, i.e.
15 only cgroup v2 ('unified' hierarchy) can be selected as build-time
18 * Previously, systemd-networkd did not explicitly remove any bridge
19 VLAN IDs assigned on bridge master and ports. Since version 256, if a
20 .network file for an interface has at least one valid setting in the
21 [BridgeVLAN] section, then all assigned VLAN IDs on the interface
22 that are not configured in the .network file are removed.
24 * systemd-gpt-auto-generator will stop generating units for ESP or
25 XBOOTLDR partitions if it finds mount entries for or below the /boot/
26 or /efi/ hierarchies in /etc/fstab. This is to prevent the generator
27 from interfering with systems where the ESP is explicitly configured
28 to be mounted at some path, for example /boot/efi/ (this type of
29 setup is obsolete but still commonly found).
31 * The behavior of systemd-sleep and systemd-homed has been updated to
32 freeze user sessions when entering the various sleep modes or when
33 locking a homed-managed home area. This is known to cause issues with
34 the proprietary NVIDIA drivers. Packagers of the NVIDIA proprietary
35 drivers may want to add drop-in configuration files that set
36 SYSTEMD_SLEEP_FREEZE_USER_SESSION=false for systemd-suspend.service
37 and related services, and SYSTEMD_HOME_LOCK_FREEZE_SESSION=false for
38 systemd-homed.service.
40 * systemd-tmpfiles and systemd-sysusers, when given a relative
41 configuration file path (with at least one directory separator '/'),
42 will open the file directly, instead of searching for the given
43 partial path in the standard locations. The old mode wasn't useful
44 because tmpfiles.d/ and sysusers.d/ configuration has a flat
45 structure with no subdirectories under the standard locations and
46 this change makes it easier to work with local files with those
49 * systemd-tmpfiles now properly applies nested configuration to 'R' and
50 'D' stanzas. For example, with the combination of 'R /foo' and 'x
51 /foo/bar', /foo/bar will now be excluded from removal.
53 General Changes and New Features:
55 * Various programs will now attempt to load the main configuration file
56 from locations below /usr/lib/, /usr/local/lib/, and /run/, not just
57 below /etc/. For example, systemd-logind will look for
58 /etc/systemd/logind.conf, /run/systemd/logind.conf,
59 /usr/local/lib/systemd/logind.conf, and /usr/lib/systemd/logind.conf,
60 and use the first file that is found. This means that the search
61 logic for the main config file and for drop-ins is now the same.
63 Similarly, kernel-install will look for the config files in
64 /usr/lib/kernel/ and the other search locations, and now also
67 systemd-udevd now supports drop-ins for udev.conf.
69 * A new 'systemd-vpick' binary has been added. It implements the new
70 vpick protocol, where a "*.v/" directory may contain multiple files
71 which have versions (following the UAPI version format specification)
72 embedded in the file name. The files are ordered by version and
73 the newest one is selected.
75 systemd-nspawn --image=/--directory=, systemd-dissect, and the
76 RootDirectory=, RootImage=, ExtensionImages=, and
77 ExtensionDirectories= settings for units now support the vpick
78 protocol and allow the latest version to be selected automatically if
79 a "*.v/" directory is specified as the source.
81 * Encrypted service credentials may now be made accessible to
82 unprivileged users. systemd-creds gained new options --user/--uid=
83 for encrypting/decrypting a credential for a specific user.
85 * New command-line tool 'importctl' to download, import, and export
86 disk images via systemd-importd is added with the following verbs:
87 pull-tar, pull-raw, import-tar, import-raw, import-fs, export-tar,
88 export-raw, list-transfers, cancel-transfer. This functionality was
89 previously available in "machinectl", where it was exclusively for
90 machine image. The new "importctl" generalizes this for sysext,
91 confext, portable service images, too.
95 * New system manager setting ProtectSystem= has been added. It is
96 analogous to the unit setting, but applies to the whole system. It is
97 enabled by default in the initrd.
99 * New unit setting WantsMountsFor= has been added. It is analogous to
100 RequiresMountsFor=, but with a Wants= dependency instead of
101 Requires=. This new logic is used in various places where mounts were
102 added as dependencies for other settings (WorkingDirectory=-…,
103 PrivateTmp=yes, cryptsetup lines with 'nofail').
105 * New unit setting MemoryZSwapWriteback= can be used to control the new
106 memory.zswap.writeback cgroup knob added in kernel 6.8.
108 * The manager gained a org.freedesktop.systemd1.StartAuxiliaryScope()
109 D-Bus method to devolve some processes from a service into a new
110 scope. This new scope will remain even if the original service unit
111 is restarted. Control group properties of the new scope are copied
112 from the originating unit, so various limits are retained.
114 * Units now expose properties EffectiveMemoryMax=,
115 EffectiveMemoryHigh=, and EffectiveTasksMax=, which report the
116 most stringent limit systemd is aware of for the given unit.
118 * A new unit file specifier %D expands to $XDG_DATA_HOME (for user
119 services) or /usr/share/ (for system services).
121 * AllowedCPUs= now supports specifier expansion.
123 * What= setting in .mount and .swap units now accepts fstab-style
124 identifiers, for example UUID=… or LABEL=….
126 * RestrictNetworkInterfaces= now supports alternative network interface
129 * PAMName= now implies SetLoginEnvironment=yes.
131 * systemd.firstboot=no can be used on the kernel command-line to
132 disable interactive queries, but allow other first boot configuration
133 to happen based on credentials.
135 * The system's hostname can be configured via the systemd.hostname
138 * The systemd binary will no longer chainload sysvinit's "telinit"
139 binary when called under the init/telinit name on a system that
140 isn't booted with systemd. This previously has been supported to make
141 sure a distribution that has both init systems installed can be
142 reasonably switched from one to the other via a simple
143 reboot. Distributions apparently have lost interest in this, and the
144 functionality has not been supported on the primary distribution this
145 was still intended for for a longer time, and hence has been removed
148 * A new concept called "capsules" has been introduced. "Capsules"
149 encapsulate additional per-user service managers, whose users are
150 transient and are only defined as long as the service manager
151 is running (implemented via DynamicUser=1). These service managers run
152 off home directories defined in /var/lib/capsules/<name>, where
153 <name> is a the capsule's name. These home directories can contain
154 regular per-user services and other units. A capsule is started via a
155 simple "systemctl start capsule@<name>.service". See the
156 capsule@.service(5) man page for further details. Various systemd
157 tools (including, and most importantly, systemctl and systemd-run)
158 have been updated to interact with capsules via the new
159 "--capsule="/"-C" switch.
161 * .socket units gained a new setting PassFileDescriptorsToExec=, taking
162 a boolean value. If set to true the file descriptors the socket unit
163 encapsulates are passed to the ExecStartPost=, ExecStopPre=,
164 ExecStopPost= using the usual $LISTEN_FDS interface. This may be used
165 for doing additional initializations on the sockets once they are
166 allocated (for example, install an additional eBPF program on them).
168 * The .socket setting MaxConnectionsPerSource= (which so far put a
169 limit on concurrent connections per IP in Accept=yes socket units),
170 now also has an effect on AF_UNIX sockets: it will put a limit on the
171 number of simultaneous connections from the same source UID (as
172 determined via SO_PEERCRED). This is useful for implementing IPC
173 services in a simple Accept=yes mode.
175 * The service manager will now maintain a counter of soft reboot cycles
176 the system went through so far. It may be queried via the D-Bus APIs.
178 * systemd's execution logic now supports the new pidfd_spawn() API
179 introduced by glibc 2.39, which allows us to invoke a subprocess in a
180 target cgroup and get a pidfd back in a single operation.
182 * systemd/PID 1 will now send an additional sd_notify() message to its
183 supervising VMM or container manager reporting the selected hostname
184 ("X_SYSTEMD_HOSTNAME=") and machine ID ("X_SYSTEMD_MACHINE_ID=") at
185 boot. Moreover, the service manager will send additional sd_notify()
186 messages ("X_SYSTEMD_UNIT_ACTIVE=") whenever a target unit is
187 reached. This can be used by VMMs/container managers to schedule
188 access to the system precisely. For example, the moment a system
189 reports "ssh-access.target" being reached a VMM/container manager
190 knows it can now connect to the system via SSH. Finally, a new
191 sd_notify() message ("X_SYSTEMD_SIGNALS_LEVEL=2") is sent the moment
192 PID 1 successfully completed installation of its various UNIX process
193 signal handlers (i.e. the moment where SIGRTMIN+4 sent to PID 1 will
194 start to have the effect of shutting down the system cleanly).
198 * systemd-journald can now forward journal entries to a socket
199 (AF_INET, AF_INET6, AF_UNIX, or AF_VSOCK). The socket can be
200 specified in journald.conf via a new option ForwardAddress= or via
201 the 'journald.forward_address' credential. Log records are sent in
202 the Journal Export Format. A related setting MaxLevelSocket= has been
203 added to control the maximum log levels for the messages sent to this
206 * systemd-vmspawn gained a new --forward-journal= option to forward the
207 virtual machine's journal entries to the host. This is done over a
208 AF_VSOCK socket, i.e. it does not require networking in the guest.
210 * journalctl gained option '-i' as a shortcut for --file=.
212 * journalctl gained a new -T/--exclude-identifier= option to filter
213 out certain syslog identifiers.
215 * journalctl gained a new --list-namespaces option.
217 * systemd-journal-remote now also accepts AF_VSOCK and AF_UNIX sockets
218 (so it can be used to receive entries forwarded by systemd-journald).
220 * systemd-journal-gatewayd allows restricting the time range of
221 retrieved entries with a new "realtime=[<since>]:[<until>]" URL
224 * systemd-cat gained a new option --namespace= to specify the target
225 journal namespace to which the output shall be connected.
227 * systemd-bsod gained a new option --tty= to specify the output TTY
231 * /dev/ now contains symlinks that combine by-path and by-{label,uuid}
234 /dev/disk/by-path/<path>/by-<label|uuid|…>/<label|uuid|…>
236 This allows distinguishing partitions with identical contents on
237 multiple storage devices. This is useful, for example, when copying
238 raw disk contents between devices.
240 * systemd-udevd now creates persistent /dev/media/by-path/ symlinks for
241 media controllers. For example, the uvcvideo driver may create
242 /dev/media0 which will be linked as
243 /dev/media/by-path/pci-0000:04:00.3-usb-0:1:1.0-media-controller.
245 * An allowlist/denylist may be specified to filter which sysfs
246 attributes are used when crafting network interface names. Those
247 lists are stored as hwdb entries
248 ID_NET_NAME_ALLOW_<sysfsattr>=0|1
250 ID_NET_NAME_ALLOW=0|1.
252 The goal is to avoid unexpected changes to interface names when the
253 kernel is updated and new sysfs attributes become visible.
255 * A new unit tpm2.target has been added to provide a synchronization
256 point for units which expect the TPM hardware to be available. A new
257 generator "systemd-tpm2-generator" has been added that will insert
258 this target whenever it detects that the firmware has initialized a
259 TPM, but Linux hasn't loaded a driver for it yet.
261 * systemd-backlight now properly supports numbered devices which the
262 kernel creates to avoid collisions in the leds subsystem.
264 * systemd-hwdb update operation can be disabled with a new environment
265 variable SYSTEMD_HWDB_UPDATE_BYPASS=1.
269 * systemd-hostnamed now exposes the machine ID and boot ID via
270 D-Bus. It also exposes the hosts AF_VSOCK CID, if available.
272 * systemd-hostnamed now provides a basic Varlink interface.
274 * systemd-hostnamed exports the full data in os-release(5) and
275 machine-info(5) via D-Bus and Varlink.
277 * hostnamectl now shows the system's product UUID and hardware serial
282 * systemd-networkd now provides a basic Varlink interface.
284 * systemd-networkd's ARP proxy support gained a new option to configure
285 a private VLAN variant of the proxy ARP supported by the kernel under
286 the name IPv4ProxyARPPrivateVLAN=.
288 * systemd-networkd now exports the NamespaceId and NamespaceNSID
289 properties via D-Bus and Varlink. (which expose the inode and NSID of
290 the network namespace the networkd instance manages)
292 * systemd-networkd now supports IPv6RetransmissionTimeSec= and
293 UseRetransmissionTime= settings in .network files to configure
294 retransmission time for IPv6 neighbor solicitation messages.
296 * networkctl gained new verbs 'mask' and 'unmask' for masking networkd
297 configuration files such as .network files.
299 * 'networkctl edit --runtime' allows editing volatile configuration
300 under /run/systemd/network/.
302 * The implementation behind TTLPropagate= network setting has been
303 removed and the setting is now ignored.
305 * systemd-network-generator will now pick up .netdev/.link/.network
306 configuration from system credentials.
308 * systemd-networkd will now pick up wireguard secrets from
311 * systemd-networkd's Varlink API now supports enumerating LLDP peers.
313 * .link files now support new Property=, ImportProperty=,
314 UnsetProperty= fields for setting udev properties on a link.
316 * The various .link files that systemd ships for interfaces that are
317 supposed to be managed by systemd-networkd only now carry a
318 ID_NET_MANAGED_BY=io.systemd.Network udev property ensuring that
319 other network management solutions honouring this udev property do
320 not come into conflict with networkd, trying to manage these
323 * .link files now support a new ReceivePacketSteeringCPUMask= setting
324 for configuring which CPUs to steer incoming packets to.
328 * systemd-nspawn now provides a /run/systemd/nspawn/unix-export/
329 directory where the container payload can expose AF_UNIX sockets to
330 allow them them to be accessed from outside.
332 * systemd-nspawn will tint the terminal background for containers in a
333 blueish color. This can be controller with the new --background=
336 * systemd-nspawn gained support for the 'owneridmap' option for --bind=
337 mounts to map the target directory owner from inside the container to
338 the owner of the directory bound from the host filesystem.
340 * systemd-nspawn now supports moving Wi-Fi network devices into a
341 container, just like other network interfaces.
345 * systemd-resolved now reads RFC 8914 EDE error codes provided by
346 upstream DNS services.
348 * systemd-resolved and resolvectl now support RFC 9460 SVCB and HTTPS
349 records, as well as RFC 2915 NAPTR records.
351 * resolvectl gained a new option --relax-single-label= to allow
352 querying single-label hostnames via unicast DNS on a per-query basis.
354 * systemd-resolved's Varlink IPC interface now supports resolving
355 DNS-SD services as well as an API for resolving raw DNS RRs.
357 * systemd-resolved's .dnssd DNS_SD service description files now
358 support DNS-SD "subtypes" via the new SubType= setting.
360 * systemd-resolved's configuration may now be reloaded without
361 restarting the service. (i.e. "systemctl reload systemd-resolved" is
366 * An sshd config drop-in to allow ssh keys acquired via userdbctl (for
367 example expose by homed accounts) to be used for authorization of
368 incoming SSH connections.
370 * A small new unit generator "systemd-ssh-generator" has been added. It
371 checks if the sshd binary is installed. If so, it binds it via
372 per-connection socket activation to various sockets depending on the
375 • If the system is run in a VM providing AF_VSOCK support, it
376 automatically binds sshd to AF_VSOCK port 22.
378 • If the system is invoked as a full-OS container and the container
379 manager pre-mounts a directory /run/host/unix-export/, it will
380 bind sshd to an AF_UNIX socket /run/host/unix-export/ssh. The
381 idea is the container manager bind mounts the directory to an
382 appropriate place on the host as well, so that the AF_UNIX socket
383 may be used to easily connect from the host to the container.
385 • sshd is also bound to an AF_UNIX socket
386 /run/ssh-unix-local/socket, which may be to use ssh/sftp in a
387 "sudo"-like fashion to access resources of other local users.
389 • Via the kernel command line option "systemd.ssh_listen=" and the
390 system credential "ssh.listen" sshd may be bound to additional,
391 explicitly configured options, including AF_INET/AF_INET6 ports.
393 In particular the first two mechanisms should make dealing with local
394 VMs and full OS containers a lot easier, as SSH connections will
395 *just* *work* from the host – even if no networking is available
398 systemd-ssh-generator optionally generates a per-connection
399 socket activation service file wrapping sshd. This is only done if
400 the distribution does not provide one on its own under the name
401 "sshd@.service". The generated unit only works correctly if the SSH
402 privilege separation ("privsep") directory exists. Unfortunately
403 distributions vary wildly where they place this directory. An
404 incomprehensive list:
406 • /usr/share/empty.sshd/ (new fedora)
409 • /run/sshd/ (debian/ubuntu?)
411 If the SSH privsep directory is placed below /var/ or /run/ care
412 needs to be taken that the directory is created automatically at boot
413 if needed, since these directories possibly or always come up
414 empty. This can be done via a tmpfiles.d/ drop-in. You may use the
415 "sshdprivsepdir" meson option provided by systemd to configure the
416 directory, in case you want systemd to create the directory as needed
417 automatically, if your distribution does not cover this natively.
419 Recommendations to distributions, in order to make things just work:
421 • Please provide a per-connection SSH service file under the name
424 • Please move the SSH privsep dir into /usr/ (so that it is truly
425 immutable on image-based operating systems, is strictly under
426 package manager control, and never requires recreation if the
427 system boots up with an empty /run/ or /var/).
429 • As an extension of this: please consider following Fedora's lead
430 here, and use /usr/share/empty.sshd/ to minimize needless
431 differences between distributions.
433 • If your distribution insists on placing the directory in /var/ or
434 /run/ then please at least provide a tmpfiles.d/ drop-in to
435 recreate it automatically at boot, so that the sshd binary just
436 works, regardless in which context it is called.
438 * A small tool "systemd-ssh-proxy" has been added, which is supposed to
439 act as counterpart to "systemd-ssh-generator". It's a small plug-in
440 for the SSH client (via ProxyCommand/ProxyUseFdpass) to allow it to
441 connect to AF_VSOCK or AF_UNIX sockets. Example: "ssh vsock/4711"
442 connects to a local VM with cid 4711, or "ssh
443 unix/run/ssh-unix-local/socket" to connect to the local host via the
444 AF_UNIX socket /run/ssh-unix-local/socket.
446 systemd-boot and systemd-stub and Related Tools:
448 * TPM 1.2 PCR measurement support has been removed from systemd-stub.
449 TPM 1.2 is obsolete and – due to the (by today's standards) weak
450 cryptographic algorithms it only supports – does not actually provide
451 the security benefits it's supposed to provide. Given that the rest
452 of systemd's codebase never supported TPM 1.2, the support has now
453 been removed from systemd-stub as well.
455 * systemd-stub will now measure its payload via the new EFI
456 Confidential Computing APIs (CC), in addition to the pre-existing
459 * confexts are loaded by systemd-stub from the ESP as well.
461 * The pcrlock policy is saved in an unencrypted credential file
462 "pcrlock.<entry-token>.cred" under XBOOTLDR/ESP in the
463 /loader/credentials/ directory. It will be picked up at boot by
464 systemd-stub and passed to the initrd, where it can be used to unlock
465 the root file system.
467 * kernel-install gained support for --root= for the 'list' verb.
469 * systemd-pcrlock gained an --entry-token= option to configure the
472 * systemd-pcrlock now provides a basic Varlink interface and can be run
473 as a daemon via a template unit.
475 * bootctl now provides a basic Varlink interface and can be run as a
476 daemon via a template unit.
478 * systemd-measure gained new options --certificate=, --private-key=,
479 and --private-key-source= to allow using OpenSSL's "engines" or
480 "providers" as the signing mechanism to use when creating signed
481 TPM2 PCR measurement values.
483 * ukify gained support for signing of PCR signatures via OpenSSL's
484 engines and providers.
486 * ukify now supports zboot kernels.
488 * systemd-boot now supports passing additional kernel command line
489 switches to invoked kernels via an SMBIOS Type #11 string
490 "io.systemd.boot.kernel-cmdline-extra". This is similar to the
491 pre-existing support for this in systemd-stub, but also applies to
492 Type #1 Boot Loader Specification Entries.
494 * systemd-boot's automatic SecureBoot enrollment support gained support
495 for enrolling "dbx" too (Previously, only db/KEK/PK enrollment was
496 supported). It also now supports UEFI "Custom" mode.
500 * systemd-run is now a multi-call binary. When invoked as 'run0', it
501 provides as interface similar to 'sudo', with all arguments starting
502 at the first non-option parameter being treated the command to invoke
503 as root. Unlike 'sudo' and similar tools, it does not make use of
504 setuid binaries or other privilege escalation methods, but instead
505 runs the specified command as a transient unit, which is started by
506 the system service manager, so privileges are dropped, rather than
507 gained, thus implementing a much more robust and safe security
508 model. As usual, authorization is managed via Polkit.
510 * systemd-run/run0 will now tint the terminal background on supported
511 terminals: in a reddish tone when invoking a root service, in a
512 yellowish tone otherwise. This may be controlled and turned off via
513 the new --background= switch.
515 * systemd-run gained a new option '--ignore-failure' to suppress
520 * 'systemctl edit --stdin' allows creation of unit files and drop-ins
521 with contents supplied via standard input. This is useful when creating
522 configuration programmatically; the tool takes care of figuring out
523 the file name, creating any directories, and reloading the manager
526 * 'systemctl disable --now' and 'systemctl mask --now' now work
527 correctly with template units.
529 * 'systemd-analyze architectures' lists known CPU architectures.
531 * 'systemd-analyze --json=…' is supported for 'architectures',
532 'capability', 'exit-status'.
534 * 'systemd-tmpfiles --purge' will purge (remove) all files and
535 directories created via tmpfiles.d configuration.
537 * systemd-id128 gained new options --no-pager, --no-legend, and
540 * hostnamectl gained '-j' as shortcut for '--json=pretty' or
543 * loginctl now supports -j/--json=.
545 * resolvectl now supports -j/--json= for --type=.
547 * systemd-tmpfiles gained a new option --dry-run to print what would be
548 done without actually taking action.
550 * varlinkctl gained a new --collect switch to collect all responses of
551 a method call that supports multiple replies and turns it into a
554 * systemd-dissect gained a new --make-archive option to generate an
555 archive file (tar.gz and similar) from a disk image.
559 * systemd-vmspawn gained a new --firmware= option to configure or list
560 firmware definitions for Qemu, a new --tpm= option to enable or
561 disable the use of a software TPM, a new --linux= option to specify a
562 kernel binary for direct kernel boot, a new --initrd= option to
563 specify an initrd for direct kernel boot, a new -D/--directory option
564 to use a plain directory as the root file system, a new
565 --private-users option similar to the one in systemd-nspawn, new
566 options --bind= and --bind-ro= to bind part of the host's file system
567 hierarchy into the guest, a new --extra-drive= option to attach
568 additional storage, and -n/--network-tap/--network-user-mode to
569 configure networking.
571 * A new systemd-vmspawn@.service can be used to launch systemd-vmspawn
574 * systemd-vmspawn gained the new --console= and --background= switches
575 that control how to interact with the VM. As before, by default an
576 interactive terminal interface is provided, but now with a background
577 tinted with a greenish hue.
579 * systemd-vmspawn can now register its VMs with systemd-machined,
580 controlled via the --register= switch.
582 * machinectl's start command (and related) can now invoke images either
583 as containers via `systemd-nspawn` (switch is --runner=nspawn, the
584 default) or as VMs via `systemd-vmspawn` (switch is --runner=vmspawn,
587 * systemd-vmspawn now supports two switches --pass-ssh-key= and
588 --ssh-key-type= to optionally set up transient SSH keys to pass to the
589 invoked VMs in order to be able to SSH into them once booted.
593 * systemd-repart gained new options --generate-fstab= and
594 --generate-crypttab= to write out fstab and crypttab files matching the
595 generated partitions.
597 * systemd-repart gained a new option --private-key-source= to allow
598 using OpenSSL's "engines" or "providers" as the signing mechanism to
599 use when creating verity signature partitions.
601 * systemd-repart gained a new DefaultSubvolume= setting in repart.d/
602 drop-ins that allow configuring the default btrfs subvolume for newly
603 formatted btrfs file systems.
607 * libsystemd gained new call sd_bus_creds_new_from_pidfd() to get a
608 credentials object for a pidfd and sd_bus_creds_get_pidfd_dup() to
609 retrieve the pidfd from a credentials object.
611 * sd-bus' credentials logic will now also acquire peer's UNIX group
612 lists and peer's pidfd if supported and requested.
614 * RPM macro %_kernel_install_dir has been added with the path
615 to the directory for kernel-install plugins.
617 * The liblz4, libzstd, liblzma, libkmod, libgcrypt dependencies have
618 been changed from regular shared library dependencies into dlopen()
621 * The sd-journal API gained a new call
622 sd_journal_stream_fd_with_namespace() which is just like
623 sd_journal_stream_fd() but creates a log stream targeted at a
624 specific log namespace.
626 systemd-cryptsetup/systemd-cryptenroll:
628 * systemd-cryptenroll can now enroll directly with a PKCS11 public key
629 (instead of a certificate).
631 * systemd-cryptsetup/systemd-cryptenroll now may lock a disk against a
632 PKCS#11 provided EC key (before it only supported RSA).
634 * systemd-cryptsetup gained support for crypttab option
635 link-volume-key= to link the volume key into the kernel keyring when
636 the volume is opened.
638 * systemd-cryptenroll will no longer enable Dictionary Attack
639 Protection (i.e. turn on NO_DA) for TPM enrollments that do not
640 involve a PIN. DA should not be necessary in that case (since key
641 entropy is high enough to make this unnecessary), but risks
642 accidental lock-out in case of unexpected PCR changes.
644 * systemd-cryptenroll now supports enrolling a new slot while unlocking
645 the old slot via TPM2 (previously unlocking only worked via password
650 * The remaining documentation that was on
651 https://freedesktop.org/wiki/Software/systemd/ has been moved to
654 * A new text describing the VM integration interfaces of systemd has
657 https://systemd.io/VM_INTERFACE
659 * The sd_notify() man page has gained an example with C code that shows
660 how to implement the interface in C without involving libsystemd.
662 systemd-homed, systemd-logind, systemd-userdbd:
664 * systemd-homed now supports unlocking of home directories when logging
665 in via SSH. Previously home directories needed to be unlocked before
666 an SSH login is attempted.
668 * JSON User Records have been extended with a separate public storage
669 area called "User Record Blob Directories". This is intended to store
670 the user's background image, avatar picture, and other similar items
671 which are too large to fit into the User Record itself.
673 systemd-homed, userdbctl, and homectl gained support for blob
674 directories. homectl gained --avatar= and --login-background= to
675 control two specific items of the blob directories.
677 * A new "additionalLanguages" field has been added to JSON user records
678 (as supported by systemd-homed and systemd-userdbd), which is closely
679 related to the pre-existing "preferredLanguage", and allows
680 specifying multiple additional languages for the user account. It is
681 used to initialize the $LANGUAGES environment variable when used.
683 * A new pair of "preferredSessionType" and "preferredSessionLauncher"
684 fields have been added to JSON user records, that may be used to
685 control which kind of desktop session to preferable activate on
688 * homectl gained a new verb 'firstboot', and a new
689 systemd-homed-firstboot.service unit uses this verb to create users
690 in a first boot environment, either from system credentials or by
691 querying interactively.
693 * systemd-logind now supports a new "background-light" session class
694 which does not pull in the user@.service unit. This is intended in
695 particular for lighter weight per-user cron jobs which do require any
696 per-user service manager to be around.
698 * The per-user service manager will now be tracked as a distinct "manager"
699 session type among logind sessions of each user.
701 * homectl now supports an --offline mode, by which certain account
702 properties can be changed without unlocking the home directory.
704 * systemd-logind gained a new
705 org.freedesktop.login1.Manager.ListSessionsEx() method that provides
706 additional metadata compared to ListSessions(). loginctl makes use of
707 this to list additional fields in list-sessions.
709 * systemd-logind gained a new org.freedesktop.login1.Manager.Sleep()
710 method that automatically redirects to SuspendThenHibernate(),
711 Suspend(), HybridSleep(), or Hibernate(), depending on what is
712 supported and configured, a new configuration setting SleepOperation=,
713 and an accompanying helper method
714 org.freedesktop.login1.Manager.CanSleep() and property
715 org.freedesktop.login1.Manager.SleepOperation.
717 'systemctl sleep' calls the new method to automatically put the
718 machine to sleep in the most appropriate way.
720 Credential Management:
722 * systemd-creds now provides a Varlink IPC API for encrypting and
723 decrypting credentials.
725 * systemd-creds' "tpm2-absent" key selection has been renamed to
726 "null", since that's what it actually does: "encrypt" and "sign"
727 with a fixed null key. --with-key=null should only be used in very
728 specific cases, as it provides zero integrity or confidentiality
729 protections. (i.e. it's only safe to use as fallback in environments
730 lacking both a TPM and access to the root fs to use the host
731 encryption key, or when integrity is provided some other way.)
733 * systemd-creds gained a new switch --allow-null. If specified, the
734 "decrypt" verb will decode encrypted credentials that use the "null"
735 key (by default this is refused, since using the "null" key defeats
736 the authenticated encryption normally done).
740 * The sleep.conf configuration file gained a new MemorySleepMode=
741 setting for configuring the sleep mode in more detail.
743 * A tiny new service systemd-hibernate-clear.service has been added
744 which clears hibernation information from the HibernateLocation EFI
745 variable, in case the resume device is gone. Normally, this variable
746 is supposed to be cleaned up by the code that initiates the resume
747 from hibernation image. But when the device is missing and that code
748 doesn't run, this service will now do the necessary work, ensuring
749 that no outdated hibernation image information remains on subsequent
752 Unprivileged User Namespaces & Mounts:
754 * A small new service systemd-nsresourced.service has been added. It
755 provides a Varlink IPC API that assigns a free, transiently allocated
756 64K UID/GID range to an uninitialized user namespace a client
757 provides. It may be used to implement unprivileged container managers
758 and other programs that need dynamic user ID ranges. It also provides
759 interfaces to then delegate mount file descriptors, control groups
760 and network interfaces to user namespaces set up this way.
762 * A small new service systemd-mountfsd.service has been added. It
763 provides a Varlink IPC API for mounting DDI images, and returning a set
764 of mount file descriptors for it. If a user namespace fd is provided
765 as input, then the mounts are registered with the user namespace. To
766 ensure trust in the image it must provide Verity information (or
767 alternatively interactive polkit authentication is required).
769 * The systemd-dissect tool now can access DDIs fully unprivileged by
770 using systemd-nsresourced/systemd-mountfsd.
772 * If the service manager runs unprivileged (i.e. systemd --user) it now
773 supports RootImage= for accessing DDI images, also implemented via
774 the systemd-nsresourced/systemd-mountfsd.
776 * systemd-nspawn may now operate without privileges, if a suitable DDI
777 is provided via --image=, again implemented via
778 systemd-nsresourced/systemd-mountfsd.
782 * timedatectl and machinectl gained option '-P', an alias for
783 '--value --property=…'.
785 * Various tools that pretty-print config files will now highlight
786 configuration directives.
788 * varlinkctl gained support for the "ssh:" transport. This requires
789 OpenSSH 9.4 or newer.
791 * systemd-sysext gained support for enabling system extensions in
792 mutable fashion, where a writeable upperdir is stored under
793 /var/lib/extensions.mutable/, and a new --mutable= option to
794 configure this behaviour. An "ephemeral" mode is not also supported
795 where the mutable layer is configured to be a tmpfs that is
796 automatically released when the system extensions are reattached.
798 * Coredumps are now retained for two weeks by default (instead of three
801 * portablectl --copy= parameter gained a new 'mixed' argument, that will
802 result in resources owned by the OS (e.g.: portable profiles) to be linked
803 but resources owned by the portable image (e.g.: the unit files and the
804 images themselves) to be copied.
806 * systemd will now register MIME types for various of its file types
807 (e.g. journal files, DDIs, encrypted credentials …) via the XDG
808 shared-mime-info infrastructure. (Files of these types will thus be
809 recognized as their own thing in desktop file managers such as GNOME
812 * systemd-dissect will now show the detected sector size of a given DDI
813 in its default output.
815 * systemd-portabled now generates recognizable structured log messages
816 whenever a portable service is attached or detached.
818 * Verity signature checking in userspace (i.e. checking against
819 /etc/verity.d/ keys) when activating DDIs can now be turned on/off
820 via a kernel command line option systemd.allow_userspace_verity= and
821 an environment variable SYSTEMD_ALLOW_USERSPACE_VERITY=.
823 * ext4/xfs file system quota handling has been reworked, so that
824 quotacheck and quotaon are now invoked as per-file-system templated
825 services (as opposed to single system-wide singletons), similar in
826 style to the fsck, growfs, pcrfs logic. This means file systems with
827 quota enabled can now be reasonably enabled at runtime of the system,
830 * "systemd-analyze dot" will now also show BindsTo= dependencies.
832 * systemd-debug-generator gained the ability add in arbitrary units
833 based on them being passed in via system credentials.
835 * A new kernel command-line option systemd.default_debug_tty= can be
836 used to specify the TTY for the debug shell, independently of
837 enabling or disabling it.
841 Announcements of Future Feature Removals and Incompatible Changes:
843 * Support for split-usr (/usr/ mounted separately during late boot,
844 instead of being mounted by the initrd before switching to the rootfs)
845 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
846 /usr/lib/, …) has been removed. For more details, see:
847 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
849 * We intend to remove cgroup v1 support from a systemd release after
850 the end of 2023. If you run services that make explicit use of
851 cgroup v1 features (i.e. the "legacy hierarchy" with separate
852 hierarchies for each controller), please implement compatibility with
853 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
854 Most of Linux userspace has been ported over already.
856 * Support for System V service scripts is now deprecated and will be
857 removed in a future release. Please make sure to update your software
858 *now* to include a native systemd unit file instead of a legacy
859 System V script to retain compatibility with future systemd releases.
861 * Support for the SystemdOptions EFI variable is deprecated.
862 'bootctl systemd-efi-options' will emit a warning when used. It seems
863 that this feature is little-used and it is better to use alternative
864 approaches like credentials and confexts. The plan is to drop support
865 altogether at a later point, but this might be revisited based on
868 * systemd-run's switch --expand-environment= which currently is disabled
869 by default when combined with --scope, will be changed in a future
870 release to be enabled by default.
872 * "systemctl switch-root" is now restricted to initrd transitions only.
874 Transitions between real systems should be done with
875 "systemctl soft-reboot" instead.
877 * The "ip=off" and "ip=none" kernel command line options interpreted by
878 systemd-network-generator will now result in IPv6RA + link-local
879 addressing being disabled, too. Previously DHCP was turned off, but
880 IPv6RA and IPv6 link-local addressing was left enabled.
882 * The NAMING_BRIDGE_MULTIFUNCTION_SLOT naming scheme has been deprecated
885 * SuspendMode=, HibernateState= and HybridSleepState= in the [Sleep]
886 section of systemd-sleep.conf are now deprecated and have no effect.
887 They did not (and could not) take any value other than the respective
888 default. HybridSleepMode= is also deprecated, and will now always use
889 the 'suspend' disk mode.
893 * The way services are spawned has been overhauled. Previously, a
894 process was forked that shared all of the manager's memory (via
895 copy-on-write) while doing all the required setup (e.g.: mount
896 namespaces, CGroup configuration, etc.) before exec'ing the target
897 executable. This was problematic for various reasons: several glibc
898 APIs were called that are not supposed to be used after a fork but
899 before an exec, copy-on-write meant that if either process (the
900 manager or the child) touched a memory page a copy was triggered, and
901 also the memory footprint of the child process was that of the
902 manager, but with the memory limits of the service. From this version
903 onward, the new process is spawned using CLONE_VM and CLONE_VFORK
904 semantics via posix_spawn(3), and it immediately execs a new internal
905 binary, systemd-executor, that receives the configuration to apply
906 via memfd, and sets up the process before exec'ing the target
907 executable. The systemd-executor binary is pinned by file descriptor
908 by each manager instance (system and users), and the reference is
909 updated on daemon-reexec - it is thus important to reexec all running
910 manager instances when the systemd-executor and/or libsystemd*
911 libraries are updated on the filesystem.
913 * Most of the internal process tracking is being changed to use PIDFDs
914 instead of PIDs when the kernel supports it, to improve robustness
917 * A new option SurviveFinalKillSignal= can be used to configure the
918 unit to be skipped in the final SIGTERM/SIGKILL spree on shutdown.
919 This is part of the required configuration to let a unit's processes
920 survive a soft-reboot operation.
922 * System extension images (sysext) can now set
923 EXTENSION_RELOAD_MANAGER=1 in their extension-release files to
924 automatically reload the service manager (PID 1) when
925 merging/refreshing/unmerging on boot. Generally, while this can be
926 used to ship services in system extension images it's recommended to
927 do that via portable services instead.
929 * The ExtensionImages= and ExtensionDirectories= options now support
930 confexts images/directories.
932 * A new option NFTSet= provides a method for integrating dynamic cgroup
933 IDs into firewall rules with NFT sets. The benefit of using this
934 setting is to be able to use control group as a selector in firewall
935 rules easily and this in turn allows more fine grained filtering.
936 Also, NFT rules for cgroup matching use numeric cgroup IDs, which
937 change every time a service is restarted, making them hard to use in
940 * A new option CoredumpReceive= can be set for service and scope units,
941 together with Delegate=yes, to make systemd-coredump on the host
942 forward core files from processes crashing inside the delegated
943 CGroup subtree to systemd-coredump running in the container. This new
944 option is by default used by systemd-nspawn containers that use the
947 * A new ConditionSecurity=measured-uki option is now available, to ensure
948 a unit can only run when the system has been booted from a measured UKI.
950 * MemoryAvailable= now considers physical memory if there are no CGroup
951 memory limits set anywhere in the tree.
953 * The $USER environment variable is now always set for services, while
954 previously it was only set if User= was specified. A new option
955 SetLoginEnvironment= is now supported to determine whether to also set
956 $HOME, $LOGNAME, and $SHELL.
958 * Socket units now support a new pair of
959 PollLimitBurst=/PollLimitInterval= options to configure a limit on
960 how often polling events on the file descriptors backing this unit
961 will be considered within a time window.
963 * Scope units can now be created using PIDFDs instead of PIDs to select
964 the processes they should include.
966 * Sending SIGRTMIN+18 with 0x500 as sigqueue() value will now cause the
967 manager to dump the list of currently pending jobs.
969 * If the kernel supports MOVE_MOUNT_BENEATH, the systemctl and
970 machinectl bind and mount-image verbs will now cause the new mount to
971 replace the old mount (if any), instead of overmounting it.
973 * Units now have MemoryPeak, MemorySwapPeak, MemorySwapCurrent and
974 MemoryZSwapCurrent properties, which respectively contain the values
975 of the cgroup v2's memory.peak, memory.swap.peak, memory.swap.current
976 and memory.zswap.current properties. This information is also shown in
977 "systemctl status" output, if available.
979 TPM2 Support + Disk Encryption & Authentication:
981 * systemd-cryptenroll now allows specifying a PCR bank and explicit hash
982 value in the --tpm2-pcrs= option.
984 * systemd-cryptenroll now allows specifying a TPM2 key handle (nv
985 index) to be used instead of the default SRK via the new
986 --tpm2-seal-key-handle= option.
988 * systemd-cryptenroll now allows TPM2 enrollment using only a TPM2
989 public key (in TPM2B_PUBLIC format) – without access to the TPM2
990 device itself – which enables offline sealing of LUKS images for a
991 specific TPM2 chip, as long as the SRK public key is known. Pass the
992 public to the tool via the new --tpm2-device-key= switch.
994 * systemd-cryptsetup is now installed in /usr/bin/ and is no longer an
995 internal-only executable.
997 * The TPM2 Storage Root Key will now be set up, if not already present,
998 by a new systemd-tpm2-setup.service early boot service. The SRK will
999 be stored in PEM format and TPM2_PUBLIC format (the latter is useful
1000 for systemd-cryptenroll --tpm2-device-key=, as mentioned above) for
1001 easier access. A new "srk" verb has been added to systemd-analyze to
1002 allow extracting it on demand if it is already set up.
1004 * The internal systemd-pcrphase executable has been renamed to
1007 * The systemd-pcrextend tool gained a new --pcr= switch to override
1008 which PCR to measure into.
1010 * systemd-pcrextend now exposes a Varlink interface at
1011 io.systemd.PCRExtend that can be used to do measurements and event
1014 * TPM measurements are now also written to an event log at
1015 /run/log/systemd/tpm2-measure.log, using a derivative of the TCG
1016 Canonical Event Log format. Previously we'd only log them to the
1017 journal, where they however were subject to rotation and similar.
1019 * A new component "systemd-pcrlock" has been added that allows managing
1020 local TPM2 PCR policies for PCRs 0-7 and similar, which are hard to
1021 predict by the OS vendor because of the inherently local nature of
1022 what measurements they contain, such as firmware versions of the
1023 system and extension cards and suchlike. pcrlock can predict PCR
1024 measurements ahead of time based on various inputs, such as the local
1025 TPM2 event log, GPT partition tables, PE binaries, UKI kernels, and
1026 various other things. It can then pre-calculate a TPM2 policy from
1027 this, which it stores in an TPM2 NV index. TPM2 objects (such as disk
1028 encryption keys) can be locked against this NV index, so that they
1029 are locked against a specific combination of system firmware and
1030 state. Alternatives for each component are supported to allowlist
1031 multiple kernel versions or boot loader version simultaneously
1032 without losing access to the disk encryption keys. The tool can also
1033 be used to analyze and validate the local TPM2 event log.
1034 systemd-cryptsetup, systemd-cryptenroll, systemd-repart have all been
1035 updated to support such policies. There's currently no support for
1036 locking the system's root disk against a pcrlock policy, this will be
1037 added soon. Moreover, it is currently not possible to combine a
1038 pcrlock policy with a signed PCR policy. This component is
1039 experimental and its public interface is subject to change.
1041 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
1043 * bootctl will now show whether the system was booted from a UKI in its
1046 * systemd-boot and systemd-stub now use different project keys in their
1047 respective SBAT sections, so that they can be revoked individually if
1050 * systemd-boot will no longer load unverified Devicetree blobs when UEFI
1051 SecureBoot is enabled. For more details see:
1052 https://github.com/systemd/systemd/security/advisories/GHSA-6m6p-rjcq-334c
1054 * systemd-boot gained new hotkeys to reboot and power off the system
1055 from the boot menu ("B" and "O"). If the "auto-poweroff" and
1056 "auto-reboot" options in loader.conf are set these entries are also
1057 shown as menu items (which is useful on devices lacking a regular
1060 * systemd-boot gained a new configuration value "menu-disabled" for the
1061 set-timeout option, to allow completely disabling the boot menu,
1062 including the hotkey.
1064 * systemd-boot will now measure the content of loader.conf in TPM2
1067 * systemd-stub will now concatenate the content of all kernel
1068 command-line addons before measuring them in TPM2 PCR 12, in a single
1069 measurement, instead of measuring them individually.
1071 * systemd-stub will now measure and load Devicetree Blob addons, which
1072 are searched and loaded following the same model as the existing
1073 kernel command-line addons.
1075 * systemd-stub will now ignore unauthenticated kernel command line options
1076 passed from systemd-boot when running inside Confidential VMs with UEFI
1079 * systemd-stub will now load a Devicetree blob even if the firmware did
1080 not load any beforehand (e.g.: for ACPI systems).
1082 * ukify is no longer considered experimental, and now ships in /usr/bin/.
1084 * ukify gained a new verb inspect to describe the sections of a UKI and
1085 print the contents of the well-known sections.
1087 * ukify gained a new verb genkey to generate a set of key pairs for
1088 signing UKIs and their PCR data.
1090 * The 90-loaderentry kernel-install hook now supports installing device
1093 * kernel-install now supports the --json=, --root=, --image=, and
1094 --image-policy= options for the inspect verb.
1096 * kernel-install now supports new list and add-all verbs. The former
1097 lists all installed kernel images (if those are available in
1098 /usr/lib/modules/). The latter will install all the kernels it can
1103 * A new option --copy-from= has been added that synthesizes partition
1104 definitions from the given image, which are then applied by the
1105 systemd-repart algorithm.
1107 * A new option --copy-source= has been added, which can be used to specify
1108 a directory to which CopyFiles= is considered relative to.
1110 * New --make-ddi=confext, --make-ddi=sysext, and --make-ddi=portable
1111 options have been added to make it easier to generate these types of
1112 DDIs, without having to provide repart.d definitions for them.
1114 * The dm-verity salt and UUID will now be derived from the specified
1117 * New VerityDataBlockSizeBytes= and VerityHashBlockSizeBytes= can now be
1118 configured in repart.d/ configuration files.
1120 * A new Subvolumes= setting is now supported in repart.d/ configuration
1121 files, to indicate which directories in the target partition should be
1124 * A new --tpm2-device-key= option can be used to lock a disk against a
1125 specific TPM2 public key. This matches the same switch the
1126 systemd-cryptenroll tool now supports (see above).
1130 * The journalctl --lines= parameter now accepts +N to show the oldest N
1131 entries instead of the newest.
1133 * journald now ensures that sealing happens once per epoch, and sets a
1134 new compatibility flag to distinguish old journal files that were
1135 created before this change, for backward compatibility.
1139 * udev will now create symlinks to loopback block devices in the
1140 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
1141 string field selected during allocation. The systemd-dissect tool and
1142 the util-linux losetup command now supports a complementing new switch
1143 --loop-ref= for selecting the string. This means a loopback block
1144 device may now be allocated under a caller-chosen reference and can
1145 subsequently be referenced without first having to look up the block
1146 device name the caller ended up with.
1148 * udev also creates symlinks to loopback block devices in the
1149 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
1150 of the inode attached to the loopback block device. This means that
1151 attaching a file to a loopback device will implicitly make a handle
1152 available to be found via that file's inode information.
1154 * udevadm info gained support for JSON output via a new --json= flag, and
1155 for filtering output using the same mechanism that udevadm trigger
1158 * The predictable network interface naming logic is extended to include
1159 the SR-IOV-R "representor" information in network interface names.
1160 This feature was intended for v254, but even though the code was
1161 merged, the part that actually enabled the feature was forgotten.
1162 It is now enabled by default and is part of the new "v255" naming
1165 * A new hwdb/rules file has been added that sets the
1166 ID_NET_AUTO_LINK_LOCAL_ONLY=1 udev property on all network interfaces
1167 that should usually only be configured with link-local addressing
1168 (IPv4LL + IPv6LL), i.e. for PC-to-PC cables ("laplink") or
1169 Thunderbolt networking. systemd-networkd and NetworkManager (soon)
1170 will make use of this information to apply an appropriate network
1171 configuration by default.
1173 * The ID_NET_DRIVER property on network interfaces is now set
1174 relatively early in the udev rule set so that other rules may rely on
1175 its use. This is implemented in a new "net-driver" udev built-in.
1179 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
1180 is now dropped, as it never worked, hence it should not be used by
1183 * The 'prefixstable' ipv6 address generation mode now considers the SSID
1184 when generating stable addresses, so that a different stable address
1185 is used when roaming between wireless networks. If you already use
1186 'prefixstable' addresses with wireless networks, the stable address
1187 will be changed by the update.
1189 * The DHCPv4 client gained a RapidCommit option, true by default, which
1190 enables RFC4039 Rapid Commit behavior to obtain a lease in a
1191 simplified 2-message exchange instead of the typical 4-message
1192 exchange, if also supported by the DHCP server.
1194 * The DHCPv4 client gained new InitialCongestionWindow= and
1195 InitialAdvertisedReceiveWindow= options for route configurations.
1197 * The DHCPv4 client gained a new RequestAddress= option that allows
1198 to send a preferred IP address in the initial DHCPDISCOVER message.
1200 * The DHCPv4 server and client gained support for IPv6-only mode
1203 * The SendHostname= and Hostname= options are now available for the
1204 DHCPv6 client, independently of the DHCPv4= option, so that these
1205 configuration values can be set independently for each client.
1207 * The DHCPv4 and DHCPv6 client state can now be queried via D-Bus,
1208 including lease information.
1210 * The DHCPv6 client can now be configured to use a custom DUID type.
1212 * .network files gained a new IPv4ReversePathFilter= setting in the
1213 [Network] section, to control sysctl's rp_filter setting.
1215 * .network files gaiend a new HopLimit= setting in the [Route] section,
1216 to configure a per-route hop limit.
1218 * .network files gained a new TCPRetransmissionTimeoutSec= setting in
1219 the [Route] section, to configure a per-route TCP retransmission
1222 * A new directive NFTSet= provides a method for integrating network
1223 configuration into firewall rules with NFT sets. The benefit of using
1224 this setting is that static network configuration or dynamically
1225 obtained network addresses can be used in firewall rules with the
1226 indirection of NFT set types.
1228 * The [IPv6AcceptRA] section supports the following new options:
1229 UsePREF64=, UseHopLimit=, UseICMP6RateLimit=, and NFTSet=.
1231 * The [IPv6SendRA] section supports the following new options:
1232 RetransmitSec=, HopLimit=, HomeAgent=, HomeAgentLifetimeSec=, and
1233 HomeAgentPreference=.
1235 * A new [IPv6PREF64Prefix] set of options, containing Prefix= and
1236 LifetimeSec=, has been introduced to append pref64 options in router
1237 advertisements (RFC8781).
1239 * The network generator now configures the interfaces with only
1240 link-local addressing if "ip=link-local" is specified on the kernel
1243 * The prefix of the configuration files generated by the network
1244 generator from the kernel command line is now prefixed with '70-',
1245 to make them have higher precedence over the default configuration
1248 * Added a new -Ddefault-network=BOOL meson option, that causes more
1249 .network files to be installed as enabled by default. These configuration
1250 files will which match generic setups, e.g. 89-ethernet.network matches
1251 all Ethernet interfaces and enables both DHCPv4 and DHCPv6 clients.
1253 * If a ID_NET_MANAGED_BY= udev property is set on a network device and
1254 it is any other string than "io.systemd.Network" then networkd will
1255 not manage this device. This may be used to allow multiple network
1256 management services to run in parallel and assign ownership of
1257 specific devices explicitly. NetworkManager will soon implement a
1262 * systemctl is-failed now checks the system state if no unit is
1265 * systemctl will now automatically soft-reboot if a new root file system
1266 is found under /run/nextroot/ when a reboot operation is invoked.
1270 * Wall messages now work even when utmp support is disabled, using
1271 systemd-logind to query the necessary information.
1273 * systemd-logind now sends a new PrepareForShutdownWithMetadata D-Bus
1274 signal before shutdown/reboot/soft-reboot that includes additional
1275 information compared to the PrepareForShutdown signal. Currently the
1276 additional information is the type of operation that is about to be
1279 Hibernation & Suspend:
1281 * The kernel and OS versions will no longer be checked on resume from
1284 * Hibernation into swap files backed by btrfs are now
1285 supported. (Previously this was supported only for other file
1290 * A new systemd-vmspawn tool has been added, that aims to provide for VMs
1291 the same interfaces and functionality that systemd-nspawn provides for
1292 containers. For now it supports QEMU as a backend, and exposes some of
1293 its options to the user. This component is experimental and its public
1294 interface is subject to change.
1296 * "systemd-analyze plot" has gained tooltips on each unit name with
1297 related-unit information in its svg output, such as Before=,
1298 Requires=, and similar properties.
1300 * A new varlinkctl tool has been added to allow interfacing with
1301 Varlink services, and introspection has been added to all such
1302 services. This component is experimental and its public interface is
1305 * systemd-sysext and systemd-confext now expose a Varlink service
1306 at io.systemd.sysext.
1308 * portable services now accept confexts as extensions.
1310 * systemd-sysupdate now accepts directories in the MatchPattern= option.
1312 * systemd-run will now output the invocation ID of the launched
1313 transient unit and its peak memory usage.
1315 * systemd-analyze, systemd-tmpfiles, systemd-sysusers, systemd-sysctl,
1316 and systemd-binfmt gained a new --tldr option that can be used instead
1317 of --cat-config to suppress uninteresting configuration lines, such as
1318 comments and whitespace.
1320 * resolvectl gained a new "show-server-state" command that shows
1321 current statistics of the resolver. This is backed by a new
1322 DumpStatistics() Varlink method provided by systemd-resolved.
1324 * systemd-timesyncd will now emit a D-Bus signal when the LinkNTPServers
1327 * vconsole now supports KEYMAP=@kernel for preserving the kernel keymap
1330 * seccomp now supports the LoongArch64 architecture.
1332 * seccomp may now be enabled for services running as a non-root User=
1333 without NoNewPrivileges=yes.
1335 * systemd-id128 now supports a new -P option to show only values. The
1336 combination of -P and --app options is also supported.
1338 * A new pam_systemd_loadkey.so PAM module is now available, which will
1339 automatically fetch the passphrase used by cryptsetup to unlock the
1340 root file system and set it as the PAM authtok. This enables, among
1341 other things, configuring auto-unlock of the GNOME Keyring / KDE
1342 Wallet when autologin is configured.
1344 * Many meson options now use the 'feature' type, which means they
1345 take enabled/disabled/auto as values.
1347 * A new meson option -Dconfigfiledir= can be used to change where
1348 configuration files with default values are installed to.
1350 * Options and verbs in man pages are now tagged with the version they
1351 were first introduced in.
1353 * A new component "systemd-storagetm" has been added, which exposes all
1354 local block devices as NVMe-TCP devices, fully automatically. It's
1355 hooked into a new target unit storage-target-mode.target that is
1356 suppsoed to be booted into via
1357 rd.systemd.unit=storage-target-mode.target on the kernel command
1358 line. This is intended to be used for installers and debugging to
1359 quickly get access to the local disk. It's inspired by MacOS "target
1360 disk mode". This component is experimental and its public interface is
1363 * A new component "systemd-bsod" has been added, which can show logged
1364 error messages full screen, if they have a log level of LOG_EMERG log
1365 level. This component is experimental and its public interface is
1368 * The systemd-dissect tool's --with command will now set the
1369 $SYSTEMD_DISSECT_DEVICE environment variable to the block device it
1370 operates on for the invoked process.
1372 * The systemd-mount tool gained a new --tmpfs switch for mounting a new
1373 'tmpfs' instance. This is useful since it does so via .mount units
1374 and thus can be executed remotely or in containers.
1376 * The various tools in systemd that take "verbs" (such as systemctl,
1377 loginctl, machinectl, …) now will suggest a close verb name in case
1378 the user specified an unrecognized one.
1380 * libsystemd now exports a new function sd_id128_get_app_specific()
1381 that generates "app-specific" 128bit IDs from any ID. It's similar to
1382 sd_id128_get_machine_app_specific() and
1383 sd_id128_get_boot_app_specific() but takes the ID to base calculation
1384 on as input. This new functionality is also exposed in the
1385 "systemd-id128" tool where you can now combine --app= with `show`.
1387 * All tools that parse timestamps now can also parse RFC3339 style
1388 timestamps that include the "T" and Z" characters.
1390 * New documentation has been added:
1392 https://systemd.io/FILE_DESCRIPTOR_STORE
1393 https://systemd.io/TPM2_PCR_MEASUREMENTS
1394 https://systemd.io/MOUNT_REQUIREMENTS
1396 * The codebase now recognizes the suffix .confext.raw and .sysext.raw
1397 as alternative to the .raw suffix generally accepted for DDIs. It is
1398 recommended to name configuration extensions and system extensions
1399 with such suffixes, to indicate their purpose in the name.
1401 * The sd-device API gained a new function
1402 sd_device_enumerator_add_match_property_required() which allows
1403 configuring matches on properties that are strictly required. This is
1404 different from the existing sd_device_enumerator_add_match_property()
1405 matches of which one one needs to apply.
1407 * The MAC address the veth side of an nspawn container shall get
1408 assigned may now be controlled via the $SYSTEMD_NSPAWN_NETWORK_MAC
1409 environment variable.
1411 * The libiptc dependency is now implemented via dlopen(), so that tools
1412 such as networkd and nspawn no longer have a hard dependency on the
1413 shared library when compiled with support for libiptc.
1415 * New rpm macros have been added: %systemd_user_daemon_reexec does
1416 daemon-reexec for all user managers, and %systemd_postun_with_reload
1417 and %systemd_user_postun_with_reload do a reload for system and user
1420 * coredumpctl now propagates SIGTERM to the debugger process.
1422 Contributions from: 김인수, Abderrahim Kitouni, Adam Goldman,
1423 Adam Williamson, Alexandre Peixoto Ferreira, Alex Hudspith,
1424 Alvin Alvarado, André Paiusco, Antonio Alvarez Feijoo,
1425 Anton Lundin, Arian van Putten, Arseny Maslennikov, Arthur Shau,
1426 Balázs Úr, beh_10257, Benjamin Peterson, Bertrand Jacquin,
1427 Brian Norris, Charles Lee, Cheng-Chia Tseng, Chris Patterson,
1428 Christian Hergert, Christian Hesse, Christian Kirbach,
1429 Clayton Craft, commondservice, cunshunxia, Curtis Klein, cvlc12,
1430 Daan De Meyer, Daniele Medri, Daniel P. Berrangé, Daniel Rusek,
1431 Daniel Thompson, Dan Nicholson, Dan Streetman, David Rheinsberg,
1432 David Santamaría Rogado, David Tardon, dependabot[bot],
1433 Diego Viola, Dmitry V. Levin, Emanuele Giuseppe Esposito,
1434 Emil Renner Berthing, Emil Velikov, Etienne Dechamps, Fabian Vogt,
1435 felixdoerre, Felix Dörre, Florian Schmaus, Franck Bui,
1436 Frantisek Sumsal, G2-Games, Gioele Barabucci, Hugo Carvalho,
1437 huyubiao, Iago López Galeiras, IllusionMan1212, Jade Lovelace,
1438 janana, Jan Janssen, Jan Kuparinen, Jan Macku, Jeremy Fleischman,
1439 Jin Liu, jjimbo137, Joerg Behrmann, Johannes Segitz, Jordan Rome,
1440 Jordan Williams, Julien Malka, Juno Computers, Khem Raj, khm,
1441 Kingbom Dou, Kiran Vemula, Krzesimir Nowak, Laszlo Gombos,
1442 Lennart Poettering, linuxlion, Luca Boccassi, Lucas Adriano Salles,
1443 Lukas, Lukáš Nykrýn, Maanya Goenka, Maarten, Malte Poll,
1444 Marc Pervaz Boocha, Martin Beneš, Martin Joerg, Martin Wilck,
1445 Mathieu Tortuyaux, Matthias Schiffer, Maxim Mikityanskiy,
1446 Max Kellermann, Michael A Cassaniti, Michael Biebl, Michael Kuhn,
1447 Michael Vasseur, Michal Koutný, Michal Sekletár, Mike Yuan,
1448 Milton D. Miller II, mordner, msizanoen, NAHO, Nandakumar Raghavan,
1449 Neil Wilson, Nick Rosbrook, Nils K, NRK, Oğuz Ersen,
1450 Omojola Joshua, onenowy, Paul Meyer, Paymon MARANDI, pelaufer,
1451 Peter Hutterer, PhylLu, Pierre GRASSER, Piotr Drąg, Priit Laes,
1452 Rahil Bhimjiani, Raito Bezarius, Raul Cheleguini, Reto Schneider,
1453 Richard Maw, Robby Red, RoepLuke, Roland Hieber, Roland Singer,
1454 Ronan Pigott, Sam James, Sam Leonard, Sergey A, Susant Sahani,
1455 Sven Joachim, Tad Fisher, Takashi Sakamoto, Thorsten Kukuk, Tj,
1456 Tomasz Świątek, Topi Miettinen, Valentin David,
1457 Valentin Lefebvre, Victor Westerhuis, Vincent Haupert,
1458 Vishal Chillara Srinivas, Vito Caputo, Warren, Weblate,
1459 Xiaotian Wu, xinpeng wang, Yaron Shahrabani, Yo-Jung Lin,
1460 Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zeroskyx,
1461 Дамјан Георгиевски, наб
1463 — Edinburgh, 2023-12-06
1467 Announcements of Future Feature Removals and Incompatible Changes:
1469 * The next release (v255) will remove support for split-usr (/usr/
1470 mounted separately during late boot, instead of being mounted by the
1471 initrd before switching to the rootfs) and unmerged-usr (parallel
1472 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
1474 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1476 * We intend to remove cgroup v1 support from a systemd release after
1477 the end of 2023. If you run services that make explicit use of
1478 cgroup v1 features (i.e. the "legacy hierarchy" with separate
1479 hierarchies for each controller), please implement compatibility with
1480 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
1481 Most of Linux userspace has been ported over already.
1483 * Support for System V service scripts is now deprecated and will be
1484 removed in a future release. Please make sure to update your software
1485 *now* to include a native systemd unit file instead of a legacy
1486 System V script to retain compatibility with future systemd releases.
1488 * Support for the SystemdOptions EFI variable is deprecated.
1489 'bootctl systemd-efi-options' will emit a warning when used. It seems
1490 that this feature is little-used and it is better to use alternative
1491 approaches like credentials and confexts. The plan is to drop support
1492 altogether at a later point, but this might be revisited based on
1495 * EnvironmentFile= now treats the line following a comment line
1496 trailing with escape as a non comment line. For details, see:
1497 https://github.com/systemd/systemd/issues/27975
1499 * PrivateNetwork=yes and NetworkNamespacePath= now imply
1500 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
1502 * Behaviour of sandboxing options for the per-user service manager
1503 units has changed. They now imply PrivateUsers=yes, which means user
1504 namespaces will be implicitly enabled when a sandboxing option is
1505 enabled in a user unit. Enabling user namespaces has the drawback
1506 that system users will no longer be visible (and processes/files will
1507 appear as owned by 'nobody') in the user unit.
1509 By definition a sandboxed user unit should run with reduced
1510 privileges, so impact should be small. This will remove a great
1511 source of confusion that has been reported by users over the years,
1512 due to how these options require an extra setting to be manually
1513 enabled when used in the per-user service manager, which is not
1514 needed in the system service manager. For more details, see:
1515 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
1517 * systemd-run's switch --expand-environment= which currently is disabled
1518 by default when combined with --scope, will be changed in a future
1519 release to be enabled by default.
1521 Security Relevant Changes:
1523 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
1524 process capability to invoked session processes of regular users on
1525 local seats (as well as to systemd --user), unless configured
1526 otherwise via data from JSON user records, or via the PAM module's
1527 parameter list. This is useful in order allow desktop tools such as
1528 GNOME's Alarm Clock application to set a timer for
1529 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
1530 per-user service unit file may thus use AmbientCapability= to pass
1531 the capability to invoked processes. Note that this capability is
1532 relatively narrow in focus (in particular compared to other process
1533 capabilities such as CAP_SYS_ADMIN) and we already — by default —
1534 permit more impactful operations such as system suspend to local
1539 * Memory limits that apply while the unit is activating are now
1540 supported. Previously IO and CPU settings were already supported via
1541 StartupCPUWeight= and similar. The same logic has been added for the
1542 various manager and unit memory settings (DefaultStartupMemoryLow=,
1543 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
1544 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
1546 * The service manager gained support for enqueuing POSIX signals to
1547 services that carry an additional integer value, exposing the
1548 sigqueue() system call. This is accessible via new D-Bus calls
1549 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
1550 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
1551 via the new --kill-value= option.
1553 * systemctl gained a new "list-paths" verb, which shows all currently
1554 active .path units, similarly to how "systemctl list-timers" shows
1555 active timers, and "systemctl list-sockets" shows active sockets.
1557 * systemctl gained a new --when= switch which is honoured by the various
1558 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
1559 scheduling these operations by time, similar in fashion to how this
1560 has been supported by SysV shutdown.
1562 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
1563 supports the new PR_SET_MDWE prctl() call, it is used instead of the
1564 seccomp()-based system call filter to achieve the same effect.
1566 * A new set of kernel command line options is now understood:
1567 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
1568 systemd.tty.columns.<name>= allow configuring the TTY type and
1569 dimensions for the tty specified via <name>. When systemd invokes a
1570 service on a tty (via TTYName=) it will look for these and configure
1571 the TTY accordingly. This is particularly useful in VM environments
1572 to propagate host terminal settings into the appropriate TTYs of the
1575 * A new RootEphemeral= setting is now understood in service units. It
1576 takes a boolean argument. If enabled for services that use RootImage=
1577 or RootDirectory= an ephemeral copy of the disk image or directory
1578 tree is made when the service is started. It is removed automatically
1579 when the service is stopped. That ephemeral copy is made using
1580 btrfs/xfs reflinks or btrfs snapshots, if available.
1582 * The service activation logic gained new settings RestartSteps= and
1583 RestartMaxDelaySec= which allow exponentially-growing restart
1584 intervals for Restart=.
1586 * The service activation logic gained a new setting RestartMode= which
1587 can be set to 'direct' to skip the inactive/failed states when
1588 restarting, so that dependent units are not notified until the service
1589 converges to a final (successful or failed) state. For example, this
1590 means that OnSuccess=/OnFailure= units will not be triggered until the
1591 service state has converged.
1593 * PID 1 will now automatically load the virtio_console kernel module
1594 during early initialization if running in a suitable VM. This is done
1595 so that early-boot logging can be written to the console if available.
1597 * Similarly, virtio-vsock support is loaded early in suitable VM
1598 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
1599 to the VMM if configured, thus loading this early is beneficial.
1601 * A new verb "fdstore" has been added to systemd-analyze to show the
1602 current contents of the file descriptor store of a unit. This is
1603 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
1604 the service manager.
1606 * The service manager will now set a new $FDSTORE environment variable
1607 when invoking processes for services that have the file descriptor
1610 * A new service option FileDescriptorStorePreserve= has been added that
1611 allows tuning the lifecycle of the per-service file descriptor store.
1612 If set to "yes", the entries in the fd store are retained even after
1613 the service has been fully stopped.
1615 * The "systemctl clean" command may now be used to clear the fdstore of
1618 * Unit *.preset files gained a new directive "ignore", in addition to
1619 the existing "enable" and "disable". As the name suggests, matching
1620 units are left unchanged, i.e. neither enabled nor disabled.
1622 * Service units gained a new setting DelegateSubgroup=. It takes the
1623 name of a sub-cgroup to place any processes the service manager forks
1624 off in. Previously, the service manager would place all service
1625 processes directly in the top-level cgroup it created for the
1626 service. This usually meant that main process in a service with
1627 delegation enabled would first have to create a subgroup and move
1628 itself down into it, in order to not conflict with the "no processes
1629 in inner cgroups" rule of cgroup v2. With this option, this step is
1630 now handled by PID 1.
1632 * The service manager will now look for .upholds/ directories,
1633 similarly to the existing support for .wants/ and .requires/
1634 directories. Symlinks in this directory result in Upholds=
1637 The [Install] section of unit files gained support for a new
1638 UpheldBy= directive to generate .upholds/ symlinks automatically when
1641 * The service manager now supports a new kernel command line option
1642 systemd.default_device_timeout_sec=, which may be used to override
1643 the default timeout for .device units.
1645 * A new "soft-reboot" mechanism has been added to the service manager.
1646 A "soft reboot" is similar to a regular reboot, except that it
1647 affects userspace only: the service manager shuts down any running
1648 services and other units, then optionally switches into a new root
1649 file system (mounted to /run/nextroot/), and then passes control to a
1650 systemd instance in the new file system which then starts the system
1651 up again. The kernel is not rebooted and neither is the hardware,
1652 firmware or boot loader. This provides a fast, lightweight mechanism
1653 to quickly reset or update userspace, without the latency that a full
1654 system reset involves. Moreover, open file descriptors may be passed
1655 across the soft reboot into the new system where they will be passed
1656 back to the originating services. This allows pinning resources
1657 across the reboot, thus minimizing grey-out time further. This new
1658 reboot mechanism is accessible via the new "systemctl soft-reboot"
1661 * Services using RootDirectory= or RootImage= will now have read-only
1662 access to a copy of the host's os-release file under
1663 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
1664 This was already the case for Portable Services, and the feature has
1665 now been extended to all services that do not run off the host's
1668 * A new service setting MemoryKSM= has been added to enable kernel
1669 same-page merging individually for services.
1671 * A new service setting ImportCredentials= has been added that augments
1672 LoadCredential= and LoadCredentialEncrypted= and searches for
1673 credentials to import from the system, and supports globbing.
1675 * A new job mode "restart-dependencies" has been added to the service
1676 manager (exposed via systemctl --job-mode=). It is only valid when
1677 used with "start" jobs, and has the effect that the "start" job will
1678 be propagated as "restart" jobs to currently running units that have
1679 a BindsTo= or Requires= dependency on the started unit.
1681 * A new verb "whoami" has been added to "systemctl" which determines as
1682 part of which unit the command is being invoked. It writes the unit
1683 name to standard output. If one or more PIDs are specified reports
1684 the unit names the processes referenced by the PIDs belong to.
1686 * The system and service credential logic has been improved: there's
1687 now a clearly defined place where system provisioning tools running
1688 in the initrd can place credentials that will be imported into the
1689 system's set of credentials during the initrd → host transition: the
1690 /run/credentials/@initrd/ directory. Once the credentials placed
1691 there are imported into the system credential set they are deleted
1692 from this directory, and the directory itself is deleted afterwards
1695 * A new kernel command line option systemd.set_credential_binary= has
1696 been added, that is similar to the pre-existing
1697 systemd.set_credential= but accepts arbitrary binary credential data,
1698 encoded in Base64. Note that the kernel command line is not a
1699 recommend way to transfer credentials into a system, since it is
1700 world-readable from userspace.
1702 * The default machine ID to use may now be configured via the
1703 system.machine_id system credential. It will only be used if no
1704 machine ID was set yet on the host.
1706 * On Linux kernel 6.4 and newer system and service credentials will now
1707 be placed in a tmpfs instance that has the "noswap" mount option
1708 set. Previously, a "ramfs" instance was used. By switching to tmpfs
1709 ACL support and overall size limits can now be enforced, without
1710 compromising on security, as the memory is never paged out either
1713 * The service manager now can detect when it is running in a
1714 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
1715 accepted by ConditionSecurity= for units that want to conditionalize
1716 themselves on this. systemd-detect-virt gained new 'cvm' and
1717 '--list-cvm' switches to respectively perform the detection or list
1718 all known flavours of confidential VM, depending on the vendor. The
1719 manager will publish a 'ConfidentialVirtualization' D-Bus property,
1720 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
1721 variable for unit generators. Finally, udev rules can match on a new
1722 'cvm' key that will be set when in a confidential VM.
1723 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
1724 strings and QEMU's fw_cfg protocol will not be used to import
1725 credentials and kernel command line parameters by the system manager,
1726 systemd-boot and systemd-stub, because the hypervisor is considered
1727 untrusted in this particular setting.
1731 * The sd-journal API gained a new call sd_journal_get_seqnum() to
1732 retrieve the current log record's sequence number and sequence number
1733 ID, which allows applications to order records the same way as
1734 journal does internally. The sequence number is now also exported in
1735 the JSON and "export" output of the journal.
1737 * journalctl gained a new switch --truncate-newline. If specified
1738 multi-line log records will be truncated at the first newline,
1739 i.e. only the first line of each log message will be shown.
1741 * systemd-journal-upload gained support for --namespace=, similar to
1742 the switch of the same name of journalctl.
1746 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
1747 may be used to exclude certain files from the effect of CopyFiles=.
1749 * systemd-repart's Verity support now implements the Minimize= setting
1750 to minimize the size of the resulting partition.
1752 * systemd-repart gained a new --offline= switch, which may be used to
1753 control whether images shall be built "online" or "offline",
1754 i.e. whether to make use of kernel facilities such as loopback block
1755 devices and device mapper or not.
1757 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
1758 partition with some files, it will now default to VFAT rather than
1761 * systemd-repart gained a new --architecture= switch. If specified, the
1762 per-architecture GPT partition types (i.e. the root and /usr/
1763 partitions) configured in the partition drop-in files are
1764 automatically adjusted to match the specified CPU architecture, in
1765 order to simplify cross-architecture DDI building.
1767 * systemd-repart will now default to a minimum size of 300MB for XFS
1768 filesystems if no size parameter is specified. This matches what the
1769 XFS tools (xfsprogs) can support.
1771 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
1773 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
1774 Instead, pyelftools is now needed, and it will be used to perform the
1775 ELF -> PE relocations at build time.
1777 * bootctl gained a new switch --print-root-device/-R that prints the
1778 block device the root file system is backed by. If specified twice,
1779 it returns the whole disk block device (as opposed to partition block
1780 device) the root file system is on. It's useful for invocations such
1781 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
1784 * systemd-stub will now look for the SMBIOS Type 1 field
1785 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
1786 kernel command line it invokes. This is useful for VMMs such as qemu
1787 to pass additional kernel command lines into the system even when
1788 booting via full UEFI. The contents of the field are measured into
1791 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
1792 value "auto". With this value, a kernel will be automatically
1793 analyzed, and if it qualifies as UKI, it will be installed as if the
1794 setting was to set to "uki", otherwise as "bls".
1796 * systemd-stub can now optionally load UEFI PE "add-on" images that may
1797 contain additional kernel command line information. These "add-ons"
1798 superficially look like a regular UEFI executable, and are expected
1799 to be signed via SecureBoot/shim. However, they do not actually
1800 contain code, but instead a subset of the PE sections that UKIs
1801 support. They are supposed to provide a way to extend UKIs with
1802 additional resources in a secure and authenticated way. Currently,
1803 only the .cmdline PE section may be used in add-ons, in which case
1804 any specified string is appended to the command line embedded into
1805 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
1806 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
1807 the future we expect other sections to be made extensible like this as
1810 * ukify has been updated to allow building these UEFI PE "add-on"
1811 images, using the new 'addon<EFI-ARCH>.efi.stub'.
1813 * ukify now accepts SBAT information to place in the .sbat PE section
1814 of UKIs and addons. If a UKI is built the SBAT information from the
1815 inner kernel is merged with any SBAT information associated with
1816 systemd-stub and the SBAT data specified on the ukify command line.
1818 * The kernel-install script has been rewritten in C, and reuses much of
1819 the infrastructure of existing tools such as bootctl. It also gained
1820 --esp-path= and --boot-path= options to override the path to the ESP,
1821 and the $BOOT partition. Options --make-entry-directory= and
1822 --entry-token= have been added as well, similar to bootctl's options
1825 * A new kernel-install plugin 60-ukify has been added which will
1826 combine kernel/initrd locally into a UKI and optionally sign them
1827 with a local key. This may be used to switch to UKI mode even on
1828 systems where a local kernel or initrd is used. (Typically UKIs are
1829 built and signed by the vendor.)
1831 * The ukify tool now supports "pesign" in addition to the pre-existing
1832 "sbsign" for signing UKIs.
1834 * systemd-measure and systemd-stub now look for the .uname PE section
1835 that should contain the kernel's "uname -r" string.
1837 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
1838 "offline", i.e. without access to a TPM (physical or
1841 Memory Pressure & Control:
1843 * The sd-event API gained new calls sd_event_add_memory_pressure(),
1844 sd_event_source_set_memory_pressure_type(),
1845 sd_event_source_set_memory_pressure_period() to create and configure
1846 an event source that is called whenever the OS signals memory
1847 pressure. Another call sd_event_trim_memory() is provided that
1848 compacts the process' memory use by releasing allocated but unused
1849 malloc() memory back to the kernel. Services can also provide their
1850 own custom callback to do memory trimming. This should improve system
1851 behaviour under memory pressure, as on Linux traditionally provided
1852 no mechanism to return process memory back to the kernel if the
1853 kernel was under memory pressure. This makes use of the kernel's PSI
1854 interface. Most long-running services in systemd have been hooked up
1855 with this, and in particular systems with low memory should benefit
1858 * Service units gained new settings MemoryPressureWatch= and
1859 MemoryPressureThresholdSec= to configure the PSI memory pressure
1860 logic individually. If these options are used, the
1861 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
1862 variables will be set for the invoked processes to inform them about
1863 the requested memory pressure behaviour. (This is used by the
1864 aforementioned sd-events API additions, if set.)
1866 * systemd-analyze gained a new "malloc" verb that shows the output
1867 generated by glibc's malloc_info() on services that support it. Right
1868 now, only the service manager has been updated accordingly. This
1869 call requires privileges.
1871 User & Session Management:
1873 * The sd-login API gained a new call sd_session_get_username() to
1874 return the user name of the owner of a login session. It also gained
1875 a new call sd_session_get_start_time() to retrieve the time the login
1876 session started. A new call sd_session_get_leader() has been added to
1877 return the PID of the "leader" process of a session. A new call
1878 sd_uid_get_login_time() returns the time since the specified user has
1879 most recently been continuously logged in with at least one session.
1881 * JSON user records gained a new set of fields capabilityAmbientSet and
1882 capabilityBoundingSet which contain a list of POSIX capabilities to
1883 set for the logged in users in the ambient and bounding sets,
1884 respectively. homectl gained the ability to configure these two sets
1885 for users via --capability-bounding-set=/--capability-ambient-set=.
1887 * pam_systemd learnt two new module options
1888 default-capability-bounding-set= and default-capability-ambient-set=,
1889 which configure the default bounding sets for users as they are
1890 logging in, if the JSON user record doesn't specify this explicitly
1891 (see above). The built-in default for the ambient set now contains
1892 the CAP_WAKE_ALARM, thus allowing regular users who may log in
1893 locally to resume from a system suspend via a timer.
1895 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
1896 call to update the TTY of a session after it has been allocated. This
1897 is useful for SSH sessions which are typically allocated first, and
1898 for which a TTY is added later.
1900 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
1901 combines the various other sd_pid_notify() flavours into one: takes a
1902 format string, an overriding PID, and a set of file descriptors to
1903 send. It also gained a new call sd_pid_notify_barrier() call which is
1904 equivalent to sd_notify_barrier() but allows the originating PID to
1907 * "loginctl list-users" and "loginctl list-sessions" will now show the
1908 state of each logged in user/session in their tabular output. It will
1909 also show the current idle state of sessions.
1913 * systemd-dissect will now show the intended CPU architecture of an
1916 * systemd-dissect will now install itself as mount helper for the "ddi"
1917 pseudo-file system type. This means you may now mount DDIs directly
1918 via /bin/mount or /etc/fstab, making full use of embedded Verity
1919 information and all other DDI features.
1921 Example: mount -t ddi myimage.raw /some/where
1923 * The systemd-dissect tool gained the new switches --attach/--detach to
1924 attach/detach a DDI to a loopback block device without mounting it.
1925 It will automatically derive the right sector size from the image
1926 and set up Verity and similar, but not mount the file systems in it.
1928 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
1929 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
1930 implied. Given that these file systems are typically untrusted, this
1931 should make mounting them automatically have less of a security
1934 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
1935 systemd-tmpfiles, …) now understand a new switch --image-policy= which
1936 takes a string encoding image dissection policy. With this mechanism
1937 automatic discovery and use of specific partition types and the
1938 cryptographic requirements on the partitions (Verity, LUKS, …) can be
1939 restricted, permitting better control of the exposed attack surfaces
1940 when mounting disk images. systemd-gpt-auto-generator will honour such
1941 an image policy too, configurable via the systemd.image_policy= kernel
1942 command line option. Unit files gained the RootImagePolicy=,
1943 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
1944 disk images a service runs off.
1946 * systemd-analyze gained a new verb "image-policy" to validate and
1947 parse image policy strings.
1949 * systemd-dissect gained support for a new --validate switch to
1950 superficially validate DDI structure, and check whether a specific
1951 image policy allows the DDI.
1953 * systemd-dissect gained support for a new --mtree-hash switch to
1954 optionally disable calculating mtree hashes, which can be slow on
1957 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
1958 are now able to operate on directories too, other than images.
1962 * networkd's GENEVE support as gained a new .network option
1963 InheritInnerProtocol=.
1965 * The [Tunnel] section in .netdev files has gained a new setting
1966 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
1968 * A new global IPv6PrivacyExtensions= setting has been added that
1969 selects the default value of the per-network setting of the same
1972 * The predictable network interface naming logic was extended to
1973 include SR-IOV-R "representor" information in network interface
1974 names. Unfortunately, this feature was not enabled by default and can
1975 only be enabled at compilation time by setting
1976 -Ddefault-net-naming-scheme=v254.
1978 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
1979 the RFC8910 captive portal option.
1983 * udevadm gained the new "verify" verb for validating udev rules files
1986 * udev gained a new tool "iocost" that can be used to configure QoS IO
1987 cost data based on hwdb information onto suitable block devices. Also
1988 see https://github.com/iocost-benchmark/iocost-benchmarks.
1990 TPM2 Support + Disk Encryption & Authentication:
1992 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
1993 ("Storage Root Key") as first step in the TPM2, and then use that
1994 for binding FDE to, if TPM2 support is used. This matches
1995 recommendations of TCG (see
1996 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
1998 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
1999 understand textual identifiers for these PCRs.
2001 * systemd-veritysetup + /etc/veritytab gained support for a series of
2002 new options: hash-offset=, superblock=, format=, data-block-size=,
2003 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
2004 fec-offset=, fec-roots= to configure various aspects of a Verity
2007 * systemd-cryptsetup + /etc/crypttab gained support for a new
2008 veracrypt-pim= option for setting the Personal Iteration Multiplier
2009 of veracrypt volumes.
2011 * systemd-integritysetup + /etc/integritytab gained support for a new
2012 mode= setting for controlling the dm-integrity mode (journal, bitmap,
2013 direct) for the volume.
2015 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
2016 registers, their symbolic names and current values.
2020 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
2021 access right is specified this is equivalent to "x" but only if the
2022 inode in question already has the executable bit set for at least
2023 some user/group. Otherwise the "x" bit will be turned off.
2025 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
2026 with C+ will result in a "merge" copy, i.e. all files of the source
2027 tree are copied into the target tree, even if that tree already
2028 exists, resulting in a combined tree of files already present in the
2029 target tree and those copied in.
2031 * systemd-tmpfiles gained a new --graceful switch. If specified lines
2032 with unknown users/groups will silently be skipped.
2036 * systemd-notify gained two new options --fd= and --fdname= for sending
2037 arbitrary file descriptors to the service manager (while specifying an
2038 explicit name for it).
2040 * systemd-notify gained a new --exec switch, which makes it execute the
2041 specified command line after sending the requested messages. This is
2042 useful for sending out READY=1 first, and then continuing invocation
2043 without changing process ID, so that the tool can be nicely used
2044 within an ExecStart= line of a unit file that uses Type=notify.
2046 sd-event + sd-bus APIs:
2048 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
2049 which may be used to explicitly end a rate-limit state an event
2050 source might be in, resetting all rate limiting counters.
2052 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
2053 sockets, it will now encode the "description" set via
2054 sd_bus_set_description() into the source socket address. It will also
2055 look for this information when accepting a connection. This is useful
2056 to track individual D-Bus connections on a D-Bus broker for debug
2061 * systemd-resolved gained a new resolved.conf setting
2062 StateRetentionSec= which may be used to retain cached DNS records
2063 even after their nominal TTL, and use them in case upstream DNS
2064 servers cannot be reached. This can be used to make name resolution
2065 more resilient in case of network problems.
2067 * resolvectl gained a new verb "show-cache" to show the current cache
2068 contents of systemd-resolved. This verb communicates with the
2069 systemd-resolved daemon and requires privileges.
2073 * Meson >= 0.60.0 is now required to build systemd.
2075 * The default keymap to apply may now be chosen at build-time via the
2076 new -Ddefault-keymap= meson option.
2078 * Most of systemd's long-running services now have a generic handler of
2079 the SIGRTMIN+18 signal handler which executes various operations
2080 depending on the sigqueue() parameter sent along. For example, values
2081 0x100…0x107 allow changing the maximum log level of such
2082 services. 0x200…0x203 allow changing the log target of such
2083 services. 0x300 make the services trim their memory similarly to the
2084 automatic PSI-triggered action, see above. 0x301 make the services
2085 output their malloc_info() data to the logs.
2087 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
2088 files, inspired by systemctl's verbs of the same name which edit unit
2089 files. Similarly, networkctl gained the same verbs for editing
2090 .network, .netdev, .link files.
2092 * A new syscall filter group "@sandbox" has been added that contains
2093 syscalls for sandboxing system calls such as those for seccomp and
2096 * New documentation has been added:
2098 https://systemd.io/COREDUMP
2099 https://systemd.io/MEMORY_PRESSURE
2102 * systemd-firstboot gained a new --reset option. If specified, the
2103 settings in /etc/ it knows how to initialize are reset.
2105 * systemd-sysext is now a multi-call binary and is also installed under
2106 the systemd-confext alias name (via a symlink). When invoked that way
2107 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
2108 powerful, atomic, secure configuration management of sorts, that
2109 locally can merge configuration from multiple confext configuration
2110 images into a single immutable tree.
2112 * The --network-macvlan=, --network-ipvlan=, --network-interface=
2113 switches of systemd-nspawn may now optionally take the intended
2114 network interface inside the container.
2116 * All our programs will now send an sd_notify() message with their exit
2117 status in the EXIT_STATUS= field when exiting, using the usual
2118 protocol, including PID 1. This is useful for VMMs and container
2119 managers to collect an exit status from a system as it shuts down, as
2120 set via "systemctl exit …". This is particularly useful in test cases
2121 and similar, as invocations via a VM can now nicely propagate an exit
2122 status to the host, similar to local processes.
2124 * systemd-run gained a new switch --expand-environment=no to disable
2125 server-side environment variable expansion in specified command
2126 lines. Expansion defaults to enabled for all execution types except
2127 --scope, where it defaults to off (and prints a warning) for backward
2128 compatibility reasons. --scope will be flipped to enabled by default
2129 too in a future release. If you are using --scope and passing a '$'
2130 character in the payload you should start explicitly using
2131 --expand-environment=yes/no according to the use case.
2133 * The systemd-system-update-generator has been updated to also look for
2134 the special flag file /etc/system-update in addition to the existing
2135 support for /system-update to decide whether to enter system update
2138 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
2139 mount options by default.
2141 * systemd-fstab-generator now understands two new kernel command line
2142 options systemd.mount-extra= and systemd.swap-extra=, which configure
2143 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
2144 will be ran on these block devices, like it already happens for
2145 'root='. It also now supports the new fstab.extra and
2146 fstab.extra.initrd credentials that may contain additional /etc/fstab
2147 lines to apply at boot.
2149 * systemd-getty-generator now understands two new credentials
2150 getty.ttys.container and getty.ttys.serial. These credentials may
2151 contain a list of TTY devices – one per line – to instantiate
2152 container-getty@.service and serial-getty@.service on.
2154 * The getty/serial-getty/container-getty units now import the 'agetty.*'
2155 and 'login.*' credentials, which are consumed by the 'login' and
2156 'agetty' programs starting from util-linux v2.40.
2158 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
2159 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
2160 which case the Path= setting is taken relative to the ESP or XBOOTLDR
2161 partitions, rather than the system's root directory /. The relevant
2162 directories are automatically discovered.
2164 * The systemd-ac-power tool gained a new switch --low, which reports
2165 whether the battery charge is considered "low", similar to how the
2166 s2h suspend logic checks this state to decide whether to enter system
2167 suspend or hibernation.
2169 * The /etc/os-release file can now have two new optional fields
2170 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
2173 * When the system hibernates, information about the device and offset
2174 used is now written to a non-volatile EFI variable. On next boot the
2175 system will attempt to resume from the location indicated in this EFI
2176 variable. This should make hibernation a lot more robust, while
2177 requiring no manual configuration of the resume location.
2179 * The $XDG_STATE_HOME environment variable (added in more recent
2180 versions of the XDG basedir specification) is now honoured to
2181 implement the StateDirectory= setting in user services.
2183 * A new component "systemd-battery-check" has been added. It may run
2184 during early boot (usually in the initrd), and checks the battery
2185 charge level of the system. In case the charge level is very low the
2186 user is notified (graphically via Plymouth – if available – as well
2187 as in text form on the console), and the system is turned off after a
2188 10s delay. The feature can be disabled by passing
2189 systemd.battery_check=0 through the kernel command line.
2191 * The 'passwdqc' library is now supported as an alternative to the
2192 'pwquality' library and can be selected at build time.
2194 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
2195 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
2196 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
2197 Arian van Putten, Arthur Shau, A S Alam,
2198 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
2199 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
2200 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
2201 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
2202 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
2203 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
2204 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
2205 David Schroeder, David Tardon, dependabot[bot],
2206 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
2207 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
2208 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
2209 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
2210 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
2211 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
2212 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
2213 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
2214 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
2215 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
2216 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
2217 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
2218 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
2219 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
2220 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
2221 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
2222 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
2223 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
2224 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
2225 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
2226 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
2227 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
2228 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
2229 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
2230 saikat0511, Samanta Navarro, Sam James, Sam Morris,
2231 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
2232 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
2233 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
2234 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
2235 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
2236 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
2237 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
2238 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
2239 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
2240 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
2241 Дамјан Георгиевски, наб
2243 — Edinburgh, 2023-07-28
2247 Announcements of Future Feature Removals and Incompatible Changes:
2249 * We intend to remove cgroup v1 support from systemd release after the
2250 end of 2023. If you run services that make explicit use of cgroup v1
2251 features (i.e. the "legacy hierarchy" with separate hierarchies for
2252 each controller), please implement compatibility with cgroup v2 (i.e.
2253 the "unified hierarchy") sooner rather than later. Most of Linux
2254 userspace has been ported over already.
2256 * We intend to remove support for split-usr (/usr mounted separately
2257 during boot) and unmerged-usr (parallel directories /bin and
2258 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
2259 half of 2023, in the first release that falls into that time window.
2260 For more details, see:
2261 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
2263 * We intend to change behaviour w.r.t. units of the per-user service
2264 manager and sandboxing options, so that they work without having to
2265 manually enable PrivateUsers= as well, which is not required for
2266 system units. To make this work, we will implicitly enable user
2267 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
2268 user unit. The drawback is that system users will no longer be visible
2269 (and appear as 'nobody') to the user unit when a sandboxing option is
2270 enabled. By definition a sandboxed user unit should run with reduced
2271 privileges, so impact should be small. This will remove a great source
2272 of confusion that has been reported by users over the years, due to
2273 how these options require an extra setting to be manually enabled when
2274 used in the per-user service manager, as opposed as to the system
2275 service manager. We plan to enable this change in the next release
2276 later this year. For more details, see:
2277 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
2279 Deprecations and incompatible changes:
2281 * systemctl will now warn when invoked without /proc/ mounted
2282 (e.g. when invoked after chroot() into an directory tree without the
2283 API mount points like /proc/ being set up.) Operation in such an
2284 environment is not fully supported.
2286 * The return value of 'systemctl is-active|is-enabled|is-failed' for
2287 unknown units is changed: previously 1 or 3 were returned, but now 4
2288 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
2290 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
2291 systemd-hwdb (added in 2014) should be used instead.
2293 * 'bootctl --json' now outputs a single JSON array, instead of a stream
2294 of newline-separated JSON objects.
2296 * Udev rules in 60-evdev.rules have been changed to load hwdb
2297 properties for all modalias patterns. Previously only the first
2298 matching pattern was used. This could change what properties are
2299 assigned if the user has more and less specific patterns that could
2300 match the same device, but it is expected that the change will have
2301 no effect for most users.
2303 * systemd-networkd-wait-online exits successfully when all interfaces
2304 are ready or unmanaged. Previously, if neither '--any' nor
2305 '--interface=' options were used, at least one interface had to be in
2306 configured state. This change allows the case where systemd-networkd
2307 is enabled, but no interfaces are configured, to be handled
2308 gracefully. It may occur in particular when a different network
2309 manager is also enabled and used.
2311 * Some compatibility helpers were dropped: EmergencyAction= in the user
2312 manager, as well as measuring kernel command line into PCR 8 in
2313 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
2316 * The '-Dupdate-helper-user-timeout=' build-time option has been
2317 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
2318 integer as parameter instead of a string.
2320 * The DDI image dissection logic (which backs RootImage= in service
2321 unit files, the --image= switch in various tools such as
2322 systemd-nspawn, as well as systemd-dissect) will now only mount file
2323 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
2324 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
2325 variable. These file systems are fairly well supported and maintained
2326 in current kernels, while others are usually more niche, exotic or
2327 legacy and thus typically do not receive the same level of security
2330 * The default per-link multicast DNS mode is changed to "yes"
2331 (that was previously "no"). As the default global multicast DNS mode
2332 has been "yes" (but can be changed by the build option), now the
2333 multicast DNS is enabled on all links by default. You can disable the
2334 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
2335 or on an interface by calling "resolvectl mdns INTERFACE no".
2339 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
2340 (UKIs) has been added. This replaces functionality provided by
2341 'dracut --uefi' and extends it with automatic calculation of PE file
2342 offsets, insertion of signed PCR policies generated by
2343 systemd-measure, support for initrd concatenation, signing of the
2344 embedded Linux image and the combined image with sbsign, and
2345 heuristics to autodetect the kernel uname and verify the splash
2348 Changes in systemd and units:
2350 * A new service type Type=notify-reload is defined. When such a unit is
2351 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
2352 service process. The manager will then wait until it receives a
2353 "RELOADING=1" followed by a "READY=1" notification from the unit as
2354 response (via sd_notify()). Otherwise, this type is the same as
2355 Type=notify. A new setting ReloadSignal= may be used to change the
2356 signal to send from the default of SIGHUP.
2358 user@.service, systemd-networkd.service, systemd-udevd.service, and
2359 systemd-logind have been updated to this type.
2361 * Initrd environments which are not on a pure memory file system (e.g.
2362 overlayfs combination as opposed to tmpfs) are now supported. With
2363 this change, during the initrd → host transition ("switch root")
2364 systemd will erase all files of the initrd only when the initrd is
2365 backed by a memory file system such as tmpfs.
2367 * New per-unit MemoryZSwapMax= option has been added to configure
2368 memory.zswap.max cgroup properties (the maximum amount of zswap
2371 * A new LogFilterPatterns= option has been added for units. It may be
2372 used to specify accept/deny regular expressions for log messages
2373 generated by the unit, that shall be enforced by systemd-journald.
2374 Rejected messages are neither stored in the journal nor forwarded.
2375 This option may be used to suppress noisy or uninteresting messages
2378 * The manager has a new
2379 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
2380 query process ownership via a PIDFD, which is more resilient against
2381 PID recycling issues.
2383 * Scope units now support OOMPolicy=. Login session scopes default to
2384 OOMPolicy=continue, allowing login scopes to survive the OOM killer
2385 terminating some processes in the scope.
2387 * systemd-fstab-generator now supports x-systemd.makefs option for
2388 /sysroot/ (in the initrd).
2390 * The maximum rate at which daemon reloads are executed can now be
2391 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
2392 options. (Or the equivalent on the kernel command line:
2393 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
2394 addition, systemd now logs the originating unit and PID when a reload
2395 request is received over D-Bus.
2397 * When enabling a swap device systemd will now reinitialize the device
2398 when the page size of the swap space does not match the page size of
2399 the running kernel. Note that this requires the 'swapon' utility to
2400 provide the '--fixpgsz' option, as implemented by util-linux, and it
2401 is not supported by busybox at the time of writing.
2403 * systemd now executes generator programs in a mount namespace
2404 "sandbox" with most of the file system read-only and write access
2405 restricted to the output directories, and with a temporary /tmp/
2406 mount provided. This provides a safeguard against programming errors
2407 in the generators, but also fixes here-docs in shells, which
2408 previously didn't work in early boot when /tmp/ wasn't available
2409 yet. (This feature has no security implications, because the code is
2410 still privileged and can trivially exit the sandbox.)
2412 * The system manager will now parse a new "vmm.notify_socket"
2413 system credential, which may be supplied to a VM via SMBIOS. If
2414 found, the manager will send a "READY=1" notification on the
2415 specified socket after boot is complete. This allows readiness
2416 notification to be sent from a VM guest to the VM host over a VSOCK
2419 * The sample PAM configuration file for systemd-user@.service now
2420 includes a call to pam_namespace. This puts children of user@.service
2421 in the expected namespace. (Many distributions replace their file
2422 with something custom, so this change has limited effect.)
2424 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
2425 can be used to override the mount units burst late limit for
2426 parsing '/proc/self/mountinfo', which was introduced in v249.
2429 * Drop-ins for init.scope changing control group resource limits are
2430 now applied, while they were previously ignored.
2432 * New build-time configuration options '-Ddefault-timeout-sec=' and
2433 '-Ddefault-user-timeout-sec=' have been added, to let distributions
2434 choose the default timeout for starting/stopping/aborting system and
2435 user units respectively.
2437 * Service units gained a new setting OpenFile= which may be used to
2438 open arbitrary files in the file system (or connect to arbitrary
2439 AF_UNIX sockets in the file system), and pass the open file
2440 descriptor to the invoked process via the usual file descriptor
2441 passing protocol. This is useful to give unprivileged services access
2442 to select files which have restrictive access modes that would
2443 normally not allow this. It's also useful in case RootDirectory= or
2444 RootImage= is used to allow access to files from the host environment
2445 (which is after all not visible from the service if these two options
2450 * The new net naming scheme "v253" has been introduced. In the new
2451 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
2452 a PCI bus. This extends the coverage of predictable interface names
2453 in some embedded systems.
2455 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
2456 a more informative path on some embedded systems.
2458 * Partition block devices will now also get symlinks in
2459 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
2460 block device nodes via the kernel's "diskseq" value. Previously those
2461 symlinks were only created for the main block device.
2463 * A new operator '-=' is supported for SYMLINK variables. This allows
2464 symlinks to be unconfigured even if an earlier rule added them.
2466 * 'udevadm --trigger --settle' now also works for network devices
2467 that are being renamed.
2469 Changes in sd-boot, bootctl, and the Boot Loader Specification:
2471 * systemd-boot now passes its random seed directly to the kernel's RNG
2472 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
2473 means the RNG gets seeded very early in boot before userspace has
2476 * systemd-boot will pass a disk-backed random seed – even when secure
2477 boot is enabled – if it can additionally get a random seed from EFI
2478 itself (via EFI's RNG protocol), or a prior seed in
2479 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
2481 * systemd-boot-system-token.service was renamed to
2482 systemd-boot-random-seed.service and extended to always save a random
2483 seed to ESP on every boot when a compatible boot loader is used. This
2484 allows a refreshed random seed to be used in the boot loader.
2486 * systemd-boot handles various seed inputs using a domain- and
2487 field-separated hashing scheme.
2489 * systemd-boot's 'random-seed-mode' option has been removed. A system
2490 token is now always required to be present for random seeds to be
2493 * systemd-boot now supports being loaded from other locations than the
2494 ESP, for example for direct kernel boot under QEMU or when embedded
2497 * systemd-boot now parses SMBIOS information to detect
2498 virtualization. This information is used to skip some warnings which
2499 are not useful in a VM and to conditionalize other aspects of
2502 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
2503 Secure Boot automated certificate enrollment from the ESP only if it
2504 is considered 'safe' to do so. At the moment 'safe' means running in
2507 * systemd-stub now processes random seeds in the same way as
2508 systemd-boot already does, in case a unified kernel image is being
2509 used from a different bootloader than systemd-boot, or without any
2512 * bootctl will now generate a system token on all EFI systems, even
2513 virtualized ones, and is activated in the case that the system token
2514 is missing from either sd-boot and sd-stub booted systems.
2516 * bootctl now implements two new verbs: 'kernel-identify' prints the
2517 type of a kernel image file, and 'kernel-inspect' provides
2518 information about the embedded command line and kernel version of
2521 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
2522 as for kernel-install.
2524 * The JSON output of "bootctl list" will now contain two more fields:
2525 isDefault and isSelected are boolean fields set to true on the
2526 default and currently booted boot menu entries.
2528 * bootctl gained a new verb "unlink" for removing a boot loader entry
2529 type #1 file from disk in a safe and robust way.
2531 * bootctl also gained a new verb "cleanup" that automatically removes
2532 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
2533 is not referenced anymore by any installed Type #1 boot loader
2534 specification entry. This is particularly useful in environments where
2535 a large number of entries reference the same or partly the same
2536 resources (for example, for snapshot-based setups).
2538 Changes in kernel-install:
2540 * A new "installation layout" can be configured as layout=uki. With
2541 this setting, a Boot Loader Specification Type#1 entry will not be
2542 created. Instead, a new kernel-install plugin 90-uki-copy.install
2543 will copy any .efi files from the staging area into the boot
2544 partition. A plugin to generate the UKI .efi file must be provided
2547 Changes in systemctl:
2549 * 'systemctl reboot' has dropped support for accepting a positional
2550 argument as the argument to the reboot(2) syscall. Please use the
2551 --reboot-argument= option instead.
2553 * 'systemctl disable' will now warn when called on units without
2554 install information. A new --no-warn option has been added that
2555 silences this warning.
2557 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
2558 of the drop-in to edit. (Previously, 'override.conf' was always
2561 * 'systemctl list-dependencies' now respects --type= and --state=.
2563 * 'systemctl kexec' now supports XEN VMM environments.
2565 * 'systemctl edit' will now tell the invoked editor to jump into the
2566 first line with actual unit file data, skipping over synthesized
2569 Changes in systemd-networkd and related tools:
2571 * The [DHCPv4] section in .network file gained new SocketPriority=
2572 setting that assigns the Linux socket priority used by the DHCPv4 raw
2573 socket. This may be used in conjunction with the
2574 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
2575 desired ethernet 802.1Q frame priority for DHCPv4 initial
2576 packets. This cannot be achieved with netfilter mangle tables because
2577 of the raw socket bypass.
2579 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
2580 new QuickAck= boolean setting that enables the TCP quick ACK mode for
2581 the routes configured by the acquired DHCPv4 lease or received router
2582 advertisements (RAs).
2584 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
2585 routes) now accepts three values, for high, medium, and low preference
2586 of the router (which can be set with the RouterPreference=) setting.
2588 * systemd-networkd-wait-online now supports matching via alternative
2591 * The [DHCPv6] section in .network file gained new SendRelease=
2592 setting which enables the DHCPv6 client to send release when
2593 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
2594 It is enabled by default.
2596 * If the Address= setting in [Network] or [Address] sections in .network
2597 specified without its prefix length, then now systemd-networkd assumes
2598 /32 for IPv4 or /128 for IPv6 addresses.
2600 * networkctl shows network and link file dropins in status output.
2602 Changes in systemd-dissect:
2604 * systemd-dissect gained a new option --list, to print the paths of
2605 all files and directories in a DDI.
2607 * systemd-dissect gained a new option --mtree, to generate a file
2608 manifest compatible with BSD mtree(5) of a DDI
2610 * systemd-dissect gained a new option --with, to execute a command with
2611 the specified DDI temporarily mounted and used as working
2612 directory. This is for example useful to convert a DDI to "tar"
2613 simply by running it within a "systemd-dissect --with" invocation.
2615 * systemd-dissect gained a new option --discover, to search for
2616 Discoverable Disk Images (DDIs) in well-known directories of the
2617 system. This will list machine, portable service and system extension
2620 * systemd-dissect now understands 2nd stage initrd images stored as a
2621 Discoverable Disk Image (DDI).
2623 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
2624 disk UUID stored in the GPT header) among the other data it can show.
2626 * systemd-dissect gained a new --in-memory switch to operate on an
2627 in-memory copy of the specified DDI file. This is useful to access a
2628 DDI with write access without persisting any changes. It's also
2629 useful for accessing a DDI without keeping the originating file
2632 * The DDI dissection logic will now automatically detect the intended
2633 sector size of disk images stored in files, based on the GPT
2634 partition table arrangement. Loopback block devices for such DDIs
2635 will then be configured automatically for the right sector size. This
2636 is useful to make dealing with modern 4K sector size DDIs fully
2637 automatic. The systemd-dissect tool will now show the detected sector
2638 size among the other DDI information in its output.
2640 Changes in systemd-repart:
2642 * systemd-repart gained new options --include-partitions= and
2643 --exclude-partitions= to filter operation on partitions by type UUID.
2644 This allows systemd-repart to be used to build images in which the
2645 type of one partition is set based on the contents of another
2646 partition (for example when the boot partition shall include a verity
2647 hash of the root partition).
2649 * systemd-repart also gained a --defer-partitions= option that is
2650 similar to --exclude-partitions=, but the size of the partition is
2651 still taken into account when sizing partitions, but without
2654 * systemd-repart gained a new --sector-size= option to specify what
2655 sector size should be used when an image is created.
2657 * systemd-repart now supports generating erofs file systems via
2658 CopyFiles= (a read-only file system similar to squashfs).
2660 * The Minimize= option was extended to accept "best" (which means the
2661 most minimal image possible, but may require multiple attempts) and
2662 "guess" (which means a reasonably small image).
2664 * The systemd-growfs binary now comes with a regular unit file template
2665 systemd-growfs@.service which can be instantiated directly for any
2666 desired file system. (Previously, the unit was generated dynamically
2667 by various generators, but no regular unit file template was
2670 Changes in journal tools:
2672 * Various systemd tools will append extra fields to log messages when
2673 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
2674 this includes information about D-Bus messages when sd-bus is used,
2675 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
2676 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
2677 Details of what is logged and when are subject to change.
2679 * The systemd-journald-audit.socket can now be disabled via the usual
2680 "systemctl disable" mechanism to stop collection of audit
2681 messages. Please note that it is not enabled statically anymore and
2682 must be handled by the preset/enablement logic in package
2683 installation scripts.
2685 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
2686 be used to curtail disk use by systemd-journal-remote. This is
2687 similar to the options supported by systemd-journald.
2689 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
2692 * When enrolling new keys systemd-cryptenroll now supports unlocking
2693 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
2694 password was strictly required to be specified.
2696 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
2697 (except for tokens with user verification, UV) to identify tokens
2698 before authentication. Multiple FIDO2 tokens can now be enrolled at
2699 the same time, and systemd-cryptsetup will automatically select one
2700 that corresponds to one of the available LUKS key slots.
2702 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
2703 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
2704 bank and number into which the volume key should be measured. This is
2705 automatically enabled for the encrypted root volume discovered and
2706 activated by systemd-gpt-auto-generator.
2708 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
2709 "noexec,nosuid,nodev".
2711 * systemd-gpt-auto-generator will now honour the rootfstype= and
2712 rootflags= kernel command line switches for root file systems it
2713 discovers, to match behaviour in case an explicit root fs is
2714 specified via root=.
2716 * systemd-pcrphase gained new options --machine-id and --file-system=
2717 to measure the machine-id and mount point information into PCR 15.
2718 New service unit files systemd-pcrmachine.service and
2719 systemd-pcrfs@.service have been added that invoke the tool with
2720 these switches during early boot.
2722 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
2723 with a success exit code even if no TPM device is detected.
2725 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
2726 making it harder to brute-force.
2728 Changes in other tools:
2730 * systemd-homed gained support for luksPbkdfForceIterations (the
2731 intended number of iterations for the PBKDF operation on LUKS).
2733 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
2734 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
2735 may now be used to specify additional arguments for mkfs when
2736 systemd-homed formats a file system.
2738 * systemd-hostnamed now exports the contents of
2739 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
2740 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
2741 unprivileged code to access those values.
2743 systemd-hostnamed also exports the SUPPORT_END= field from
2744 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
2745 this to show the status of the installed system.
2747 * systemd-measure gained an --append= option to sign multiple phase
2748 paths with different signing keys. This allows secrets to be
2749 accessible only in certain parts of the boot sequence. Note that
2750 'ukify' provides similar functionality in a more accessible form.
2752 * systemd-timesyncd will now write a structured log message with
2753 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
2754 on a on-disk timestamp, similarly to what it did when reaching
2755 synchronization via NTP.
2757 * systemd-timesyncd will now update the on-disk timestamp file on each
2758 boot at least once, making it more likely that the system time
2759 increases in subsequent boots.
2761 * systemd-vconsole-setup gained support for system/service credentials:
2762 vconsole.keymap/vconsole.keymap_toggle and
2763 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
2764 the similarly-named options in vconsole.conf.
2766 * systemd-localed will now save the XKB keyboard configuration to
2767 /etc/vconsole.conf, and also read it from there with a higher
2768 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
2769 file. Previously, this information was stored in the former file in
2770 converted form, and only in latter file in the original form. Tools
2771 which want to access keyboard configuration can now do so from a
2774 * systemd-resolved gained support for configuring the nameservers and
2775 search domains via kernel command line (nameserver=, domain=) and
2776 credentials (network.dns, network.search_domains).
2778 * systemd-resolved will now synthesize host names for the DNS stub
2779 addresses it supports. Specifically when "_localdnsstub" is resolved,
2780 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
2781 127.0.0.54 is returned.
2783 * systemd-notify will now send a "RELOADING=1" notification when called
2784 with --reloading, and "STOPPING=1" when called with --stopping. This
2785 can be used to implement notifications from units where it's easier
2786 to call a program than to use the sd-daemon library.
2788 * systemd-analyze's 'plot' command can now output its information in
2789 JSON, controlled via the --json= switch. Also, new --table, and
2790 --no-legend options have been added.
2792 * 'machinectl enable' will now automatically enable machines.target
2793 unit in addition to adding the machine unit to the target.
2795 Similarly, 'machinectl start|stop' gained a --now option to enable or
2796 disable the machine unit when starting or stopping it.
2798 * systemd-sysusers will now create /etc/ if it is missing.
2800 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
2801 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
2802 added to provide the new initial value for the new automated battery
2803 estimation functionality. If 'HibernateDelaySec=' is set to any value,
2804 the automated estimate (and thus the automated hibernation on low
2805 battery to avoid data loss) functionality will be disabled.
2807 * Default tmpfiles.d/ configuration will now automatically create
2808 credentials storage directory '/etc/credstore/' with the appropriate,
2809 secure permissions. If '/run/credstore/' exists, its permissions will
2810 be fixed too in case they are not correct.
2812 Changes in libsystemd and shared code:
2814 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
2815 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
2817 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
2818 128-bit ID in files such as /etc/machine-id has an invalid
2819 format. They also accept NULL as output parameter in more places,
2820 which is useful when the caller only wants to validate the inputs and
2821 does not need the output value.
2823 * sd-login gained new functions sd_pidfd_get_session(),
2824 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
2825 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
2826 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
2827 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
2828 but accept a PIDFD instead of a PID.
2830 * sd-path (and systemd-path) now export four new paths:
2831 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
2832 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
2833 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
2834 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
2836 * sd_notify() now supports AF_VSOCK as transport for notification
2837 messages (in addition to the existing AF_UNIX support). This is
2838 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
2840 * Detection of chroot() environments now works if /proc/ is not
2841 mounted. This affects systemd-detect-virt --chroot, but also means
2842 that systemd tools will silently skip various operations in such an
2845 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
2846 virtualization is now detected.
2848 Changes in the build system:
2850 * Standalone variants of systemd-repart and systemd-shutdown may now be
2851 built (if -Dstandalone=true).
2853 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
2854 example, allow scripts to conditionalize execution on AC power
2857 * The libp11kit library is now loaded through dlopen(3).
2859 Changes in the documentation:
2861 * Specifications that are not closely tied to systemd have moved to
2862 https://uapi-group.org/specifications/: the Boot Loader Specification
2863 and the Discoverable Partitions Specification.
2865 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
2866 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
2867 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
2868 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
2869 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
2870 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
2871 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
2872 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
2873 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
2874 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
2875 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
2876 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
2877 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
2878 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
2879 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
2880 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
2881 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
2882 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
2883 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
2884 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
2885 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
2886 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
2887 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
2888 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
2889 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
2890 reuben olinsky, Richard E. van der Luit, Richard Phibel,
2891 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
2892 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
2893 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
2894 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
2895 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
2896 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
2897 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
2898 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
2901 — Warsaw, 2023-02-15
2905 Announcements of Future Feature Removals:
2907 * We intend to remove cgroup v1 support from systemd release after the
2908 end of 2023. If you run services that make explicit use of cgroup v1
2909 features (i.e. the "legacy hierarchy" with separate hierarchies for
2910 each controller), please implement compatibility with cgroup v2 (i.e.
2911 the "unified hierarchy") sooner rather than later. Most of Linux
2912 userspace has been ported over already.
2914 * We intend to remove support for split-usr (/usr mounted separately
2915 during boot) and unmerged-usr (parallel directories /bin and
2916 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
2917 half of 2023, in the first release that falls into that time window.
2918 For more details, see:
2919 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
2921 Compatibility Breaks:
2923 * ConditionKernelVersion= checks that use the '=' or '!=' operators
2924 will now do simple string comparisons (instead of version comparisons
2925 à la stverscmp()). Version comparisons are still done for the
2926 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
2927 specified, a shell-style glob match is now done. This creates a minor
2928 incompatibility compared to older systemd versions when the '*', '?',
2929 '[', ']' characters are used, as these will now match as shell globs
2930 instead of literally. Given that kernel version strings typically do
2931 not include these characters we expect little breakage through this
2934 * The service manager will now read the SELinux label used for SELinux
2935 access checks from the unit file at the time it loads the file.
2936 Previously, the label would be read at the moment of the access
2937 check, which was problematic since at that time the unit file might
2938 already have been updated or removed.
2942 * systemd-measure is a new tool for calculating and signing expected
2943 TPM2 PCR values for a given unified kernel image (UKI) booted via
2944 sd-stub. The public key used for the signature and the signed
2945 expected PCR information can be embedded inside the UKI. This
2946 information can be extracted from the UKI by external tools and code
2947 in the image itself and is made available to userspace in the booted
2950 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
2951 updated to make use of this information if available in the booted
2952 kernel: when locking an encrypted volume/credential to the TPM
2953 systemd-cryptenroll/systemd-creds will use the public key to bind the
2954 volume/credential to any kernel that carries PCR information signed
2955 by the same key pair. When unlocking such volumes/credentials
2956 systemd-cryptsetup/systemd-creds will use the signature embedded in
2957 the booted UKI to gain access.
2959 Binding TPM-based disk encryption to public keys/signatures of PCR
2960 values — instead of literal PCR values — addresses the inherent
2961 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
2962 disks remain accessible even if the UKI is updated, without any TPM
2963 specific preparation during the OS update — as long as each UKI
2964 carries the necessary PCR signature information.
2966 Net effect: if you boot a properly prepared kernel, TPM-bound disk
2967 encryption now defaults to be locked to kernels which carry PCR
2968 signatures from the same key pair. Example: if a hypothetical distro
2969 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
2970 by default – bound to only FooOS kernels, and encrypted volumes bound
2971 to the TPM cannot be unlocked on kernels from other sources. (But do
2972 note this behaviour requires preparation/enabling in the UKI, and of
2973 course users can always enroll non-TPM ways to unlock the volume.)
2975 * systemd-pcrphase is a new tool that is invoked at six places during
2976 system runtime, and measures additional words into TPM2 PCR 11, to
2977 mark milestones of the boot process. This allows binding access to
2978 specific TPM2-encrypted secrets to specific phases of the boot
2979 process. (Example: LUKS2 disk encryption key only accessible in the
2980 initrd, but not later.)
2982 Changes in systemd itself, i.e. the manager and units
2984 * The cpu controller is delegated to user manager units by default, and
2985 CPUWeight= settings are applied to the top-level user slice units
2986 (app.slice, background.slice, session.slice). This provides a degree
2987 of resource isolation between different user services competing for
2990 * Systemd can optionally do a full preset in the "first boot" condition
2991 (instead of just enable-only). This behaviour is controlled by the
2992 compile-time option -Dfirst-boot-full-preset. Right now it defaults
2993 to 'false', but the plan is to switch it to 'true' for the subsequent
2996 * Drop-ins are now allowed for transient units too.
2998 * Systemd will set the taint flag 'support-ended' if it detects that
2999 the OS image is past its end-of-support date. This date is declared
3000 in a new /etc/os-release field SUPPORT_END= described below.
3002 * Two new settings ConditionCredential= and AssertCredential= can be
3003 used to skip or fail units if a certain system credential is not
3006 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
3008 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
3009 specify the SMACK security label to use when not specified in a unit
3012 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
3013 specify the default timeout when waiting for device units to
3016 * C.UTF-8 is used as the default locale if nothing else has been
3019 * [Condition|Assert]Firmware= have been extended to support certain
3020 SMBIOS fields. For example
3022 ConditionFirmware=smbios-field(board_name = "Custom Board")
3024 conditionalizes the unit to run only when
3025 /sys/class/dmi/id/board_name contains "Custom Board" (without the
3028 * ConditionFirstBoot= now correctly evaluates as true only during the
3029 boot phase of the first boot. A unit executed later, after booting
3030 has completed, will no longer evaluate this condition as true.
3032 * Socket units will now create sockets in the SELinuxContext= of the
3033 associated service unit, if any.
3035 * Boot phase transitions (start initrd → exit initrd → boot complete →
3036 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
3037 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
3038 unsealed only in the initrd.
3040 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
3041 also be provided to ExecStartPre= processes.
3043 * Various units are now correctly ordered against
3044 initrd-switch-root.target where previously a conflict without
3045 ordering was configured. A stop job for those units would be queued,
3046 but without the ordering it could be executed only after
3047 initrd-switch-root.service, leading to units not being restarted in
3048 the host system as expected.
3050 * In order to fully support the IPMI watchdog driver, which has not yet
3051 been ported to the new common watchdog device interface,
3052 /dev/watchdog0 will be tried first and systemd will silently fallback
3053 to /dev/watchdog if it is not found.
3055 * New watchdog-related D-Bus properties are now published by systemd:
3056 WatchdogDevice, WatchdogLastPingTimestamp,
3057 WatchdogLastPingTimestampMonotonic.
3059 * At shutdown, API virtual files systems (proc, sys, etc.) will be
3062 * At shutdown, systemd will now log about processes blocking unmounting
3065 * A new meson build option 'clock-valid-range-usec-max' was added to
3066 allow disabling system time correction if RTC returns a timestamp far
3069 * Propagated restart jobs will no longer be discarded while a unit is
3072 * PID 1 will now import system credentials from SMBIOS Type 11 fields
3073 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
3074 simple, fast and generic path for supplying credentials to a VM,
3075 without involving external tools such as cloud-init/ignition.
3077 * The CPUWeight= setting of unit files now accepts a new special value
3078 "idle", which configures "idle" level scheduling for the unit.
3080 * Service processes that are activated due to a .timer or .path unit
3081 triggering will now receive information about this via environment
3082 variables. Note that this is information is lossy, as activation
3083 might be coalesced and only one of the activating triggers will be
3084 reported. This is hence more suited for debugging or tracing rather
3085 than for behaviour decisions.
3087 * The riscv_flush_icache(2) system call has been added to the list of
3088 system calls allowed by default when SystemCallFilter= is used.
3090 * The selinux context derived from the target executable, instead of
3091 'init_t' used for the manager itself, is now used when creating
3092 listening sockets for units that specify SELinuxContextFromNet=yes.
3094 Changes in sd-boot, bootctl, and the Boot Loader Specification:
3096 * The Boot Loader Specification has been cleaned up and clarified.
3097 Various corner cases in version string comparisons have been fixed
3098 (e.g. comparisons for empty strings). Boot counting is now part of
3099 the main specification.
3101 * New PCRs measurements are performed during boot: PCR 11 for the
3102 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
3103 took place this is now reported to userspace via the new
3104 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
3106 * As before, systemd-stub will measure kernel parameters and system
3107 credentials into PCR 12. It will now report this fact via the
3108 StubPcrKernelParameters EFI variable to userspace.
3110 * The UEFI monotonic boot counter is now included in the updated random
3111 seed file maintained by sd-boot, providing some additional entropy.
3113 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
3114 of arranging the image manually and jumping to the kernel entry
3115 point. sd-stub also installs a temporary UEFI SecurityOverride to
3116 allow the (unsigned) nested image to be booted. This is safe because
3117 the outer (signed) stub+kernel binary must have been verified before
3118 the stub was executed.
3120 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
3121 is now supported by sd-boot.
3123 * bootctl gained a bunch of new options: --all-architectures to install
3124 binaries for all supported EFI architectures, --root= and --image=
3125 options to operate on a directory or disk image, and
3126 --install-source= to specify the source for binaries to install,
3127 --efi-boot-option-description= to control the name of the boot entry.
3129 * The sd-boot stub exports a StubFeatures flag, which is used by
3130 bootctl to show features supported by the stub that was used to boot.
3132 * The PE section offsets that are used by tools that assemble unified
3133 kernel images have historically been hard-coded. This may lead to
3134 overlapping PE sections which may break on boot. The UKI will now try
3135 to detect and warn about this.
3137 Any tools that assemble UKIs must update to calculate these offsets
3138 dynamically. Future sd-stub versions may use offsets that will not
3139 work with the currently used set of hard-coded offsets!
3141 * sd-stub now accepts (and passes to the initrd and then to the full
3142 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
3143 signatures of expected PCR values, to allow sealing secrets via the
3144 TPM2 against pre-calculated PCR measurements.
3146 Changes in the hardware database:
3148 * 'systemd-hwdb query' now supports the --root= option.
3150 Changes in systemctl:
3152 * systemctl now supports --state= and --type= options for the 'show'
3155 * systemctl gained a new verb 'list-automounts' to list automount
3158 * systemctl gained support for a new --image= switch to be able to
3159 operate on the specified disk image (similar to the existing --root=
3160 which operates relative to some directory).
3162 Changes in systemd-networkd:
3164 * networkd can set Linux NetLabel labels for integration with the
3165 network control in security modules via a new NetLabel= option.
3167 * The RapidCommit= is (re-)introduced to enable faster configuration
3168 via DHCPv6 (RFC 3315).
3170 * networkd gained a new option TCPCongestionControlAlgorithm= that
3171 allows setting a per-route TCP algorithm.
3173 * networkd gained a new option KeepFileDescriptor= to allow keeping a
3174 reference (file descriptor) open on TUN/TAP interfaces, which is
3175 useful to avoid link flaps while the underlying service providing the
3176 interface is being serviced.
3178 * RouteTable= now also accepts route table names.
3180 Changes in systemd-nspawn:
3182 * The --bind= and --overlay= options now support relative paths.
3184 * The --bind= option now supports a 'rootidmap' value, which will
3185 use id-mapped mounts to map the root user inside the container to the
3186 owner of the mounted directory on the host.
3188 Changes in systemd-resolved:
3190 * systemd-resolved now persists DNSOverTLS in its state file too. This
3191 fixes a problem when used in combination with NetworkManager, which
3192 sends the setting only once, causing it to be lost if resolved was
3193 restarted at any point.
3195 * systemd-resolved now exposes a Varlink socket at
3196 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
3197 root. Processed DNS requests in a JSON format will be published to
3198 any clients connected to this socket.
3200 resolvectl gained a 'monitor' verb to make use of this.
3202 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
3203 instead of returning SERVFAIL, as per RFC:
3204 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
3206 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
3207 is still supported.)
3209 Changes in libsystemd and other libraries:
3211 * libsystemd now exports sd_bus_error_setfv() (a convenience function
3212 for setting bus errors), sd_id128_string_equal (a convenience
3213 function for 128-bit ID string comparisons), and
3214 sd_bus_message_read_strv_extend() (a function to incrementally read
3217 * libsystemd now exports sd_device_get_child_first()/_next() as a
3218 high-level interface for enumerating child devices. It also supports
3219 sd_device_new_child() for opening a child device given a device
3222 * libsystemd now exports sd_device_monitor_set()/get_description()
3223 which allow setting a custom description that will be used in log
3224 messages by sd_device_monitor*.
3226 * Private shared libraries (libsystemd-shared-nnn.so,
3227 libsystemd-core-nnn.so) are now installed into arch-specific
3228 directories to allow multi-arch installs.
3230 * A new sd-gpt.h header is now published, listing GUIDs from the
3231 Discoverable Partitions specification. For more details see:
3232 https://systemd.io/DISCOVERABLE_PARTITIONS/
3234 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
3235 database given an explicit path to the file.
3237 * The signal number argument to sd_event_add_signal() now can now be
3238 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
3239 be automatically invoked to block the specified signal. This is
3240 useful to simplify invocations as the caller doesn't have to do this
3243 * A new convenience call sd_event_set_signal_exit() has been added to
3244 sd-event to set up signal handling so that the event loop
3245 automatically terminates cleanly on SIGTERM/SIGINT.
3247 Changes in other components:
3249 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
3250 can now be provided via the credential mechanism.
3252 * systemd-analyze gained a new verb 'compare-versions' that implements
3253 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
3254 'dpkg --compare-versions').
3256 * 'systemd-analyze dump' is extended to accept glob patterns for unit
3257 names to limit the output to matching units.
3259 * tmpfiles.d/ lines can read file contents to write from a credential.
3260 The new modifier char '^' is used to specify that the argument is a
3261 credential name. This mechanism is used to automatically populate
3262 /etc/motd, /etc/issue, and /etc/hosts from credentials.
3264 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
3265 an inode if the specification is prefixed with ':' and the inode
3268 * Default tmpfiles.d/ configuration now carries a line to automatically
3269 use an 'ssh.authorized_keys.root' credential if provided to set up
3270 the SSH authorized_keys file for the root user.
3272 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
3275 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
3276 in base64. This is useful to write arbitrary binary data into files.
3278 * The pkgconfig and rpm macros files now export the directory for user
3279 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
3281 * Detection of Apple Virtualization and detection of Parallels and
3282 KubeVirt virtualization on non-x86 archs have been added.
3284 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
3285 user when their system will become unsupported.
3287 * When performing suspend-then-hibernate, the system will estimate the
3288 discharge rate and use that to set the delay until hibernation and
3289 hibernate immediately instead of suspending when running from a
3290 battery and the capacity is below 5%.
3292 * systemd-sysctl gained a --strict option to fail when a sysctl
3293 setting is unknown to the kernel.
3295 * machinectl supports --force for the 'copy-to' and 'copy-from'
3298 * coredumpctl gained the --root and --image options to look for journal
3299 files under the specified root directory, image, or block device.
3301 * 'journalctl -o' and similar commands now implement a new output mode
3302 "short-delta". It is similar to "short-monotonic", but also shows the
3303 time delta between subsequent messages.
3305 * journalctl now respects the --quiet flag when verifying consistency
3308 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
3309 will indicate whether a message was logged in the 'initrd' phase or
3310 in the 'system' phase of the boot process.
3312 * Journal files gained a new compatibility flag
3313 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
3314 to the storage format that allow reducing size on disk. As with other
3315 compatibility flags, older journalctl versions will not be able to
3316 read journal files using this new format. The environment variable
3317 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
3318 disable this functionality. It is enabled by default.
3320 * systemd-run's --working-directory= switch now works when used in
3321 combination with --scope.
3323 * portablectl gained a --force flag to skip certain sanity checks. This
3324 is implemented using new flags accepted by systemd-portabled for the
3325 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
3326 flag now means that the attach/detach checks whether the units are
3327 already present and running will be skipped. Similarly,
3328 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
3329 image name matches the name declared inside of the image will be
3330 skipped. Callers must be sure to do those checks themselves if
3333 * systemd-portabled will now use the original filename to check
3334 extension-release.NAME for correctness, in case it is passed a
3337 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
3340 * sysext's extension-release files now support '_any' as a special
3341 value for the ID= field, to allow distribution-independent extensions
3342 (e.g.: fully statically compiled binaries, scripts). It also gained
3343 support for a new ARCHITECTURE= field that may be used to explicitly
3344 restrict an image to hosts of a specific architecture.
3346 * systemd-repart now supports creating squashfs partitions. This
3347 requires mksquashfs from squashfs-tools.
3349 * systemd-repart gained a --split flag to also generate split
3350 artifacts, i.e. a separate file for each partition. This is useful in
3351 conjunction with systemd-sysupdate or other tools, or to generate
3352 split dm-verity artifacts.
3354 * systemd-repart is now able to generate dm-verity partitions, including
3357 * systemd-repart can now set a partition UUID to zero, allowing it to
3358 be filled in later, such as when using verity partitions.
3360 * systemd-repart now supports drop-ins for its configuration files.
3362 * Package metadata logged by systemd-coredump in the system journal is
3365 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
3367 * systemd-oomd now automatically links against libatomic, if available.
3369 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
3372 * scope units now also provide oom-kill status.
3374 * systemd-pstore will now try to load only the efi_pstore kernel module
3375 before running, ensuring that pstore can be used.
3377 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
3378 session after a preconfigure timeout.
3380 * systemd-homed will now wait up to 30 seconds for workers to terminate,
3381 rather than indefinitely.
3383 * homectl gained a new '--luks-sector-size=' flag that allows users to
3384 select the preferred LUKS sector size. Must be a power of 2 between 512
3385 and 4096. systemd-userdbd records gained a corresponding field.
3387 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
3388 variable when generating the 'sp_lstchg' field, to ensure an image
3389 build can be reproducible.
3391 * 'udevadm wait' will now listen to kernel uevents too when called with
3394 * When naming network devices udev will now consult the Devicetree
3395 "alias" fields for the device.
3397 * systemd-udev will now create infiniband/by-path and
3398 infiniband/by-ibdev links for Infiniband verbs devices.
3400 * systemd-udev-trigger.service will now also prioritize input devices.
3402 * ConditionACPower= and systemd-ac-power will now assume the system is
3403 running on AC power if no battery can be found.
3405 * All features and tools using the TPM2 will now communicate with it
3406 using a bind key. Beforehand, the tpm2 support used encrypted sessions
3407 by creating a primary key that was used to encrypt traffic. This
3408 creates a problem as the key created for encrypting the traffic could
3409 be faked by an active interposer on the bus. In cases when a pin is
3410 used, a bind key will be used. The pin is used as the auth value for
3411 the seal key, aka the disk encryption key, and that auth value will be
3412 used in the session establishment. An attacker would need the pin
3413 value to create the secure session and thus an active interposer
3414 without the pin cannot interpose on TPM2 traffic.
3416 * systemd-growfs no longer requires udev to run.
3418 * systemd-backlight now will better support systems with multiple
3421 * systemd-cryptsetup's keyfile-timeout= option now also works when a
3422 device is used as a keyfile.
3424 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
3425 unlocking key from a key file (instead of prompting the user). Note
3426 that this is the key for unlocking the volume in order to be able to
3427 enroll a new key, but it is not the key that is enrolled.
3429 * systemd-dissect gained a new --umount switch that will safely and
3430 synchronously unmount all partitions of an image previously mounted
3431 with 'systemd-dissect --mount'.
3433 * When using gcrypt, all systemd tools and services will now configure
3434 it to prefer the OS random number generator if present.
3436 * All example code shipped with documentation has been relicensed from CC0
3439 * Unit tests will no longer fail when running on a system without
3442 Experimental features:
3444 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
3445 and bpftool >= 7.0).
3447 * sd-boot can automatically enroll SecureBoot keys from files found on
3448 the ESP. This enrollment can be either automatic ('force' mode) or
3449 controlled by the user ('manual' mode). It is sufficient to place the
3450 SecureBoot keys in the right place in the ESP and they will be picked
3451 up by sd-boot and shown in the boot menu.
3453 * The mkosi config in systemd gained support for automatically
3454 compiling a kernel with the configuration appropriate for testing
3455 systemd. This may be useful when developing or testing systemd in
3456 tandem with the kernel.
3458 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
3459 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
3460 Alexander Graf, Alexander Shopov, Alexander Wilson,
3461 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
3462 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
3463 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
3464 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
3465 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
3466 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
3467 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
3468 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
3469 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
3470 David Jaša, David Rheinsberg, David Seifert, David Tardon,
3471 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
3472 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
3473 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
3474 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
3475 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
3476 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
3477 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
3478 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
3479 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
3480 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
3481 JeroenHD, jiangchuangang, João Loureiro,
3482 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
3483 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
3484 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
3485 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
3486 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
3487 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
3488 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
3489 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
3490 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
3491 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
3492 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
3493 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
3494 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
3495 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
3496 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
3497 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
3498 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
3499 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
3500 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
3501 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
3502 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
3503 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
3504 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
3505 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
3507 – The Great Beyond, 2022-10-31 👻
3511 Backwards-incompatible changes:
3513 * The minimum kernel version required has been bumped from 3.13 to 4.15,
3514 and CLOCK_BOOTTIME is now assumed to always exist.
3516 * C11 with GNU extensions (aka "gnu11") is now used to build our
3517 components. Public API headers are still restricted to ISO C89.
3519 * In v250, a systemd-networkd feature that automatically configures
3520 routes to addresses specified in AllowedIPs= was added and enabled by
3521 default. However, this causes network connectivity issues in many
3522 existing setups. Hence, it has been disabled by default since
3523 systemd-stable 250.3. The feature can still be used by explicitly
3524 configuring RouteTable= setting in .netdev files.
3526 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
3527 when a Condition*= check does not succeed, restoring the JobRemoved
3528 signal to the behaviour it had before v250.
3530 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
3531 GetImageMetadataWithExtensions() have been fixed to provide an extra
3532 return parameter, containing the actual extension release metadata.
3533 The current implementation was judged to be broken and unusable, and
3534 thus the usual procedure of adding a new set of methods was skipped,
3535 and backward compatibility broken instead on the assumption that
3536 nobody can be affected given the current state of this interface.
3538 * All kernels supported by systemd mix bytes returned by RDRAND (or
3539 similar) into the entropy pool at early boot. This means that on
3540 those systems, even if /dev/urandom is not yet initialized, it still
3541 returns bytes that are of at least RDRAND quality. For that reason,
3542 we no longer have reason to invoke RDRAND from systemd itself, which
3543 has historically been a source of bugs. Furthermore, kernels ≥5.6
3544 provide the getrandom(GRND_INSECURE) interface for returning random
3545 bytes before the entropy pool is initialized without warning into
3546 kmsg, which is what we attempt to use if available. systemd's direct
3547 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
3548 running an older kernel may experience kmsg warnings that were not
3549 seen with 250. For newer kernels, non-x86 systems, or older x86
3550 systems, there should be no visible changes.
3552 * sd-boot will now measure the kernel command line into TPM PCR 12
3553 rather than PCR 8. This improves usefulness of the measurements on
3554 systems where sd-boot is chainloaded from Grub. Grub measures all
3555 commands its executes into PCR 8, which makes it very hard to use
3556 reasonably, hence separate ourselves from that and use PCR 12
3557 instead, which is what certain Ubuntu editions already do. To retain
3558 compatibility with systems running older systemd systems a new meson
3559 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
3560 If enabled, the measurement is done twice: into the new-style PCR 12
3561 *and* the old-style PCR 8. It's strongly advised to migrate all users
3562 to PCR 12 for this purpose in the long run, as we intend to remove
3563 this compatibility feature in two years' time.
3565 * busctl capture now writes output in the newer pcapng format instead
3568 * A udev rule that imported hwdb matches for USB devices with lowercase
3569 hexadecimal vendor/product ID digits was added in systemd 250. This
3570 has been reverted, since uppercase hexadecimal digits are supposed to
3571 be used, and we already had a rule with the appropriate match.
3573 Users might need to adjust their local hwdb entries.
3575 * arch_prctl(2) has been moved to the @default set in the syscall filters
3576 (as exposed via the SystemCallFilter= setting in service unit files).
3577 It is apparently used by the linker now.
3579 * The tmpfiles entries that create the /run/systemd/netif directory and
3580 its subdirectories were moved from tmpfiles.d/systemd.conf to
3581 tmpfiles.d/systemd-network.conf.
3583 Users might need to adjust their files that override tmpfiles.d/systemd.conf
3584 to account for this change.
3586 * The requirement for Portable Services images to contain a well-formed
3587 os-release file (i.e.: contain at least an ID field) is now enforced.
3588 This applies to base images and extensions, and also to systemd-sysext.
3590 Changes in the Boot Loader Specification, kernel-install and sd-boot:
3592 * kernel-install's and bootctl's Boot Loader Specification Type #1
3593 entry generation logic has been reworked. The user may now pick
3594 explicitly by which "token" string to name the installation's boot
3595 entries, via the new /etc/kernel/entry-token file or the new
3596 --entry-token= switch to bootctl. By default — as before — the
3597 entries are named after the local machine ID. However, in "golden
3598 image" environments, where the machine ID shall be initialized on
3599 first boot (as opposed to at installation time before first boot) the
3600 machine ID will not be available at build time. In this case the
3601 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
3602 file) may be used to override the "token" for the entries, for
3603 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
3604 make the OS images independent of any machine ID, and ensure that the
3605 images will not carry any identifiable information before first boot,
3606 but on the other hand means that multiple parallel installations of
3607 the very same image on the same disk cannot be supported.
3609 Summary: if you are building golden images that shall acquire
3610 identity information exclusively on first boot, make sure to both
3611 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
3612 value of the IMAGE_ID= or ID= field of /etc/os-release or another
3613 suitable identifier before deploying the image.
3615 * The Boot Loader Specification has been extended with
3616 /loader/entries.srel file located in the EFI System Partition (ESP)
3617 that disambiguates the format of the entries in the /loader/entries/
3618 directory (in order to discern them from incompatible uses of this
3619 directory by other projects). For entries that follow the
3620 Specification, the string "type1" is stored in this file.
3622 bootctl will now write this file automatically when installing the
3623 systemd-boot boot loader.
3625 * kernel-install supports a new initrd_generator= setting in
3626 /etc/kernel/install.conf, that is exported as
3627 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
3628 allows choosing different initrd generators.
3630 * kernel-install will now create a "staging area" (an initially-empty
3631 directory to gather files for a Boot Loader Specification Type #1
3632 entry). The path to this directory is exported as
3633 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
3634 drop files there instead of writing them directly to the final
3635 location. kernel-install will move them when all files have been
3636 prepared successfully.
3638 * New option sort-key= has been added to the Boot Loader Specification
3639 to override the sorting order of the entries in the boot menu. It is
3640 read by sd-boot and bootctl, and will be written by kernel-install,
3641 with the default value of IMAGE_ID= or ID= fields from
3642 os-release. Together, this means that on multiboot installations,
3643 entries should be grouped and sorted in a predictable way.
3645 * The sort order of boot entries has been updated: entries which have
3646 the new field sort-key= are sorted by it first, and all entries
3647 without it are ordered later. After that, entries are sorted by
3648 version so that newest entries are towards the beginning of the list.
3650 * The kernel-install tool gained a new 'inspect' verb which shows the
3651 paths and other settings used.
3653 * sd-boot can now optionally beep when the menu is shown and menu
3654 entries are selected, which can be useful on machines without a
3655 working display. (Controllable via a loader.conf setting.)
3657 * The --make-machine-id-directory= switch to bootctl has been replaced
3658 by --make-entry-directory=, given that the entry directory is not
3659 necessarily named after the machine ID, but after some other suitable
3660 ID as selected via --entry-token= described above. The old name of
3661 the option is still understood to maximize compatibility.
3663 * 'bootctl list' gained support for a new --json= switch to output boot
3664 menu entries in JSON format.
3666 * 'bootctl is-installed' now supports the --graceful, and various verbs
3667 omit output with the new option --quiet.
3669 Changes in systemd-homed:
3671 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
3672 of activated home directories it manages (if the kernel and selected
3673 file systems support it). So far it mapped three UID ranges: the
3674 range from 0…60000, the user's own UID, and the range 60514…65534,
3675 leaving everything else unmapped (in other words, the 16-bit UID range
3676 is mapped almost fully, with the exception of the UID subrange used
3677 for systemd-homed users, with one exception: the user's own UID).
3678 Unmapped UIDs may not be used for file ownership in the home
3679 directory — any chown() attempts with them will fail. With this
3680 release a fourth range is added to these mappings:
3681 524288…1879048191. This range is the UID range intended for container
3684 https://systemd.io/UIDS-GIDS
3686 This range may be used for container managers that place container OS
3687 trees in the home directory (which is a questionable approach, for
3688 quota, permission, SUID handling and network file system
3689 compatibility reasons, but nonetheless apparently commonplace). Note
3690 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
3691 UID assignments from the range are not managed or mapped by
3692 `systemd-homed`, and must be managed with other mechanisms, in the
3693 context of the local system.
3695 Typically, a better approach to user namespacing in relevant
3696 container managers would be to leave container OS trees on disk at
3697 UID offset 0, but then map them to a dynamically allocated runtime
3698 UID range via another UID mount map at container invocation
3699 time. That way user namespace UID ranges become strictly a runtime
3700 concept, and do not leak into persistent file systems, persistent
3701 user databases or persistent configuration, thus greatly simplifying
3702 handling, and improving compatibility with home directories intended
3703 to be portable like the ones managed by systemd-homed.
3705 Changes in shared libraries:
3707 * A new libsystemd-core-<version>.so private shared library is
3708 installed under /usr/lib/systemd/system, mirroring the existing
3709 libsystemd-shared-<version>.so library. This allows the total
3710 installation size to be reduced by binary code reuse.
3712 * The <version> tag used in the name of libsystemd-shared.so and
3713 libsystemd-core.so can be configured via the meson option
3714 'shared-lib-tag'. Distributions may build subsequent versions of the
3715 systemd package with unique tags (e.g. the full package version),
3716 thus allowing multiple installations of those shared libraries to be
3717 available at the same time. This is intended to fix an issue where
3718 programs that link to those libraries would fail to execute because
3719 they were installed earlier or later than the appropriate version of
3722 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
3723 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
3724 format instead of as a simple series of hex characters.
3726 * The sd-device API gained two new calls sd_device_new_from_devname()
3727 and sd_device_new_from_path() which permit allocating an sd_device
3728 object from a device node name or file system path.
3730 * sd-device also gained a new call sd_device_open() which will open the
3731 device node associated with a device for which an sd_device object
3732 has been allocated. The call is supposed to address races around
3733 device nodes being removed/recycled due to hotplug events, or media
3734 change events: the call checks internally whether the major/minor of
3735 the device node and the "diskseq" (in case of block devices) match
3736 with the metadata loaded in the sd_device object, thus ensuring that
3737 the device once opened really matches the provided sd_device object.
3739 Changes in PID1, systemctl, and systemd-oomd:
3741 * A new set of service monitor environment variables will be passed to
3742 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
3743 handler unit as OnFailure=/OnSuccess=. The variables are:
3744 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
3745 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
3746 handler needs to watch multiple units, use a templated handler.
3748 * A new ExtensionDirectories= setting in service unit files allows
3749 system extensions to be loaded from a directory. (It is similar to
3750 ExtensionImages=, but takes paths to directories, instead of
3753 'portablectl attach --extension=' now also accepts directory paths.
3755 * The user.delegate and user.invocation_id extended attributes on
3756 cgroups are used in addition to trusted.delegate and
3757 trusted.invocation_id. The latter pair requires privileges to set,
3758 but the former doesn't and can be also set by the unprivileged user
3761 (Only supported on kernels ≥5.6.)
3763 * Units that were killed by systemd-oomd will now have a service result
3764 of 'oom-kill'. The number of times a service was killed is tallied
3765 in the 'user.oomd_ooms' extended attribute.
3767 The OOMPolicy= unit file setting is now also honoured by
3770 * In unit files the new %y/%Y specifiers can be used to refer to
3771 normalized unit file path, which is particularly useful for symlinked
3774 The new %q specifier resolves to the pretty hostname
3775 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
3777 The new %d specifier resolves to the credentials directory of a
3778 service (same as $CREDENTIALS_DIRECTORY).
3780 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
3781 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
3782 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
3783 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
3784 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
3785 MountFlags= service settings now also work in unprivileged user
3786 services, i.e. those run by the user's --user service manager, as long
3787 as user namespaces are enabled on the system.
3789 * Services with Restart=always and a failing ExecCondition= will no
3790 longer be restarted, to bring ExecCondition= behaviour in line with
3791 Condition*= settings.
3793 * LoadCredential= now accepts a directory as the argument; all files
3794 from the directory will be loaded as credentials.
3796 * A new D-Bus property ControlGroupId is now exposed on service units,
3797 that encapsulates the service's numeric cgroup ID that newer kernels
3798 assign to each cgroup.
3800 * PID 1 gained support for configuring the "pre-timeout" of watchdog
3801 devices and the associated governor, via the new
3802 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
3803 options in /etc/systemd/system.conf.
3805 * systemctl's --timestamp= option gained a new choice "unix", to show
3806 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
3808 * A new "taint" flag named "old-kernel" is introduced which is set when
3809 the kernel systemd runs on is older then the current baseline version
3810 (see above). The flag is shown in "systemctl status" output.
3812 * Two additional taint flags "short-uid-range" and "short-gid-range"
3813 have been added as well, which are set when systemd notices it is run
3814 within a userns namespace that does not define the full 0…65535 UID
3817 * A new "unmerged-usr" taint flag has been added that is set whenever
3818 running on systems where /bin/ + /sbin/ are *not* symlinks to their
3819 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
3822 * Generators invoked by PID 1 will now have a couple of useful
3823 environment variables set describing the execution context a
3824 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
3825 system service manager, or from the per-user service
3826 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
3827 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
3828 systemd considers the current boot to be a "first"
3829 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
3830 detected and which type of hypervisor/container
3831 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
3832 kernel is built for.
3834 * PID 1 will now automatically pick up system credentials from qemu's
3835 fw_cfg interface, thus allowing passing arbitrary data into VM
3836 systems similar to how this is already supported for passing them
3837 into `systemd-nspawn` containers. Credentials may now also be passed
3838 in via the new kernel command line option `systemd.set_credential=`
3839 (note that kernel command line options are world-readable during
3840 runtime, and only useful for credentials that require no
3841 confidentiality). The credentials that can be passed to unified
3842 kernels that use the `systemd-stub` UEFI stub are now similarly
3843 picked up automatically. Automatic importing of system credentials
3844 this way can be turned off via the new
3845 `systemd.import_credentials=no` kernel command line option.
3847 * LoadCredential= will now automatically look for credentials in the
3848 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
3849 the argument is not an absolute path. Similarly,
3850 LoadCredentialEncrypted= will check the same directories plus
3851 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
3852 /usr/lib/credstore.encrypted/. The idea is to use those directories
3853 as the system-wide location for credentials that services should pick
3856 * System and service credentials are described in great detail in a new
3859 https://systemd.io/CREDENTIALS
3861 Changes in systemd-journald:
3863 * The journal JSON export format has been added to listed of stable
3864 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
3866 * journalctl --list-boots now supports JSON output and the --reverse option.
3868 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
3869 updated, BUILDING_IMAGES is new:
3871 https://systemd.io/JOURNAL_EXPORT_FORMATS
3872 https://systemd.io/BUILDING_IMAGES
3876 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
3877 calculators, etc.), the other AV production devices (DJ tables,
3878 keypads, etc.) that should accessible to the seat owner user by
3881 * udevadm trigger gained a new --prioritized-subsystem= option to
3882 process certain subsystems (and all their parent devices) earlier.
3884 systemd-udev-trigger.service now uses this new option to trigger
3885 block and TPM devices first, hopefully making the boot a bit faster.
3887 * udevadm trigger now implements --type=all, --initialized-match,
3888 --initialized-nomatch to trigger both subsystems and devices, only
3889 already-initialized devices, and only devices which haven't been
3890 initialized yet, respectively.
3892 * udevadm gained a new "wait" command for safely waiting for a specific
3893 device to show up in the udev device database. This is useful in
3894 scripts that asynchronously allocate a block device (e.g. through
3895 repartitioning, or allocating a loopback device or similar) and need
3896 to synchronize on the creation to complete.
3898 * udevadm gained a new "lock" command for locking one or more block
3899 devices while formatting it or writing a partition table to it. It is
3900 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
3901 usable in scripts dealing with block devices.
3903 * udevadm info will show a couple of additional device fields in its
3904 output, and will not apply a limited set of coloring to line types.
3906 * udevadm info --tree will now show a tree of objects (i.e. devices and
3907 suchlike) in the /sys/ hierarchy.
3909 * Block devices will now get a new set of device symlinks in
3910 /dev/disk/by-diskseq/<nr>, which may be used to reference block
3911 device nodes via the kernel's "diskseq" value. Note that this does
3912 not guarantee that opening a device by a symlink like this will
3913 guarantee that the opened device actually matches the specified
3914 diskseq value. To be safe against races, the actual diskseq value of
3915 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
3916 the one in the symlink path.
3918 * .link files gained support for setting MDI/MID-X on a link.
3920 * .link files gained support for [Match] Firmware= setting to match on
3921 the device firmware description string. By mistake, it was previously
3922 only supported in .network files.
3924 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
3925 and [SR-IOV] section to configure SR-IOV virtual functions.
3927 Changes in systemd-networkd:
3929 * The default scope for unicast routes configured through [Route]
3930 section is changed to "link", to make the behavior consistent with
3931 "ip route" command. The manual configuration of [Route] Scope= is
3934 * A new unit systemd-networkd-wait-online@<interface>.service has been
3935 added that can be used to wait for a specific network interface to be
3938 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
3939 that configures the eponymous kernel attribute on the bridge.
3941 * .netdev files now can be used to create virtual WLAN devices, and
3942 configure various settings on them, via the [WLAN] section.
3944 * .link/.network files gained support for [Match] Kind= setting to match
3945 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
3947 This value is also shown by 'networkctl status'.
3949 * The Local= setting in .netdev files for various virtual network
3950 devices gained support for specifying, in addition to the network
3951 address, the name of a local interface which must have the specified
3954 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
3955 files, to configure tunnels in external mode (a.k.a. collect metadata
3958 * [Network] L2TP= setting was removed. Please use interface specifier in
3959 Local= setting in .netdev files of corresponding L2TP interface.
3961 * New [DHCPServer] BootServerName=, BootServerAddress=, and
3962 BootFilename= settings can be used to configure the server address,
3963 server name, and file name sent in the DHCP packet (e.g. to configure
3966 Changes in systemd-resolved:
3968 * systemd-resolved is started earlier (in sysinit.target), so it
3969 available earlier and will also be started in the initrd if installed
3972 Changes in disk encryption:
3974 * systemd-cryptenroll can now control whether to require the user to
3975 enter a PIN when using TPM-based unlocking of a volume via the new
3976 --tpm2-with-pin= option.
3978 Option tpm2-pin= can be used in /etc/crypttab.
3980 * When unlocking devices via TPM, TPM2 parameter encryption is now
3981 used, to ensure that communication between CPU and discrete TPM chips
3982 cannot be eavesdropped to acquire disk encryption keys.
3984 * A new switch --fido2-credential-algorithm= has been added to
3985 systemd-cryptenroll allowing selection of the credential algorithm to
3986 use when binding encryption to FIDO2 tokens.
3988 Changes in systemd-hostnamed:
3990 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
3991 to override the values gleaned from the hwdb.
3993 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
3994 /sys/class/dmi/id) to override the chassis that is reported by
3997 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
3998 for reading the hardware serial number, as reportd by DMI. It also
3999 exposes a new method D-Bus property FirmwareVersion that encode the
4000 firmware version of the system.
4002 Changes in other components:
4004 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
4005 handling with the values that were configured during systemd build
4006 (if /etc/locale.conf has not been created through some other
4007 mechanism). This means that /etc/locale.conf should always have
4008 reasonable contents and we avoid a potential mismatch in defaults.
4010 * The userdbctl tool will now show UID range information as part of the
4011 list of known users.
4013 * A new build-time configuration setting default-user-shell= can be
4014 used to set the default shell for user records and nspawn shell
4015 invocations (instead of the default /bin/bash).
4017 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
4018 information dynamically at runtime via IPC.
4020 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
4021 whether a functioning TPM2 infrastructure is available, i.e. if
4022 firmware, kernel driver and systemd all have TPM2 support enabled and
4025 * The systemd-creds tool gained support for generating encrypted
4026 credentials that are using an empty encryption key. While this
4027 provides no integrity nor confidentiality it's useful to implement
4028 codeflows that work the same on TPM-ful and TPM2-less systems. The
4029 service manager will only accept credentials "encrypted" that way if
4030 a TPM2 device cannot be detected, to ensure that credentials
4031 "encrypted" like that cannot be used to trick TPM2 systems.
4033 * When deciding whether to colorize output, all systemd programs now
4034 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
4037 * Meson's new install_tag feature is now in use for several components,
4038 allowing to build and install select binaries only: pam, nss, devel
4039 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
4040 $ meson build systemd-boot
4041 $ meson install --tags systemd-boot --no-rebuild
4042 https://mesonbuild.com/Installing.html#installation-tags
4044 * A new build configuration option has been added, to allow selecting the
4045 default compression algorithm used by systemd-journald and systemd-coredump.
4046 This allows to build-in support for decompressing all supported formats,
4047 but choose a specific one for compression. E.g.:
4048 $ meson -Ddefault-compression=xz
4050 Experimental features:
4052 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
4053 loader.conf that implements booting Microsoft Windows from the
4054 sd-boot in a way that first reboots the system, to reset the TPM
4055 PCRs. This improves compatibility with BitLocker's TPM use, as the
4056 PCRs will only record the Windows boot process, and not sd-boot
4057 itself, thus retaining the PCR measurements not involving sd-boot.
4058 Note that this feature is experimental for now, and is likely going
4059 to be generalized and renamed in a future release, without retaining
4060 compatibility with the current implementation.
4062 * A new systemd-sysupdate component has been added that automatically
4063 discovers, downloads, and installs A/B-style updates for the host
4064 installation itself, or container images, portable service images,
4065 and other assets. See the new systemd-sysupdate man page for updates.
4067 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
4068 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
4069 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
4070 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
4071 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
4072 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
4073 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
4074 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
4075 David, David Bond, Davide Cavalca, David Tardon, davijosw,
4076 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
4077 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
4078 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
4079 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
4080 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
4081 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
4082 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
4083 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
4084 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
4085 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
4086 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
4087 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
4088 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
4089 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
4090 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
4091 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
4092 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
4093 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
4094 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
4095 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
4096 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
4097 Simon Ellmann, Sonali Srivastava, Stefan Seering,
4098 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
4099 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
4100 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
4101 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
4102 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
4103 Zbigniew Jędrzejewski-Szmek, наб
4105 — Edinburgh, 2022-05-21
4109 * Support for encrypted and authenticated credentials has been added.
4110 This extends the credential logic introduced with v247 to support
4111 non-interactive symmetric encryption and authentication, based on a
4112 key that is stored on the /var/ file system or in the TPM2 chip (if
4113 available), or the combination of both (by default if a TPM2 chip
4114 exists the combination is used, otherwise the /var/ key only). The
4115 credentials are automatically decrypted at the moment a service is
4116 started, and are made accessible to the service itself in unencrypted
4117 form. A new tool 'systemd-creds' encrypts credentials for this
4118 purpose, and two new service file settings LoadCredentialEncrypted=
4119 and SetCredentialEncrypted= configure such credentials.
4121 This feature is useful to store sensitive material such as SSL
4122 certificates, passwords and similar securely at rest and only decrypt
4123 them when needed, and in a way that is tied to the local OS
4124 installation or hardware.
4126 * systemd-gpt-auto-generator can now automatically set up discoverable
4127 LUKS2 encrypted swap partitions.
4129 * The GPT Discoverable Partitions Specification has been substantially
4130 extended with support for root and /usr/ partitions for the majority
4131 of architectures systemd supports. This includes platforms that do
4132 not natively support UEFI, because even though GPT is specified under
4133 UEFI umbrella, it is useful on other systems too. Specifically,
4134 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
4135 Portable Services use the concept without requiring UEFI.
4137 * The GPT Discoverable Partitions Specifications has been extended with
4138 a new set of partitions that may carry PKCS#7 signatures for Verity
4139 partitions, encoded in a simple JSON format. This implements a simple
4140 mechanism for building disk images that are fully authenticated and
4141 can be tested against a set of cryptographic certificates. This is
4142 now implemented for the various systemd tools that can operate with
4143 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
4144 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
4145 The PKCS#7 signatures are passed to the kernel (where they are
4146 checked against certificates from the kernel keyring), or can be
4147 verified against certificates provided in userspace (via a simple
4148 drop-in file mechanism).
4150 * systemd-dissect's inspection logic will now report for which uses a
4151 disk image is intended. Specifically, it will display whether an
4152 image is suitable for booting on UEFI or in a container (using
4153 systemd-nspawn's --image= switch), whether it can be used as portable
4154 service, or attached as system extension.
4156 * The system-extension.d/ drop-in files now support a new field
4157 SYSEXT_SCOPE= that may encode which purpose a system extension image
4158 is for: one of "initrd", "system" or "portable". This is useful to
4159 make images more self-descriptive, and to ensure system extensions
4160 cannot be attached in the wrong contexts.
4162 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
4163 be used in portable service images to indicate which unit prefixes
4166 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
4167 now is able to decode images for non-native architectures as well.
4168 This allows systemd-nspawn to boot images of non-native architectures
4169 if the corresponding user mode emulator is installed and
4170 systemd-binfmtd is running.
4172 * systemd-logind gained new settings HandlePowerKeyLongPress=,
4173 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
4174 HandleHibernateKeyLongPress= which may be used to configure actions
4175 when the relevant keys are pressed for more than 5s. This is useful
4176 on devices that only have hardware for a subset of these keys. By
4177 default, if the reboot key is pressed long the poweroff operation is
4178 now triggered, and when the suspend key is pressed long the hibernate
4179 operation is triggered. Long pressing the other two keys currently
4180 does not trigger any operation by default.
4182 * When showing unit status updates on the console during boot and
4183 shutdown, and a service is slow to start so that the cylon animation
4184 is shown, the most recent sd_notify() STATUS= text is now shown as
4185 well. Services may use this to make the boot/shutdown output easier
4186 to understand, and to indicate what precisely a service that is slow
4187 to start or stop is waiting for. In particular, the per-user service
4188 manager instance now reports what it is doing and which service it is
4189 waiting for this way to the system service manager.
4191 * The service manager will now re-execute on reception of the
4192 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
4193 only when running as PID 1. There was no signal to request this when
4194 running as per-user service manager, i.e. as any other PID than 1.
4195 SIGRTMIN+25 works for both system and user managers.
4197 * The hardware watchdog logic in PID 1 gained support for operating
4198 with the default timeout configured in the hardware, instead of
4199 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
4200 request this behavior.
4202 * A new kernel command line option systemd.watchdog_sec= is now
4203 understood which may be used to override the hardware watchdog
4204 time-out for the boot.
4206 * A new setting DefaultOOMScoreAdjust= is now supported in
4207 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
4208 to set the default process OOM score adjustment value for processes
4209 started by the service manager. For per-user service managers this
4210 now defaults to 100, but for per-system service managers is left as
4211 is. This means that by default now services forked off the user
4212 service manager are more likely to be killed by the OOM killer than
4213 system services or the managers themselves.
4215 * A new per-service setting RestrictFileSystems= as been added that
4216 restricts the file systems a service has access to by their type.
4217 This is based on the new BPF LSM of the Linux kernel. It provides an
4218 effective way to make certain API file systems unavailable to
4219 services (and thus minimizing attack surface). A new command
4220 "systemd-analyze filesystems" has been added that lists all known
4221 file system types (and how they are grouped together under useful
4224 * Services now support a new setting RestrictNetworkInterfaces= for
4225 restricting access to specific network interfaces.
4227 * Service unit files gained new settings StartupAllowedCPUs= and
4228 StartupAllowedMemoryNodes=. These are similar to their counterparts
4229 without the "Startup" prefix and apply during the boot process
4230 only. This is useful to improve boot-time behavior of the system and
4231 assign resources differently during boot than during regular
4232 runtime. This is similar to the preexisting StartupCPUWeight=
4235 * Related to this: the various StartupXYZ= settings
4236 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
4237 during shutdown. The settings not prefixed with "Startup" hence apply
4238 during regular runtime, and those that are prefixed like that apply
4239 during boot and shutdown.
4241 * A new per-unit set of conditions/asserts
4242 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
4243 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
4244 pressure is above the configured threshold, using the kernel PSI
4245 feature. For more details see systemd.unit(5) and
4246 https://docs.kernel.org/accounting/psi.html
4248 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
4249 ProtectKernelLogs=yes can now be used.
4251 * The default maximum numbers of inodes have been raised from 64k to 1M
4252 for /dev/, and from 400k to 1M for /tmp/.
4254 * The per-user service manager learnt support for communicating with
4255 systemd-oomd to acquire OOM kill information.
4257 * A new service setting ExecSearchPath= has been added that allows
4258 changing the search path for executables for services. It affects
4259 where we look for the binaries specified in ExecStart= and similar,
4260 and the specified directories are also added the $PATH environment
4261 variable passed to invoked processes.
4263 * A new setting RuntimeRandomizedExtraSec= has been added for service
4264 and scope units that allows extending the runtime time-out as
4265 configured by RuntimeMaxSec= with a randomized amount.
4267 * The syntax of the service unit settings RuntimeDirectory=,
4268 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
4269 if the specified value is now suffixed with a colon, followed by
4270 another filename, the latter will be created as symbolic link to the
4271 specified directory. This allows creating these service directories
4272 together with alias symlinks to make them available under multiple
4275 * Service unit files gained two new settings TTYRows=/TTYColumns= for
4276 configuring rows/columns of the TTY device passed to
4277 stdin/stdout/stderr of the service. This is useful to propagate TTY
4278 dimensions to a virtual machine.
4280 * A new service unit file setting ExitType= has been added that
4281 specifies when to assume a service has exited. By default systemd
4282 only watches the main process of a service. By setting
4283 ExitType=cgroup it can be told to wait for the last process in a
4286 * Automount unit files gained a new setting ExtraOptions= that can be
4287 used to configure additional mount options to pass to the kernel when
4288 mounting the autofs instance.
4290 * "Urlification" (generation of ESC sequences that generate clickable
4291 hyperlinks in modern terminals) may now be turned off altogether
4294 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
4295 settings that default to 200 and 2 s respectively. The ratelimit
4296 ensures that a path unit cannot cause PID1 to busy-loop when it is
4297 trying to trigger a service that is skipped because of a Condition*=
4298 not being satisfied. This matches the configuration and behaviour of
4301 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
4302 as a plug-in for cryptsetup. This means the plain cryptsetup command
4303 may now be used to unlock volumes set up this way.
4305 * The TPM2 logic in cryptsetup will now automatically detect systems
4306 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
4307 updates the SHA1 banks. In such a case PCR policies will be
4308 automatically bound to the latter, not the former. This makes the PCR
4309 policies reliable, but of course do not provide the same level of
4310 trust as SHA256 banks.
4312 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
4313 RSA primary keys in addition to ECC, improving compatibility with
4314 TPM2 chips that do not support ECC. RSA keys are much slower to use
4315 than ECC, and hence are only used if ECC is not available.
4317 * /etc/crypttab gained support for a new token-timeout= setting for
4318 encrypted volumes that allows configuration of the maximum time to
4319 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
4320 the logic will query the user for a regular passphrase/recovery key
4323 * Support for activating dm-integrity volumes at boot via a new file
4324 /etc/integritytab and the tool systemd-integritysetup have been
4325 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
4326 with dm-integrity instead of dm-crypt/dm-verity.
4328 * The systemd-veritysetup-generator now understands a new usrhash=
4329 kernel command line option for specifying the Verity root hash for
4330 the partition backing the /usr/ file system. A matching set of
4331 systemd.verity_usr_* kernel command line options has been added as
4332 well. These all work similar to the corresponding options for the
4335 * The sd-device API gained a new API call sd_device_get_diskseq() to
4336 return the DISKSEQ property of a device structure. The "disk
4337 sequence" concept is a new feature recently introduced to the Linux
4338 kernel that allows detecting reuse cycles of block devices, i.e. can
4339 be used to recognize when loopback block devices are reused for a
4340 different purpose or CD-ROM drives get their media changed.
4342 * A new unit systemd-boot-update.service has been added. If enabled
4343 (the default) and the sd-boot loader is detected to be installed, it
4344 is automatically updated to the newest version when out of date. This
4345 is useful to ensure the boot loader remains up-to-date, and updates
4346 automatically propagate from the OS tree in /usr/.
4348 * sd-boot will now build with SBAT by default in order to facilitate
4349 working with recent versions of Shim that require it to be present.
4351 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
4352 This is used to robustly generate boot entry titles for Windows.
4354 * A new generic target unit factory-reset.target has been added. It is
4355 hooked into systemd-logind similar in fashion to
4356 reboot/poweroff/suspend/hibernate, and is supposed to be used to
4357 initiate a factory reset operation. What precisely this operation
4358 entails is up for the implementer to decide, the primary goal of the
4359 new unit is provide a framework where to plug in the implementation
4360 and how to trigger it.
4362 * A new meson build-time option 'clock-valid-range-usec-max' has been
4363 added which takes a time in µs and defaults to 15 years. If the RTC
4364 time is noticed to be more than the specified time ahead of the
4365 built-in epoch of systemd (which by default is the release timestamp
4366 of systemd) it is assumed that the RTC is not working correctly, and
4367 the RTC is reset to the epoch. (It already is reset to the epoch when
4368 noticed to be before it.) This should increase the chance that time
4369 doesn't accidentally jump too far ahead due to faulty hardware or
4372 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
4373 which may be used to automatically save the current system time to
4374 disk in regular intervals. This is useful to maintain a roughly
4375 monotonic clock even without RTC hardware and with some robustness
4376 against abnormal system shutdown.
4378 * systemd-analyze verify gained support for a pair of new --image= +
4379 --root= switches for verifying units below a specific root
4380 directory/image instead of on the host.
4382 * systemd-analyze verify gained support for verifying unit files under
4383 an explicitly specified unit name, independently of what the filename
4386 * systemd-analyze verify gained a new switch --recursive-errors= which
4387 controls whether to only fail on errors found in the specified units
4388 or recursively any dependent units.
4390 * systemd-analyze security now supports a new --offline mode for
4391 analyzing unit files stored on disk instead of loaded units. It may
4392 be combined with --root=/--image to analyze unit files under a root
4393 directory or disk image. It also learnt a new --threshold= parameter
4394 for specifying an exposure level threshold: if the exposure level
4395 exceeds the specified value the call will fail. It also gained a new
4396 --security-policy= switch for configuring security policies to
4397 enforce on the units. A policy is a JSON file that lists which tests
4398 shall be weighted how much to determine the overall exposure
4399 level. Altogether these new features are useful for fully automatic
4400 analysis and enforcement of security policies on unit files.
4402 * systemd-analyze security gain a new --json= switch for JSON output.
4404 * systemd-analyze learnt a new --quiet switch for reducing
4405 non-essential output. It's honored by the "dot", "syscall-filter",
4406 "filesystems" commands.
4408 * systemd-analyze security gained a --profile= option that can be used
4409 to take into account a portable profile when analyzing portable
4410 services, since a lot of the security-related settings are enabled
4413 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
4414 files, binaries and executables and prints metadata information,
4415 including the build-id and other info described on:
4416 https://systemd.io/COREDUMP_PACKAGE_METADATA/
4418 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
4419 section, for automatically propagating DNS settings from other
4422 * The static lease DHCP server logic in systemd-networkd may now serve
4423 IP addresses outside of the configured IP pool range for the server.
4425 * CAN support in systemd-networkd gained four new settings Loopback=,
4426 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
4427 control modes. It gained a number of further settings for tweaking
4430 * The [CAN] section in .network file gained new TimeQuantaNSec=,
4431 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
4432 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
4433 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
4434 DataSyncJumpWidth= settings to control bit-timing processed by the
4437 * DHCPv4 client support in systemd-networkd learnt a new Label= option
4438 for configuring the address label to apply to configure IPv4
4441 * The [IPv6AcceptRA] section of .network files gained support for a new
4442 UseMTU= setting that may be used to control whether to apply the
4443 announced MTU settings to the local interface.
4445 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
4446 setting to control whether the DHCPv4 client request and process the
4449 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
4450 [DHCPPrefixDelegation], as now the prefix delegation is also supported
4451 with DHCPv4 protocol by enabling the Use6RD= setting.
4453 * The [DHCPPrefixDelegation] section in .network file gained a new
4454 setting UplinkInterface= to specify the upstream interface.
4456 * The [DHCPv6] section in .network file gained a new setting
4457 UseDelegatedPrefix= to control whether the delegated prefixes will be
4458 propagated to the downstream interfaces.
4460 * The [IPv6AcceptRA] section of .network files now understands two new
4461 settings UseGateway=/UseRoutePrefix= for explicitly configuring
4462 whether to use the relevant fields from the IPv6 Router Advertisement
4465 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
4466 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
4467 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
4468 [IPv6AcceptRA] section to control when the DHCPv6 client is started
4469 and how the delegated prefixes are handled by the DHCPv6 client.
4471 * The IPv6Token= section in the [Network] section is deprecated, and
4472 the [IPv6AcceptRA] section gained the Token= setting for its
4473 replacement. The [IPv6Prefix] section also gained the Token= setting.
4474 The Token= setting gained 'eui64' mode to explicitly configure an
4475 address with the EUI64 algorithm based on the interface MAC address.
4476 The 'prefixstable' mode can now optionally take a secret key. The
4477 Token= setting in the [DHCPPrefixDelegation] section now supports all
4478 algorithms supported by the same settings in the other sections.
4480 * The [RoutingPolicyRule] section of .network file gained a new
4481 SuppressInterfaceGroup= setting.
4483 * The IgnoreCarrierLoss= setting in the [Network] section of .network
4484 files now allows a duration to be specified, controlling how long to
4485 wait before reacting to carrier loss.
4487 * The [DHCPServer] section of .network file gained a new Router=
4488 setting to specify the router address.
4490 * The [CAKE] section of .network files gained various new settings
4491 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
4492 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
4493 and UseRawPacketSize= for configuring CAKE.
4495 * systemd-networkd now ships with new default .network files:
4496 80-container-vb.network which matches host-side network bridge device
4497 created by systemd-nspawn's --network-bridge or --network-zone
4498 switch, and 80-6rd-tunnel.network which matches automatically created
4499 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
4501 * systemd-networkd's handling of Endpoint= resolution for WireGuard
4502 interfaces has been improved.
4504 * systemd-networkd will now automatically configure routes to addresses
4505 specified in AllowedIPs=. This feature can be controlled via
4506 RouteTable= and RouteMetric= settings in [WireGuard] or
4507 [WireGuardPeer] sections.
4509 * systemd-networkd will now once again automatically generate persistent
4510 MAC addresses for batadv and bridge interfaces. Users can disable this
4511 by using MACAddress=none in .netdev files.
4513 * systemd-networkd and systemd-udevd now support IP over InfiniBand
4514 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
4515 systemd.netdev files gained the [IPoIB] section.
4517 * systemd-networkd and systemd-udevd now support net.ifname_policy=
4518 option on the kernel command-line. This is implemented through the
4519 systemd-network-generator service that automatically generates
4520 appropriate .link, .network, and .netdev files.
4522 * The various systemd-udevd "ethtool" buffer settings now understand
4523 the special value "max" to configure the buffers to the maximum the
4526 * systemd-udevd's .link files may now configure a large variety of
4527 NIC coalescing settings, plus more hardware offload settings.
4529 * .link files gained a new WakeOnLanPassword= setting in the [Link]
4530 section that allows to specify a WoL "SecureOn" password on hardware
4533 * systemd-nspawn's --setenv= switch now supports an additional syntax:
4534 if only a variable name is specified (i.e. without being suffixed by
4535 a '=' character and a value) the current value of the environment
4536 variable is propagated to the container. e.g. --setenv=FOO will
4537 lookup the current value of $FOO in the environment, and pass it down
4538 to the container. Similar behavior has been added to homectl's,
4539 machinectl's and systemd-run's --setenv= switch.
4541 * systemd-nspawn gained a new switch --suppress-sync= which may be used
4542 to optionally suppress the effect of the sync()/fsync()/fdatasync()
4543 system calls for the container payload. This is useful for build
4544 system environments where safety against abnormal system shutdown is
4545 not essential as all build artifacts can be regenerated any time, but
4546 the performance win is beneficial.
4548 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
4549 same value that PID 1 uses for most forked off processes.
4551 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
4552 uidmap/nouidmap options as last parameter. If "uidmap" is used the
4553 bind mounts are created with UID mapping taking place that ensures
4554 the host's file ownerships are mapped 1:1 to container file
4555 ownerships, even if user namespacing is used. This way
4556 files/directories bound into containers will no longer show up as
4557 owned by the nobody user as they typically did if no special care was
4558 taken to shift them manually.
4560 * When discovering Windows installations sd-boot will now attempt to
4561 show the Windows version.
4563 * The color scheme to use in sd-boot may now be configured at
4566 * sd-boot gained the ability to change screen resolution during
4567 boot-time, by hitting the "r" key. This will cycle through available
4568 resolutions and save the last selection.
4570 * sd-boot learnt a new hotkey "f". When pressed the system will enter
4571 firmware setup. This is useful in environments where it is difficult
4572 to hit the right keys early enough to enter the firmware, and works
4573 on any firmware regardless which key it natively uses.
4575 * sd-boot gained support for automatically booting into the menu item
4576 selected on the last boot (using the "@saved" identifier for menu
4579 * sd-boot gained support for automatically loading all EFI drivers
4580 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
4581 Partition (ESP). These drivers are loaded before the menu entries are
4582 loaded. This is useful e.g. to load additional file system drivers
4583 for the XBOOTLDR partition.
4585 * systemd-boot will now paint the input cursor on its own instead of
4586 relying on the firmware to do so, increasing compatibility with broken
4587 firmware that doesn't make the cursor reasonably visible.
4589 * sd-boot now embeds a .osrel PE section like we expect from Boot
4590 Loader Specification Type #2 Unified Kernels. This means sd-boot
4591 itself may be used in place of a Type #2 Unified Kernel. This is
4592 useful for debugging purposes as it allows chain-loading one a
4593 (development) sd-boot instance from another.
4595 * sd-boot now supports a new "devicetree" field in Boot Loader
4596 Specification Type #1 entries: if configured the specified device
4597 tree file is installed before the kernel is invoked. This is useful
4598 for installing/applying new devicetree files without updating the
4601 * Similarly, sd-stub now can read devicetree data from a PE section
4602 ".dtb" and apply it before invoking the kernel.
4604 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
4605 gained the ability to pick up credentials and sysext files, wrap them
4606 in a cpio archive, and pass as an additional initrd to the invoked
4607 Linux kernel, in effect placing those files in the /.extra/ directory
4608 of the initrd environment. This is useful to implement trusted initrd
4609 environments which are fully authenticated but still can be extended
4610 (via sysexts) and parameterized (via encrypted/authenticated
4611 credentials, see above).
4613 Credentials can be located next to the kernel image file (credentials
4614 specific to a single boot entry), or in one of the shared directories
4615 (credentials applicable to multiple boot entries).
4617 * sd-stub now comes with a full man page, that explains its feature set
4618 and how to combine a kernel image, an initrd and the stub to build a
4619 complete EFI unified kernel image, implementing Boot Loader
4620 Specification Type #2.
4622 * sd-stub may now provide the initrd to the executed kernel via the
4623 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
4624 non-x86 architectures.
4626 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
4627 may be used to set the boot menu time-out of the boot loader (for all
4628 or just the subsequent boot).
4630 * bootctl and kernel-install will now read variables
4631 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
4632 /etc/kernel/install.conf. When set, it specifies the layout to use
4633 for installation directories on the boot partition, so that tools
4634 don't need to guess it based on the already-existing directories. The
4635 only value that is defined natively is "bls", corresponding to the
4637 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
4638 kernel-install that implement a different layout can declare other
4639 values for this variable.
4641 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
4642 assumption that if the user installed sd-boot to the ESP, they intend
4643 to use the entry layout understood by sd-boot. It'll also write
4644 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
4645 (and it wasn't specified in the config file yet). Similarly,
4646 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
4647 wasn't specified in the config file yet). Effectively, those changes
4648 mean that the machine-id used for boot loader entry installation is
4649 "frozen" upon first use and becomes independent of the actual
4652 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
4653 images created for distribution ("golden images") are built with no
4654 machine-id, so that a unique machine-id can be created on the first
4655 boot. But those images may contain boot loader entries with the
4656 machine-id used during build included in paths. Using a "frozen"
4657 value allows unambiguously identifying entries that match the
4658 specific installation, while still permitting parallel installations
4661 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
4662 kernel-install to guess the installation layout. This fixes the
4663 problem where a (possibly empty) directory in the boot partition is
4664 created from a different layout causing kernel-install plugins to
4665 assume the wrong layout. A particular example of how this may happen
4666 is the grub2 package in Fedora which includes directories under /boot
4667 directly in its file list. Various other packages pull in grub2 as a
4668 dependency, so it may be installed even if unused, breaking
4669 installations that use the bls layout.
4671 * bootctl and systemd-bless-boot can now be linked statically.
4673 * systemd-sysext now optionally doesn't insist on extension-release.d/
4674 files being placed in the image under the image's file name. If the
4675 file system xattr user.extension-release.strict is set on the
4676 extension release file, it is accepted regardless of its name. This
4677 relaxes security restrictions a bit, as system extension may be
4678 attached under a wrong name this way.
4680 * udevadm's test-builtin command learnt a new --action= switch for
4681 testing the built-in with the specified action (in place of the
4684 * udevadm info gained new switches --property=/--value for showing only
4685 specific udev properties/values instead of all.
4687 * A new hwdb database has been added that contains matches for various
4688 types of signal analyzers (protocol analyzers, logic analyzers,
4689 oscilloscopes, multimeters, bench power supplies, etc.) that should
4690 be accessible to regular users.
4692 * A new hwdb database entry has been added that carries information
4693 about types of cameras (regular or infrared), and in which direction
4694 they point (front or back).
4696 * A new rule to allow console users access to rfkill by default has been
4699 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
4700 now also owned by the system group "sgx".
4702 * A new build-time meson option "extra-net-naming-schemes=" has been
4703 added to define additional naming schemes for udev's network
4704 interface naming logic. This is useful for enterprise distributions
4705 and similar which want to pin the schemes of certain distribution
4706 releases under a specific name and previously had to patch the
4707 sources to introduce new named schemes.
4709 * The predictable naming logic for network interfaces has been extended
4710 to generate stable names from Xen netfront device information.
4712 * hostnamed's chassis property can now be sourced from chassis-type
4713 field encoded in devicetree (in addition to the existing DMI
4716 * systemd-cgls now optionally displays cgroup IDs and extended
4717 attributes for each cgroup. (Controllable via the new --xattr= +
4718 --cgroup-id= switches.)
4720 * coredumpctl gained a new --all switch for operating on all
4721 Journal files instead of just the local ones.
4723 * systemd-coredump will now use libdw/libelf via dlopen() rather than
4724 directly linking, allowing users to easily opt-out of backtrace/metadata
4725 analysis of core files, and reduce image sizes when this is not needed.
4727 * systemd-coredump will now analyze core files with libdw/libelf in a
4728 forked, sandboxed process.
4730 * systemd-homed will now try to unmount an activate home area in
4731 regular intervals once the user logged out fully. Previously this was
4732 attempted exactly once but if the home directory was busy for some
4733 reason it was not tried again.
4735 * systemd-homed's LUKS2 home area backend will now create a BSD file
4736 system lock on the image file while the home area is active
4737 (i.e. mounted). If a home area is found to be locked, logins are
4738 politely refused. This should improve behavior when using home areas
4739 images that are accessible via the network from multiple clients, and
4740 reduce the chance of accidental file system corruption in that case.
4742 * Optionally, systemd-homed will now drop the kernel buffer cache once
4743 a user has fully logged out, configurable via the new --drop-caches=
4746 * systemd-homed now makes use of UID mapped mounts for the home areas.
4747 If the kernel and used file system support it, files are now
4748 internally owned by the "nobody" user (i.e. the user typically used
4749 for indicating "this ownership is not mapped"), and dynamically
4750 mapped to the UID used locally on the system via the UID mapping
4751 mount logic of recent kernels. This makes migrating home areas
4752 between different systems cheaper because recursively chown()ing file
4753 system trees is no longer necessary.
4755 * systemd-homed's CIFS backend now optionally supports CIFS service
4756 names with a directory suffix, in order to place home directories in
4757 a subdirectory of a CIFS share, instead of the top-level directory.
4759 * systemd-homed's CIFS backend gained support for specifying additional
4760 mount options in the JSON user record (cifsExtraMountOptions field,
4761 and --cifs-extra-mount-options= homectl switch). This is for example
4762 useful for configuring mount options such as "noserverino" that some
4763 SMB3 services require (use that to run a homed home directory from a
4764 FritzBox SMB3 share this way).
4766 * systemd-homed will now default to btrfs' zstd compression for home
4767 areas. This is inspired by Fedora's recent decision to switch to zstd
4770 * Additional mount options to use when mounting the file system of
4771 LUKS2 volumes in systemd-homed has been added. Via the
4772 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
4773 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
4774 systemd-homed or via the luksExtraMountOptions user record JSON
4775 property. (Exposed via homectl --luks-extra-mount-options)
4777 * homectl's resize command now takes the special size specifications
4778 "min" and "max" to shrink/grow the home area to the minimum/maximum
4779 size possible, taking disk usage/space constraints and file system
4780 limitations into account. Resizing is now generally graceful: the
4781 logic will try to get as close to the specified size as possible, but
4782 not consider it a failure if the request couldn't be fulfilled
4785 * systemd-homed gained the ability to automatically shrink home areas
4786 on logout to their minimal size and grow them again on next
4787 login. This ensures that while inactive, a home area only takes up
4788 the minimal space necessary, but once activated, it provides
4789 sufficient space for the user's needs. This behavior is only
4790 supported if btrfs is used as file system inside the home area
4791 (because only for btrfs online growing/shrinking is implemented in
4792 the kernel). This behavior is now enabled by default, but may be
4793 controlled via the new --auto-resize-mode= setting of homectl.
4795 * systemd-homed gained support for automatically re-balancing free disk
4796 space among active home areas, in case the LUKS2 backends are used,
4797 and no explicit disk size was requested. This way disk space is
4798 automatically managed and home areas resized in regular intervals and
4799 manual resizing when disk space becomes scarce should not be
4800 necessary anymore. This behavior is only supported if btrfs is used
4801 within the home areas (as only then online shrinking and growing is
4802 supported), and may be configured via the new rebalanceWeight JSON
4803 user record field (as exposed via the new --rebalance-weight= homectl
4804 setting). Re-balancing is mostly automatic, but can also be requested
4805 explicitly via "homectl rebalance", which is synchronous, and thus
4806 may be used to wait until the rebalance run is complete.
4808 * userdbctl gained a --json= switch for configured the JSON formatting
4809 to use when outputting user or group records.
4811 * userdbctl gained a new --multiplexer= switch for explicitly
4812 configuring whether to use the systemd-userdbd server side user
4813 record resolution logic.
4815 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
4816 for chaining up another command to execute after completing the
4817 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
4818 configuration of a single command to invoke, this maybe used to
4819 invoke multiple: first userdbctl's own implementation, and then any
4820 other also configured in the command line.
4822 * The sd-event API gained a new function sd_event_add_inotify_fd() that
4823 is similar to sd_event_add_inotify() but accepts a file descriptor
4824 instead of a path in the file system for referencing the inode to
4827 * The sd-event API gained a new function
4828 sd_event_source_set_ratelimit_expire_callback() that may be used to
4829 define a callback function that is called whenever an event source
4830 leaves the rate limiting phase.
4832 * New documentation has been added explaining which steps are necessary
4833 to port systemd to a new architecture:
4835 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
4837 * The x-systemd.makefs option in /etc/fstab now explicitly supports
4838 ext2, ext3, and f2fs file systems.
4840 * Mount units and units generated from /etc/fstab entries with 'noauto'
4841 are now ordered the same as other units. Effectively, they will be
4842 started earlier (if something actually pulled them in) and stopped
4843 later, similarly to normal mount units that are part of
4844 fs-local.target. This change should be invisible to users, but
4845 should prevent those units from being stopped too early during
4848 * The systemd-getty-generator now honors a new kernel command line
4849 argument systemd.getty_auto= and a new environment variable
4850 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
4851 example useful to turn off gettys inside of containers or similar
4854 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
4855 (in addition to 127.0.0.53, as before). If DNS requests are sent to
4856 this address they are propagated in "bypass" mode only, i.e. are
4857 almost not processed locally, but mostly forwarded as-is to the
4858 current upstream DNS servers. This provides a stable DNS server
4859 address that proxies all requests dynamically to the right upstream
4860 DNS servers even if these dynamically change. This stub does not do
4861 mDNS/LLMNR resolution. However, it will translate look-ups to
4862 DNS-over-TLS if necessary. This new stub is particularly useful in
4863 container/VM environments, or for tethering setups: use DNAT to
4864 redirect traffic to any IP address to this stub.
4866 * systemd-importd now honors new environment variables
4867 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
4868 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
4869 generation, btrfs quota setup and disk synchronization.
4871 * systemd-importd and systemd-resolved can now be optionally built with
4872 OpenSSL instead of libgcrypt.
4874 * systemd-repart no longer requires OpenSSL.
4876 * systemd-sysusers will no longer create the redundant 'nobody' group
4877 by default, as the 'nobody' user is already created with an
4878 appropriate primary group.
4880 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
4882 * systemctl show-environment gained support for --output=json.
4884 * pam_systemd will now first try to use the X11 abstract socket, and
4885 fallback to the socket file in /tmp/.X11-unix/ only if that does not
4888 * systemd-journald will no longer go back to volatile storage
4889 regardless of configuration when its unit is restarted.
4891 * Initial support for the LoongArch architecture has been added (system
4892 call lists, GPT partition table UUIDs, etc).
4894 * systemd-journald's own logging messages are now also logged to the
4895 journal itself when systemd-journald logs to /dev/kmsg.
4897 * systemd-journald now re-enables COW for archived journal files on
4898 filesystems that support COW. One benefit of this change is that
4899 archived journal files will now get compressed on btrfs filesystems
4900 that have compression enabled.
4902 * systemd-journald now deduplicates fields in a single log message
4903 before adding it to the journal. In archived journal files, it will
4904 also punch holes for unused parts and truncate the file as
4905 appropriate, leading to reductions in disk usage.
4907 * journalctl --verify was extended with more informative error
4910 * More of sd-journal's functions are now resistant against journal file
4913 * The shutdown command learnt a new option --show, to display the
4916 * A LICENSES/ directory is now included in the git tree. It contains a
4917 README.md file that explains the licenses used by source files in
4918 this repository. It also contains the text of all applicable
4919 licenses as they appear on spdx.org.
4921 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
4922 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
4923 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
4924 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
4925 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
4926 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
4927 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
4928 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
4929 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
4930 Christian Brauner, Christian Göttsche, Christian Wehrli,
4931 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
4932 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
4933 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
4934 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
4935 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
4936 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
4937 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
4938 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
4939 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
4940 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
4941 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
4942 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
4943 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
4944 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
4945 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
4946 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
4947 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
4948 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
4949 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
4950 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
4951 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
4952 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
4953 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
4954 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
4955 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
4956 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
4957 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
4958 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
4959 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
4960 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
4961 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
4962 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
4963 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
4964 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
4965 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
4966 Дамјан Георгиевски, наб
4968 — Warsaw, 2021-12-23
4972 * When operating on disk images via the --image= switch of various
4973 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
4974 no 'root=' parameter on the kernel command line, and multiple
4975 suitable root or /usr/ partitions exist in the image, then a simple
4976 comparison inspired by strverscmp() is done on the GPT partition
4977 label, and the newest partition is picked. This permits a simple and
4978 generic whole-file-system A/B update logic where new operating system
4979 versions are dropped into partitions whose label is then updated with
4980 a matching version identifier.
4982 * systemd-sysusers now supports querying the passwords to set for the
4983 users it creates via the "credentials" logic introduced in v247: the
4984 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
4985 credentials are consulted for the password to use (either in UNIX
4986 hashed form, or literally). By default these credentials are inherited
4987 down from PID1 (which in turn imports it from a container manager if
4988 there is one). This permits easy configuration of user passwords
4989 during first boot. Example:
4991 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
4993 Note that systemd-sysusers operates in purely additive mode: it
4994 executes no operation if the declared users already exist, and hence
4995 doesn't set any passwords as effect of the command line above if the
4996 specified root user exists already in the image. (Note that
4997 --volatile=yes ensures it doesn't, though.)
4999 * systemd-firstboot now also supports querying various system
5000 parameters via the credential subsystems. Thus, as above this may be
5001 used to initialize important system parameters on first boot of
5002 previously unprovisioned images (i.e. images with a mostly empty
5005 * PID 1 may now show both the unit name and the unit description
5006 strings in its status output during boot. This may be configured with
5007 StatusUnitFormat=combined in system.conf or
5008 systemd.status-unit-format=combined on the kernel command line.
5010 * The systemd-machine-id-setup tool now supports a --image= switch for
5011 provisioning a machine ID file into an OS disk image, similar to how
5012 --root= operates on an OS file tree. This matches the existing switch
5013 of the same name for systemd-tmpfiles, systemd-firstboot, and
5014 systemd-sysusers tools.
5016 * Similarly, systemd-repart gained support for the --image= switch too.
5017 In combination with the existing --size= option, this makes the tool
5018 particularly useful for easily growing disk images in a single
5019 invocation, following the declarative rules included in the image
5022 * systemd-repart's partition configuration files gained support for a
5023 new switch MakeDirectories= which may be used to create arbitrary
5024 directories inside file systems that are created, before registering
5025 them in the partition table. This is useful in particular for root
5026 partitions to create mount point directories for other partitions
5027 included in the image. For example, a disk image that contains a
5028 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
5029 create /home/ and /var/ as empty directories in the root file system
5030 on its creation, so that the resulting image can be mounted
5031 immediately, even in read-only mode.
5033 * systemd-repart's CopyBlocks= setting gained support for the special
5034 value "auto". If used, a suitable matching partition on the booted OS
5035 is found as source to copy blocks from. This is useful when
5036 implementing replicating installers, that are booted from one medium
5037 and then stream their own root partition onto the target medium.
5039 * systemd-repart's partition configuration files gained support for a
5040 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
5041 GPT partition flags for the created partitions: this is useful for
5042 marking newly created partitions as read-only, or as not being
5043 subject for automatic mounting from creation on.
5045 * The /etc/os-release file has been extended with two new (optional)
5046 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
5047 information for OS images that are updated comprehensively and
5048 atomically as one image. Two new specifiers %M, %A now resolve to
5049 these two fields in the various configuration options that resolve
5052 * portablectl gained a new switch --extension= for enabling portable
5053 service images with extensions that follow the extension image
5054 concept introduced with v248, and thus allows layering multiple
5055 images when setting up the root filesystem of the service.
5057 * systemd-coredump will now extract ELF build-id information from
5058 processes dumping core and include it in the coredump report.
5059 Moreover, it will look for ELF .note.package sections with
5060 distribution packaging meta-information about the crashing process.
5061 This is useful to directly embed the rpm or deb (or any other)
5062 package name and version in ELF files, making it easy to match
5063 coredump reports with the specific package for which the software was
5064 compiled. This is particularly useful on environments with ELF files
5065 from multiple vendors, different distributions and versions, as is
5066 common today in our containerized and sand-boxed world. For further
5069 https://systemd.io/COREDUMP_PACKAGE_METADATA
5071 * A new udev hardware database has been added for FireWire devices
5074 * The "net_id" built-in of udev has been updated with three
5075 backwards-incompatible changes:
5077 - PCI hotplug slot names on s390 systems are now parsed as
5078 hexadecimal numbers. They were incorrectly parsed as decimal
5079 previously, or ignored if the name was not a valid decimal
5082 - PCI onboard indices up to 65535 are allowed. Previously, numbers
5083 above 16383 were rejected. This primarily impacts s390 systems,
5084 where values up to 65535 are used.
5086 - Invalid characters in interface names are replaced with "_".
5088 The new version of the net naming scheme is "v249". The previous
5089 scheme can be selected via the "net.naming_scheme=v247" kernel
5090 command line parameter.
5092 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
5093 NULL bus object, for which they will return false. Or in other words,
5094 an unallocated bus connection is neither ready nor open.
5096 * The sd-device API acquired a new API function
5097 sd_device_get_usec_initialized() that returns the monotonic time when
5098 the udev device first appeared in the database.
5100 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
5101 sd_device_get_trigger_uuid(). The former is similar to
5102 sd_device_trigger() but returns a randomly generated UUID that is
5103 associated with the synthetic uevent generated by the call. This UUID
5104 may be read from the sd_device object a monitor eventually receives,
5105 via the sd_device_get_trigger_uuid(). This interface requires kernel
5106 4.13 or above to work, and allows tracking a synthetic uevent through
5107 the entire device management stack. The "udevadm trigger --settle"
5108 logic has been updated to make use of this concept if available to
5109 wait precisely for the uevents it generates. "udevadm trigger" also
5110 gained a new parameter --uuid that prints the UUID for each generated
5113 * sd-device also gained new APIs sd_device_new_from_ifname() and
5114 sd_device_new_from_ifindex() for allocating an sd-device object for
5115 the specified network interface. The former accepts an interface name
5116 (either a primary or an alternative name), the latter an interface
5119 * The native Journal protocol has been documented. Clients may talk
5120 this as alternative to the classic BSD syslog protocol for locally
5121 delivering log records to the Journal. The protocol has been stable
5122 for a long time and in fact been implemented already in a variety
5123 of alternative client libraries. This documentation makes the support
5126 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
5128 * A new BPFProgram= setting has been added to service files. It may be
5129 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
5130 file, or a bind mount or symlink to one. This may be used to upload
5131 and manage BPF programs externally and then hook arbitrary systemd
5134 * The "home.arpa" domain that has been officially declared as the
5135 choice for domain for local home networks per RFC 8375 has been added
5136 to the default NTA list of resolved, since DNSSEC is generally not
5137 available on private domains.
5139 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
5141 * A new ManageForeignRoutingPolicyRules= setting has been added to
5142 .network files which may be used to exclude foreign-created routing
5143 policy rules from systemd-networkd management.
5145 * systemd-network-wait-online gained two new switches -4 and -6 that
5146 may be used to tweak whether to wait for only IPv4 or only IPv6
5149 * .network files gained a new RequiredFamilyForOnline= setting to
5150 fine-tune whether to require an IPv4 or IPv6 address in order to
5151 consider an interface "online".
5153 * networkctl will now show an over-all "online" state in the per-link
5156 * In .network files a new OutgoingInterface= setting has been added to
5157 specify the output interface in bridge FDB setups.
5159 * In .network files the Multipath group ID may now be configured for
5160 [NextHop] entries, via the new Group= setting.
5162 * The DHCP server logic configured in .network files gained a new
5163 setting RelayTarget= that turns the server into a DHCP server relay.
5164 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
5165 to further tweak the DHCP relay behaviour.
5167 * The DHCP server logic also gained a new ServerAddress= setting in
5168 .network files that explicitly specifies the server IP address to
5169 use. If not specified, the address is determined automatically, as
5172 * The DHCP server logic in systemd-networkd gained support for static
5173 DHCP leases, configurable via the [DHCPServerStaticLease]
5174 section. This allows explicitly mapping specific MAC addresses to
5175 fixed IP addresses and vice versa.
5177 * The RestrictAddressFamilies= setting in service files now supports a
5178 new special value "none". If specified sockets of all address
5179 families will be made unavailable to services configured that way.
5181 * systemd-fstab-generator and systemd-repart have been updated to
5182 support booting from disks that carry only a /usr/ partition but no
5183 root partition yet, and where systemd-repart can add it in on the
5184 first boot. This is useful for implementing systems that ship with a
5185 single /usr/ file system, and whose root file system shall be set up
5186 and formatted on a LUKS-encrypted volume whose key is generated
5187 locally (and possibly enrolled in the TPM) during the first boot.
5189 * The [Address] section of .network files now accepts a new
5190 RouteMetric= setting that configures the routing metric to use for
5191 the prefix route created as effect of the address configuration.
5192 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
5193 gained matching settings for their prefix routes. (The option of the
5194 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
5195 it conceptually belongs there; the old option is still understood for
5198 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
5201 * A new udev property ID_NET_DHCP_BROADCAST on network interface
5202 devices is now honoured by systemd-networkd, controlling whether to
5203 issue DHCP offers via broadcasting. This is used to ensure that s390
5204 layer 3 network interfaces work out-of-the-box with systemd-networkd.
5206 * nss-myhostname and systemd-resolved will now synthesize address
5207 records for a new special hostname "_outbound". The name will always
5208 resolve to the local IP addresses most likely used for outbound
5209 connections towards the default routes. On multi-homed hosts this is
5210 useful to have a stable handle referring to "the" local IP address
5211 that matters most, to the point where this is defined.
5213 * The Discoverable Partition Specification has been updated with a new
5214 GPT partition flag "grow-file-system" defined for its partition
5215 types. Whenever partitions with this flag set are automatically
5216 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
5217 of systemd-nspawn or other tools; and as opposed to explicit mounting
5218 via /etc/fstab), the file system within the partition is
5219 automatically grown to the full size of the partition. If the file
5220 system size already matches the partition size this flag has no
5221 effect. Previously, this functionality has been available via the
5222 explicit x-systemd.growfs mount option, and this new flag extends
5223 this to automatically discovered mounts. A new GrowFileSystem=
5224 setting has been added to systemd-repart drop-in files that allows
5225 configuring this partition flag. This new flag defaults to on for
5226 partitions automatically created by systemd-repart, except if they
5227 are marked read-only. See the specification for further details:
5229 https://systemd.io/DISCOVERABLE_PARTITIONS
5231 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
5232 section. If enabled (which is the default), and an NTP server address
5233 is acquired through a DHCP lease on this interface an explicit route
5234 to this address is created on this interface to ensure that NTP
5235 traffic to the NTP server acquired on an interface is also routed
5236 through that interface. The pre-existing RoutesToDNS= setting that
5237 implements the same for DNS servers is now enabled by default.
5239 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
5240 been added that may be used to restrict the network interfaces
5241 sockets created by the service may be bound to. This is implemented
5244 * A new ConditionFirmware= setting has been added to unit files to
5245 conditionalize on certain firmware features. At the moment it may
5246 check whether running on a UEFI system, a device.tree system, or if
5247 the system is compatible with some specified device-tree feature.
5249 * A new ConditionOSRelease= setting has been added to unit files to
5250 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
5251 operators may be used to check if some field has some specific value
5252 or do an alphanumerical comparison. Equality comparisons are useful
5253 for fields like ID, but relative comparisons for fields like
5254 VERSION_ID or IMAGE_VERSION.
5256 * hostnamed gained a new Describe() D-Bus method that returns a JSON
5257 serialization of the host data it exposes. This is exposed via
5258 "hostnamectl --json=" to acquire a host identity description in JSON.
5259 It's our intention to add a similar features to most services and
5260 objects systemd manages, in order to simplify integration with
5261 program code that can consume JSON.
5263 * Similarly, networkd gained a Describe() method on its Manager and
5264 Link bus objects. This is exposed via "networkctl --json=".
5266 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
5267 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
5268 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
5269 that is used both to get the value (when no argument is given), and
5270 to set the value (when an argument is specified). The old names
5271 continue to be supported for compatibility.
5273 * systemd-detect-virt and ConditionVirtualization= are now able to
5274 correctly identify Amazon EC2 environments.
5276 * The LogLevelMax= setting of unit files now applies not only to log
5277 messages generated *by* the service, but also to log messages
5278 generated *about* the service by PID 1. To suppress logs concerning a
5279 specific service comprehensively, set this option to a high log
5282 * bootctl gained support for a new --make-machine-id-directory= switch
5283 that allows precise control on whether to create the top-level
5284 per-machine directory in the boot partition that typically contains
5285 Type 1 boot loader entries.
5287 * During build SBAT data to include in the systemd-boot EFI PE binaries
5288 may be specified now.
5290 * /etc/crypttab learnt a new option "headless". If specified any
5291 requests to query the user interactively for passwords or PINs will
5292 be skipped. This is useful on systems that are headless, i.e. where
5293 an interactive user is generally not present.
5295 * /etc/crypttab also learnt a new option "password-echo=" that allows
5296 configuring whether the encryption password prompt shall echo the
5297 typed password and if so, do so literally or via asterisks. (The
5298 default is the same behaviour as before: provide echo feedback via
5301 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
5302 systemd-homed has been updated to allow explicit configuration of the
5303 "user presence" and "user verification" checks, as well as whether a
5304 PIN is required for authentication, via the new switches
5305 --fido2-with-user-presence=, --fido2-with-user-verification=,
5306 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
5307 features are available, and may be enabled or disabled depends on the
5310 * systemd-nspawn's --private-user= switch now accepts the special value
5311 "identity" which configures a user namespacing environment with an
5312 identity mapping of 65535 UIDs. This means the container UID 0 is
5313 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
5314 this doesn't appear to be useful, however it does reduce the attack
5315 surface a bit, since the resulting container will possess process
5316 capabilities only within its namespace and not on the host.
5318 * systemd-nspawn's --private-user-chown switch has been replaced by a
5319 more generic --private-user-ownership= switch that accepts one of
5320 three values: "chown" is equivalent to the old --private-user-chown,
5321 and "off" is equivalent to the absence of the old switch. The value
5322 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
5323 of files and directories of the underlying image to the chosen UID
5324 range for the container. "auto" is equivalent to "map" if UID mapping
5325 mount are supported, otherwise it is equivalent to "chown". The short
5326 -U switch systemd-nspawn now implies --private-user-ownership=auto
5327 instead of the old --private-user-chown. Effectively this means: if
5328 the backing file system supports UID mapping mounts the feature is
5329 now used by default if -U is used. Generally, it's a good idea to use
5330 UID mapping mounts instead of recursive chown()ing, since it allows
5331 running containers off immutable images (since no modifications of
5332 the images need to take place), and share images between multiple
5333 instances. Moreover, the recursive chown()ing operation is slow and
5334 can be avoided. Conceptually it's also a good thing if transient UID
5335 range uses do not leak into persistent file ownership anymore. TLDR:
5336 finally, the last major drawback of user namespacing has been
5337 removed, and -U should always be used (unless you use btrfs, where
5338 UID mapped mounts do not exist; or your container actually needs
5339 privileges on the host).
5341 * nss-systemd now synthesizes user and group shadow records in addition
5342 to the main user and group records. Thus, hashed passwords managed by
5343 systemd-homed are now accessible via the shadow database.
5345 * The userdb logic (and thus nss-systemd, and so on) now read
5346 additional user/group definitions in JSON format from the drop-in
5347 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
5348 /usr/lib/userdb/. This is a simple and powerful mechanism for making
5349 additional users available to the system, with full integration into
5350 NSS including the shadow databases. Since the full JSON user/group
5351 record format is supported this may also be used to define users with
5352 resource management settings and other runtime settings that
5353 pam_systemd and systemd-logind enforce at login.
5355 * The userdbctl tool gained two new switches --with-dropin= and
5356 --with-varlink= which can be used to fine-tune the sources used for
5357 user database lookups.
5359 * systemd-nspawn gained a new switch --bind-user= for binding a host
5360 user account into the container. This does three things: the user's
5361 home directory is bind mounted from the host into the container,
5362 below the /run/userdb/home/ hierarchy. A free UID is picked in the
5363 container, and a user namespacing UID mapping to the host user's UID
5364 installed. And finally, a minimal JSON user and group record (along
5365 with its hashed password) is dropped into /run/host/userdb/. These
5366 records are picked up automatically by the userdb drop-in logic
5367 describe above, and allow the user to login with the same password as
5368 on the host. Effectively this means: if host and container run new
5369 enough systemd versions making a host user available to the container
5370 is trivially simple.
5372 * systemd-journal-gatewayd now supports the switches --user, --system,
5373 --merge, --file= that are equivalent to the same switches of
5374 journalctl, and permit exposing only the specified subset of the
5377 * The OnFailure= dependency between units is now augmented with a
5378 implicit reverse dependency OnFailureOf= (this new dependency cannot
5379 be configured directly it's only created as effect of an OnFailure=
5380 dependency in the reverse order — it's visible in "systemctl show"
5381 however). Similar, Slice= now has an reverse dependency SliceOf=,
5382 that is also not configurable directly, but useful to determine all
5383 units that are members of a slice.
5385 * A pair of new dependency types between units PropagatesStopTo= +
5386 StopPropagatedFrom= has been added, that allows propagation of unit
5387 stop events between two units. It operates similar to the existing
5388 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
5390 * A new dependency type OnSuccess= has been added (plus the reverse
5391 dependency OnSuccessOf=, which cannot be configured directly, but
5392 exists only as effect of the reverse OnSuccess=). It is similar to
5393 OnFailure=, but triggers in the opposite case: when a service exits
5394 cleanly. This allows "chaining up" of services where one or more
5395 services are started once another service has successfully completed.
5397 * A new dependency type Upholds= has been added (plus the reverse
5398 dependency UpheldBy=, which cannot be configured directly, but exists
5399 only as effect of Upholds=). This dependency type is a stronger form
5400 of Wants=: if a unit has an UpHolds= dependency on some other unit
5401 and the former is active then the latter is started whenever it is
5402 found inactive (and no job is queued for it). This is an alternative
5403 to Restart= inside service units, but less configurable, and the
5404 request to uphold a unit is not encoded in the unit itself but in
5405 another unit that intends to uphold it.
5407 * The systemd-ask-password tool now also supports reading passwords
5408 from the credentials subsystem, via the new --credential= switch.
5410 * The systemd-ask-password tool learnt a new switch --emoji= which may
5411 be used to explicit control whether the lock and key emoji (🔐) is
5412 shown in the password prompt on suitable TTYs.
5414 * The --echo switch of systemd-ask-password now optionally takes a
5415 parameter that controls character echo. It may either show asterisks
5416 (default, as before), turn echo off entirely, or echo the typed
5417 characters literally.
5419 * The systemd-ask-password tool also gained a new -n switch for
5420 suppressing output of a trailing newline character when writing the
5421 acquired password to standard output, similar to /bin/echo's -n
5424 * New documentation has been added that describes the organization of
5425 the systemd source code tree:
5427 https://systemd.io/ARCHITECTURE
5429 * Units using ConditionNeedsUpdate= will no longer be activated in
5432 * It is now possible to list a template unit in the WantedBy= or
5433 RequiredBy= settings of the [Install] section of another template
5434 unit, which will be instantiated using the same instance name.
5436 * A new MemoryAvailable property is available for units. If the unit,
5437 or the slices it is part of, have a memory limit set via MemoryMax=/
5438 MemoryHigh=, MemoryAvailable will indicate how much more memory the
5439 unit can claim before hitting the limits.
5441 * systemd-coredump will now try to stay below the cgroup memory limit
5442 placed on itself or one of the slices it runs under, if the storage
5443 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
5444 since files written on such filesystems count toward the cgroup memory
5445 limit. If there is not enough available memory in such cases to store
5446 the core file uncompressed, systemd-coredump will skip to compressed
5447 storage directly (if enabled) and it will avoid analyzing the core file
5448 to print backtrace and metadata in the journal.
5450 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
5451 of a path matches the configured expectations, and remove it if not.
5453 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
5454 specify which of the several available filesystem timestamps (access
5455 time, birth time, change time, modification time) to look at when
5456 deciding whether a path has aged enough to be cleaned.
5458 * A new IPv6StableSecretAddress= setting has been added to .network
5459 files, which takes an IPv6 address to use as secret for IPv6 address
5462 * The [DHCPServer] logic in .network files gained support for a new
5463 UplinkInterface= setting that permits configuration of the uplink
5464 interface name to propagate DHCP lease information from.
5466 * The WakeOnLan= setting in .link files now accepts a list of flags
5467 instead of a single one, to configure multiple wake-on-LAN policies.
5469 * User-space defined tracepoints (USDT) have been added to udev at
5470 strategic locations. This is useful for tracing udev behaviour and
5471 performance with bpftrace and similar tools.
5473 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
5474 setting a network timeout time.
5476 * If a system service is running in a new mount namespace (RootDirectory=
5477 and friends), all file systems will be mounted with MS_NOSUID by
5478 default, unless the system is running with SELinux enabled.
5480 * When enumerating time zones the timedatectl tool will now consult the
5481 'tzdata.zi' file shipped by the IANA time zone database package, in
5482 addition to 'zone1970.tab', as before. This makes sure time zone
5483 aliases are now correctly supported. Some distributions so far did
5484 not install this additional file, most do however. If you
5485 distribution does not install it yet, it might make sense to change
5488 * Intel HID rfkill event is no longer masked, since it's the only
5489 source of rfkill event on newer HP laptops. To have both backward and
5490 forward compatibility, userspace daemon needs to debounce duplicated
5491 events in a short time window.
5493 Contributions from: Aakash Singh, adrian5, Albert Brox,
5494 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
5495 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
5496 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
5497 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
5498 Christian Hesse, Daniel Schaefer, Dan Streetman,
5499 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
5500 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
5501 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
5502 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
5503 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
5504 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
5505 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
5506 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
5507 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
5508 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
5509 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
5510 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
5511 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
5512 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
5513 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
5514 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
5515 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
5516 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
5517 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
5518 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
5519 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
5520 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
5521 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
5522 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
5523 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
5525 — Edinburgh, 2021-07-07
5529 * A concept of system extension images is introduced. Such images may
5530 be used to extend the /usr/ and /opt/ directory hierarchies at
5531 runtime with additional files (even if the file system is read-only).
5532 When a system extension image is activated, its /usr/ and /opt/
5533 hierarchies and os-release information are combined via overlayfs
5534 with the file system hierarchy of the host OS.
5536 A new systemd-sysext tool can be used to merge, unmerge, list, and
5537 refresh system extension hierarchies. See
5538 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
5540 The systemd-sysext.service automatically merges installed system
5541 extensions during boot (before basic.target, but not in very early
5542 boot, since various file systems have to be mounted first).
5544 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
5545 supported system extension level.
5547 * A new ExtensionImages= unit setting can be used to apply the same
5548 system extension image concept from systemd-sysext to the namespaced
5549 file hierarchy of specific services, following the same rules and
5552 * Support for a new special "root=tmpfs" kernel command-line option has
5553 been added. When specified, a tmpfs is mounted on /, and mount.usr=
5554 should be used to point to the operating system implementation.
5556 * A new configuration file /etc/veritytab may be used to configure
5557 dm-verity integrity protection for block devices. Each line is in the
5558 format "volume-name data-device hash-device roothash options",
5559 similar to /etc/crypttab.
5561 * A new kernel command-line option systemd.verity.root_options= may be
5562 used to configure dm-verity behaviour for the root device.
5564 * The key file specified in /etc/crypttab (the third field) may now
5565 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
5566 acquired by connecting to that socket and reading from it. This
5567 allows the implementation of a service to provide key information
5568 dynamically, at the moment when it is needed.
5570 * When the hostname is set explicitly to "localhost", systemd-hostnamed
5571 will respect this. Previously such a setting would be mostly silently
5572 ignored. The goal is to honour configuration as specified by the
5575 * The fallback hostname that will be used by the system manager and
5576 systemd-hostnamed can now be configured in two new ways: by setting
5577 DEFAULT_HOSTNAME= in os-release(5), or by setting
5578 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
5579 also be configured during compilation. The environment variable is
5580 intended for testing and local overrides, the os-release(5) field is
5581 intended to allow customization by different variants of a
5582 distribution that share the same compiled packages.
5584 * The environment block of the manager itself may be configured through
5585 a new ManagerEnvironment= setting in system.conf or user.conf. This
5586 complements existing ways to set the environment block (the kernel
5587 command line for the system manager, the inherited environment and
5588 user@.service unit file settings for the user manager).
5590 * systemd-hostnamed now exports the default hostname and the source of
5591 the configured hostname ("static", "transient", or "default") as
5594 * systemd-hostnamed now exports the "HardwareVendor" and
5595 "HardwareModel" D-Bus properties, which are supposed to contain a
5596 pair of cleaned up, human readable strings describing the system's
5597 vendor and model. It's typically sourced from the firmware's DMI
5598 tables, but may be augmented from a new hwdb database. hostnamectl
5599 shows this in the status output.
5601 * Support has been added to systemd-cryptsetup for extracting the
5602 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
5603 metadata header. This allows the information how to open the
5604 encrypted device to be embedded directly in the device and obviates
5605 the need for configuration in an external file.
5607 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
5608 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
5609 pre-existing support for PKCS#11 security tokens).
5611 * systemd-repart may enroll encrypted partitions using TPM2
5612 hardware. This may be useful for example to create an encrypted /var
5613 partition bound to the machine on first boot.
5615 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
5616 and PKCS#11 security tokens to LUKS volumes, list and destroy
5619 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
5621 It also supports enrolling "recovery keys" and regular passphrases.
5623 * The libfido2 dependency is now based on dlopen(), so that the library
5624 is used at runtime when installed, but is not a hard runtime
5627 * systemd-cryptsetup gained support for two new options in
5628 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
5629 request synchronous processing of encryption/decryption IO.
5631 * The manager may be configured at compile time to use the fexecve()
5632 instead of the execve() system call when spawning processes. Using
5633 fexecve() closes a window between checking the security context of an
5634 executable and spawning it, but unfortunately the kernel displays
5635 stale information in the process' "comm" field, which impacts ps
5638 * The configuration option -Dcompat-gateway-hostname has been dropped.
5639 "_gateway" is now the only supported name.
5641 * The ConditionSecurity=tpm2 unit file setting may be used to check if
5642 the system has at least one TPM2 (tpmrm class) device.
5644 * A new ConditionCPUFeature= has been added that may be used to
5645 conditionalize units based on CPU features. For example,
5646 ConditionCPUFeature=rdrand will condition a unit so that it is only
5647 run when the system CPU supports the RDRAND opcode.
5649 * The existing ConditionControlGroupController= setting has been
5650 extended with two new values "v1" and "v2". "v2" means that the
5651 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
5652 hierarchy or the hybrid hierarchy are used.
5654 * A new PrivateIPC= setting on a unit file allows executed processes to
5655 be moved into a private IPC namespace, with separate System V IPC
5656 identifiers and POSIX message queues.
5658 A new IPCNamespacePath= allows the unit to be joined to an existing
5661 * The tables of system calls in seccomp filters are now automatically
5662 generated from kernel lists exported on
5663 https://fedora.juszkiewicz.com.pl/syscalls.html.
5665 The following architectures should now have complete lists:
5666 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
5667 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
5669 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
5670 on /run/ if it is not already a mount point. A writable /run/ has
5671 always been a requirement for a functioning system, but this was not
5672 guaranteed when using a read-only image.
5674 Users can always specify BindPaths= or InaccessiblePaths= as
5675 overrides, and they will take precedence. If the host's root mount
5676 point is used, there is no change in behaviour.
5678 * New bind mounts and file system image mounts may be injected into the
5679 mount namespace of a service (without restarting it). This is exposed
5680 respectively as 'systemctl bind <unit> <path>…' and
5681 'systemctl mount-image <unit> <image>…'.
5683 * The StandardOutput= and StandardError= settings can now specify files
5684 to be truncated for output (as "truncate:<path>").
5686 * The ExecPaths= and NoExecPaths= settings may be used to specify
5687 noexec for parts of the file system.
5689 * sd-bus has a new function sd_bus_open_user_machine() to open a
5690 connection to the session bus of a specific user in a local container
5691 or on the local host. This is exposed in the existing -M switch to
5692 systemctl and similar tools:
5694 systemctl --user -M lennart@foobar start foo
5696 This will connect to the user bus of a user "lennart" in container
5697 "foobar". If no container name is specified, the specified user on
5698 the host itself is connected to
5700 systemctl --user -M lennart@ start quux
5702 * sd-bus also gained a convenience function sd_bus_message_send() to
5703 simplify invocations of sd_bus_send(), taking only a single
5704 parameter: the message to send.
5706 * sd-event allows rate limits to be set on event sources, for dealing
5707 with high-priority event sources that might starve out others. See
5708 the new man page sd_event_source_set_ratelimit(3) for details.
5710 * systemd.link files gained a [Link] Promiscuous= switch, which allows
5711 the device to be raised in promiscuous mode.
5713 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
5714 number of TX and RX queues to be configured.
5716 New [Link] TransmitQueueLength= setting allows the size of the TX
5717 queue to be configured.
5719 New [Link] GenericSegmentOffloadMaxBytes= and
5720 GenericSegmentOffloadMaxSegments= allow capping the packet size and
5721 the number of segments accepted in Generic Segment Offload.
5723 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
5724 wireless routing protocol that operates on ISO/OSI Layer 2 only and
5725 uses ethernet frames to route/bridge packets. This encompasses a new
5726 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
5727 new settings in .netdev files, and a new BatmanAdvanced= setting in
5730 * systemd.network files gained a [Network] RouteTable= configuration
5731 switch to select the routing policy table.
5733 systemd.network files gained a [RoutingPolicyRule] Type=
5734 configuration switch (one of "blackhole, "unreachable", "prohibit").
5736 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
5737 RouteAllowList= settings to ignore/accept route advertisements from
5738 routers matching specified prefixes. The DenyList= setting has been
5739 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
5742 systemd.network files gained a [DHCPv6] UseAddress= setting to
5743 optionally ignore the address provided in the lease.
5745 systemd.network files gained a [DHCPv6PrefixDelegation]
5746 ManageTemporaryAddress= switch.
5748 systemd.network files gained a new ActivationPolicy= setting which
5749 allows configuring how the UP state of an interface shall be managed,
5750 i.e. whether the interface is always upped, always downed, or may be
5751 upped/downed by the user using "ip link set dev".
5753 * The default for the Broadcast= setting in .network files has slightly
5754 changed: the broadcast address will not be configured for wireguard
5757 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
5758 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
5759 configuration options for VLAN packet handling.
5761 * udev rules may now set log_level= option. This allows debug logs to
5762 be enabled for select events, e.g. just for a specific subsystem or
5763 even a single device.
5765 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
5766 DATA_PREPARED_ID properties for block devices with ISO9660 file
5769 * udev now exports decoded DMI information about installed memory slots
5770 as device properties under the /sys/class/dmi/id/ pseudo device.
5772 * /dev/ is not mounted noexec anymore. This didn't provide any
5773 significant security benefits and would conflict with the executable
5774 mappings used with /dev/sgx device nodes. The previous behaviour can
5775 be restored for individual services with NoExecPaths=/dev (or by allow-
5776 listing and excluding /dev from ExecPaths=).
5778 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
5779 and /dev/vhost-net are owned by the kvm group.
5781 * The hardware database has been extended with a list of fingerprint
5782 readers that correctly support USB auto-suspend using data from
5785 * systemd-resolved can now answer DNSSEC questions through the stub
5786 resolver interface in a way that allows local clients to do DNSSEC
5787 validation themselves. For a question with DO+CD set, it'll proxy the
5788 DNS query and respond with a mostly unmodified packet received from
5789 the upstream server.
5791 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
5792 resolved.conf. If true the service will provide caching even for DNS
5793 lookups made to an upstream DNS server on the 127.0.0.1/::1
5794 addresses. By default (and when the option is false) systemd-resolved
5795 will not cache such lookups, in order to avoid duplicate local
5796 caching, under the assumption the local upstream server caches
5799 * systemd-resolved now implements RFC5001 NSID in its local DNS
5800 stub. This may be used by local clients to determine whether they are
5801 talking to the DNS resolver stub or a different DNS server.
5803 * When resolving host names and other records resolvectl will now
5804 report where the data was acquired from (i.e. the local cache, the
5805 network, locally synthesized, …) and whether the network traffic it
5806 effected was encrypted or not. Moreover the tool acquired a number of
5807 new options --cache=, --synthesize=, --network=, --zone=,
5808 --trust-anchor=, --validate= that take booleans and may be used to
5809 tweak a lookup, i.e. whether it may be answered from cached
5810 information, locally synthesized information, information acquired
5811 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
5812 anchor, and whether DNSSEC validation shall be executed for the
5815 * systemd-nspawn gained a new --ambient-capability= setting
5816 (AmbientCapability= in .nspawn files) to configure ambient
5817 capabilities passed to the container payload.
5819 * systemd-nspawn gained the ability to configure the firewall using the
5820 nftables subsystem (in addition to the existing iptables
5821 support). Similarly, systemd-networkd's IPMasquerade= option now
5822 supports nftables as back-end, too. In both cases NAT on IPv6 is now
5823 supported too, in addition to IPv4 (the iptables back-end still is
5826 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
5827 retains its meaning, but has been deprecated. Please switch to either
5828 "ivp4" or "both" (if covering IPv6 is desired).
5830 * systemd-importd will now download .verity and .roothash.p7s files
5831 along with the machine image (as exposed via machinectl pull-raw).
5833 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
5834 setting to configure the time a unit's cgroup needs to exceed memory
5835 pressure limits before action will be taken, and a new
5836 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
5839 systemd-oomd is now considered fully supported (the usual
5840 backwards-compatibility promises apply). Swap is not required for
5841 operation, but it is still recommended.
5843 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
5844 configures the retry delay when trying to contact servers.
5846 * systemd-stdio-bridge gained --system/--user options to connect to the
5847 system bus (previous default) or the user session bus.
5849 * systemd-localed may now call locale-gen to generate missing locales
5850 on-demand (UTF-8-only). This improves integration with Debian-based
5851 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
5853 * systemctl --check-inhibitors=true may now be used to obey inhibitors
5854 even when invoked non-interactively. The old --ignore-inhibitors
5855 switch is now deprecated and replaced by --check-inhibitors=false.
5857 * systemctl import-environment will now emit a warning when called
5858 without any arguments (i.e. to import the full environment block of
5859 the called program). This command will usually be invoked from a
5860 shell, which means that it'll inherit a bunch of variables which are
5861 specific to that shell, and usually to the TTY the shell is connected
5862 to, and don't have any meaning in the global context of the system or
5863 user service manager. Instead, only specific variables should be
5864 imported into the manager environment block.
5866 Similarly, programs which update the manager environment block by
5867 directly calling the D-Bus API of the manager, should also push
5868 specific variables, and not the full inherited environment.
5870 * systemctl's status output now shows unit state with a more careful
5871 choice of Unicode characters: units in maintenance show a "○" symbol
5872 instead of the usual "●", failed units show "×", and services being
5875 * coredumpctl gained a --debugger-arguments= switch to pass arguments
5876 to the debugger. It also gained support for showing coredump info in
5877 a simple JSON format.
5879 * systemctl/loginctl/machinectl's --signal= option now accept a special
5880 value "list", which may be used to show a brief table with known
5881 process signals and their numbers.
5883 * networkctl now shows the link activation policy in status.
5885 * Various tools gained --pager/--no-pager/--json= switches to
5886 enable/disable the pager and provide JSON output.
5888 * Various tools now accept two new values for the SYSTEMD_COLORS
5889 environment variable: "16" and "256", to configure how many terminal
5890 colors are used in output.
5892 * less 568 or newer is now required for the auto-paging logic of the
5893 various tools. Hyperlink ANSI sequences in terminal output are now
5894 used even if a pager is used, and older versions of less are not able
5895 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
5896 disable this output again.
5898 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
5899 builds, non-merged-usr builds) are now officially deprecated. A
5900 warning is emitted during build. Support is slated to be removed in
5901 about a year (when the Debian Bookworm release development starts).
5903 * Systems with the legacy cgroup v1 hierarchy are now marked as
5904 "tainted", to make it clearer that using the legacy hierarchy is not
5907 * systemd-localed will now refuse to configure a keymap which is not
5908 installed in the file system. This is intended as a bug fix, but
5909 could break cases where systemd-localed was used to configure the
5910 keymap in advanced of it being installed. It is necessary to install
5911 the keymap file first.
5913 * The main git development branch has been renamed to 'main'.
5915 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
5916 for partitions, as in the vast majority of cases they contain none
5917 and are used internally by the bootloader (eg: uboot).
5919 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
5920 spawned processes to the PID of the process itself. This may be used
5921 by programs for detecting whether they were forked off by the service
5922 manager itself or are a process forked off further down the tree.
5924 * The sd-device API gained four new calls: sd_device_get_action() to
5925 determine the uevent add/remove/change/… action the device object has
5926 been seen for, sd_device_get_seqno() to determine the uevent sequence
5927 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
5928 object from stat(2) data of a device node, and sd_device_trigger() to
5929 write to the 'uevent' attribute of a device.
5931 * For most tools the --no-legend= switch has been replaced by
5932 --legend=no and --legend=yes, to force whether tables are shown with
5935 * Units acquired a new property "Markers" that takes a list of zero,
5936 one or two of the following strings: "needs-reload" and
5937 "needs-restart". These markers may be set via "systemctl
5938 set-property". Once a marker is set, "systemctl reload-or-restart
5939 --marked" may be invoked to execute the operation the units are
5940 marked for. This is useful for package managers that want to mark
5941 units for restart/reload while updating, but effect the actual
5942 operations at a later step at once.
5944 * The sd_bus_message_read_strv() API call of sd-bus may now also be
5945 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
5948 * bootctl will now report whether the UEFI firmware used a TPM2 device
5949 and measured the boot process into it.
5951 * systemd-tmpfiles learnt support for a new environment variable
5952 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
5953 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
5954 even if the root fs of the system is not itself a btrfs volume.
5956 * systemd-detect-virt/ConditionVirtualization= will now explicitly
5957 detect Docker/Podman environments where possible. Moreover, they
5958 should be able to generically detect any container manager as long as
5959 it assigns the container a cgroup.
5961 * portablectl gained a new "reattach" verb for detaching/reattaching a
5962 portable service image, useful for updating images on-the-fly.
5964 * Intel SGX enclave device nodes (which expose a security feature of
5965 newer Intel CPUs) will now be owned by a new system group "sgx".
5967 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
5968 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
5969 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
5970 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
5971 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
5972 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
5973 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
5974 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
5975 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
5976 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
5977 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
5978 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
5979 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
5980 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
5981 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
5982 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
5983 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
5984 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
5985 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
5986 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
5987 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
5988 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
5989 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
5990 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
5991 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
5992 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
5993 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
5994 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
5995 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
5996 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
5997 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
5998 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
5999 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
6000 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
6001 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
6002 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
6003 Zmicer Turok, Дамјан Георгиевски
6005 — Berlin, 2021-03-30
6009 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
6010 "bind" and "unbind" to the Linux device model. When this kernel
6011 change was made, systemd-udevd was only minimally updated to handle
6012 and propagate these new event types. The introduction of these new
6013 uevents (which are typically generated for USB devices and devices
6014 needing a firmware upload before being functional) resulted in a
6015 number of issues which we so far didn't address. We hoped the kernel
6016 maintainers would themselves address these issues in some form, but
6017 that did not happen. To handle them properly, many (if not most) udev
6018 rules files shipped in various packages need updating, and so do many
6019 programs that monitor or enumerate devices with libudev or sd-device,
6020 or otherwise process uevents. Please note that this incompatibility
6021 is not fault of systemd or udev, but caused by an incompatible kernel
6022 change that happened back in Linux 4.14, but is becoming more and
6023 more visible as the new uevents are generated by more kernel drivers.
6025 To minimize issues resulting from this kernel change (but not avoid
6026 them entirely) starting with systemd-udevd 247 the udev "tags"
6027 concept (which is a concept for marking and filtering devices during
6028 enumeration and monitoring) has been reworked: udev tags are now
6029 "sticky", meaning that once a tag is assigned to a device it will not
6030 be removed from the device again until the device itself is removed
6031 (i.e. unplugged). This makes sure that any application monitoring
6032 devices that match a specific tag is guaranteed to both see uevents
6033 where the device starts being relevant, and those where it stops
6034 being relevant (the latter now regularly happening due to the new
6035 "unbind" uevent type). The udev tags concept is hence now a concept
6036 tied to a *device* instead of a device *event* — unlike for example
6037 udev properties whose lifecycle (as before) is generally tied to a
6038 device event, meaning that the previously determined properties are
6039 forgotten whenever a new uevent is processed.
6041 With the newly redefined udev tags concept, sometimes it's necessary
6042 to determine which tags are the ones applied by the most recent
6043 uevent/database update, in order to discern them from those
6044 originating from earlier uevents/database updates of the same
6045 device. To accommodate for this a new automatic property CURRENT_TAGS
6046 has been added that works similar to the existing TAGS property but
6047 only lists tags set by the most recent uevent/database
6048 update. Similarly, the libudev/sd-device API has been updated with
6049 new functions to enumerate these 'current' tags, in addition to the
6050 existing APIs that now enumerate the 'sticky' ones.
6052 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
6053 essential that all udev rules files and applications are updated to
6054 handle the new events. Specifically:
6056 • All rule files that currently use a header guard similar to
6057 ACTION!="add|change",GOTO="xyz_end" should be updated to use
6058 ACTION=="remove",GOTO="xyz_end" instead, so that the
6059 properties/tags they add are also applied whenever "bind" (or
6060 "unbind") is seen. (This is most important for all physical device
6061 types — those for which "bind" and "unbind" are currently
6062 generated, for all other device types this change is still
6063 recommended but not as important — but certainly prepares for
6064 future kernel uevent type additions).
6066 • Similarly, all code monitoring devices that contains an 'if' branch
6067 discerning the "add" + "change" uevent actions from all other
6068 uevents actions (i.e. considering devices only relevant after "add"
6069 or "change", and irrelevant on all other events) should be reworked
6070 to instead negatively check for "remove" only (i.e. considering
6071 devices relevant after all event types, except for "remove", which
6072 invalidates the device). Note that this also means that devices
6073 should be considered relevant on "unbind", even though conceptually
6074 this — in some form — invalidates the device. Since the precise
6075 effect of "unbind" is not generically defined, devices should be
6076 considered relevant even after "unbind", however I/O errors
6077 accessing the device should then be handled gracefully.
6079 • Any code that uses device tags for deciding whether a device is
6080 relevant or not most likely needs to be updated to use the new
6081 udev_device_has_current_tag() API (or sd_device_has_current_tag()
6082 in case sd-device is used), to check whether the tag is set at the
6083 moment an uevent is seen (as opposed to the existing
6084 udev_device_has_tag() API which checks if the tag ever existed on
6085 the device, following the API concept redefinition explained
6088 We are very sorry for this breakage and the requirement to update
6089 packages using these interfaces. We'd again like to underline that
6090 this is not caused by systemd/udev changes, but result of a kernel
6093 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
6094 packages have not retriggered devices once the udev package (or any
6095 auxiliary package installing additional udev rules) is updated. We
6096 intend to work with major distributions to change this, so that
6097 "udevadm trigger -c change" is issued on such upgrades, ensuring that
6098 the updated ruleset is applied to the devices already discovered, so
6099 that (asynchronously) after the upgrade completed the udev database
6100 is consistent with the updated rule set. This means udev rules must
6101 be ready to be retriggered with a "change" action any time, and
6102 result in correct and complete udev database entries. While the
6103 majority of udev rule files known to us currently get this right,
6104 some don't. Specifically, there are udev rules files included in
6105 various packages that only set udev properties on the "add" action,
6106 but do not handle the "change" action. If a device matching those
6107 rules is retriggered with the "change" action (as is intended here)
6108 it would suddenly lose the relevant properties. This always has been
6109 problematic, but as soon as all udev devices are triggered on relevant
6110 package upgrades this will become particularly so. It is strongly
6111 recommended to fix offending rules so that they can handle a "change"
6112 action at any time, and acquire all necessary udev properties even
6113 then. Or in other words: the header guard mentioned above
6114 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
6115 this, as it makes sure rules are rerun on "change" correctly, and
6116 accumulate the correct and complete set of udev properties. udev rule
6117 definitions that cannot handle "change" events being triggered at
6118 arbitrary times should be considered buggy.
6120 * The MountAPIVFS= service file setting now defaults to on if
6121 RootImage= and RootDirectory= are used, which means that with those
6122 two settings /proc/, /sys/ and /dev/ are automatically properly set
6123 up for services. Previous behaviour may be restored by explicitly
6124 setting MountAPIVFS=off.
6126 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
6127 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
6128 latter it takes precedence over the former, similar to how most of
6129 systemd's own configuration is handled. Given that PAM stack
6130 definitions are primarily put together by OS vendors/distributions
6131 (though possibly overridden by users), this systemd release moves its
6132 own PAM stack configuration for the "systemd-user" PAM service (i.e.
6133 for the PAM session invoked by the per-user user@.service instance)
6134 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
6135 packages' vendor versions of their PAM stack definitions from
6136 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
6137 desired the location to which systemd installs its PAM stack
6138 configuration may be changed via the -Dpamconfdir Meson option.
6140 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
6141 libpwquality and libcryptsetup have been changed to be based on
6142 dlopen(): instead of regular dynamic library dependencies declared in
6143 the binary ELF headers, these libraries are now loaded on demand
6144 only, if they are available. If the libraries cannot be found the
6145 relevant operations will fail gracefully, or a suitable fallback
6146 logic is chosen. This is supposed to be useful for general purpose
6147 distributions, as it allows minimizing the list of dependencies the
6148 systemd packages pull in, permitting building of more minimal OS
6149 images, while still making use of these "weak" dependencies should
6150 they be installed. Since many package managers automatically
6151 synthesize package dependencies from ELF shared library dependencies,
6152 some additional manual packaging work has to be done now to replace
6153 those (slightly downgraded from "required" to "recommended" or
6154 whatever is conceptually suitable for the package manager). Note that
6155 this change does not alter build-time behaviour: as before the
6156 build-time dependencies have to be installed during build, even if
6157 they now are optional during runtime.
6159 * sd-event.h gained a new call sd_event_add_time_relative() for
6160 installing timers relative to the current time. This is mostly a
6161 convenience wrapper around the pre-existing sd_event_add_time() call
6162 which installs absolute timers.
6164 * sd-event event sources may now be placed in a new "exit-on-failure"
6165 mode, which may be controlled via the new
6166 sd_event_source_get_exit_on_failure() and
6167 sd_event_source_set_exit_on_failure() functions. If enabled, any
6168 failure returned by the event source handler functions will result in
6169 exiting the event loop (unlike the default behaviour of just
6170 disabling the event source but continuing with the event loop). This
6171 feature is useful to set for all event sources that define "primary"
6172 program behaviour (where failure should be fatal) in contrast to
6173 "auxiliary" behaviour (where failure should remain local).
6175 * Most event source types sd-event supports now accept a NULL handler
6176 function, in which case the event loop is exited once the event
6177 source is to be dispatched, using the userdata pointer — converted to
6178 a signed integer — as exit code of the event loop. Previously this
6179 was supported for IO and signal event sources already. Exit event
6180 sources still do not support this (simply because it makes little
6181 sense there, as the event loop is already exiting when they are
6184 * A new per-unit setting RootImageOptions= has been added which allows
6185 tweaking the mount options for any file system mounted as effect of
6186 the RootImage= setting.
6188 * Another new per-unit setting MountImages= has been added, that allows
6189 mounting additional disk images into the file system tree accessible
6192 * Timer units gained a new FixedRandomDelay= boolean setting. If
6193 enabled, the random delay configured with RandomizedDelaySec= is
6194 selected in a way that is stable on a given system (though still
6195 different for different units).
6197 * Socket units gained a new setting Timestamping= that takes "us", "ns"
6198 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
6201 * systemd-repart now generates JSON output when requested with the new
6204 * systemd-machined's OpenMachineShell() bus call will now pass
6205 additional policy metadata data fields to the PolicyKit
6206 authentication request.
6208 * systemd-tmpfiles gained a new -E switch, which is equivalent to
6209 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
6210 --exclude=/sys. It's particularly useful in combination with --root=,
6211 when operating on OS trees that do not have any of these four runtime
6212 directories mounted, as this means no files below these subtrees are
6213 created or modified, since those mount points should probably remain
6216 * systemd-tmpfiles gained a new --image= switch which is like --root=,
6217 but takes a disk image instead of a directory as argument. The
6218 specified disk image is mounted inside a temporary mount namespace
6219 and the tmpfiles.d/ drop-ins stored in the image are executed and
6220 applied to the image. systemd-sysusers similarly gained a new
6221 --image= switch, that allows the sysusers.d/ drop-ins stored in the
6222 image to be applied onto the image.
6224 * Similarly, the journalctl command also gained an --image= switch,
6225 which is a quick one-step solution to look at the log data included
6228 * journalctl's --output=cat option (which outputs the log content
6229 without any metadata, just the pure text messages) will now make use
6230 of terminal colors when run on a suitable terminal, similarly to the
6233 * JSON group records now support a "description" string that may be
6234 used to add a human-readable textual description to such groups. This
6235 is supposed to match the user's GECOS field which traditionally
6236 didn't have a counterpart for group records.
6238 * The "systemd-dissect" tool that may be used to inspect OS disk images
6239 and that was previously installed to /usr/lib/systemd/ has now been
6240 moved to /usr/bin/, reflecting its updated status of an officially
6241 supported tool with a stable interface. It gained support for a new
6242 --mkdir switch which when combined with --mount has the effect of
6243 creating the directory to mount the image to if it is missing
6244 first. It also gained two new commands --copy-from and --copy-to for
6245 copying files and directories in and out of an OS image without the
6246 need to manually mount it. It also acquired support for a new option
6247 --json= to generate JSON output when inspecting an OS image.
6249 * The cgroup2 file system is now mounted with the
6250 "memory_recursiveprot" mount option, supported since kernel 5.7. This
6251 means that the MemoryLow= and MemoryMin= unit file settings now apply
6252 recursively to whole subtrees.
6254 * systemd-homed now defaults to using the btrfs file system — if
6255 available — when creating home directories in LUKS volumes. This may
6256 be changed with the DefaultFileSystemType= setting in homed.conf.
6257 It's now the default file system in various major distributions and
6258 has the major benefit for homed that it can be grown and shrunk while
6259 mounted, unlike the other contenders ext4 and xfs, which can both be
6260 grown online, but not shrunk (in fact xfs is the technically most
6261 limited option here, as it cannot be shrunk at all).
6263 * JSON user records managed by systemd-homed gained support for
6264 "recovery keys". These are basically secondary passphrases that can
6265 unlock user accounts/home directories. They are computer-generated
6266 rather than user-chosen, and typically have greater entropy.
6267 homectl's --recovery-key= option may be used to add a recovery key to
6268 a user account. The generated recovery key is displayed as a QR code,
6269 so that it can be scanned to be kept in a safe place. This feature is
6270 particularly useful in combination with systemd-homed's support for
6271 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
6272 security tokens are lost. Recovery keys may be entered wherever the
6273 system asks for a password.
6275 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
6276 home directory which indicates that a home directory has not been
6277 deactivated cleanly when offline. This flag is useful to identify
6278 home directories for which the offline discard logic did not run when
6279 offlining, and where it would be a good idea to log in again to catch
6282 * systemctl gained a new parameter --timestamp= which may be used to
6283 change the style in which timestamps are output, i.e. whether to show
6284 them in local timezone or UTC, or whether to show µs granularity.
6286 * Alibaba's "pouch" container manager is now detected by
6287 systemd-detect-virt, ConditionVirtualization= and similar
6288 constructs. Similar, they now also recognize IBM PowerVM machine
6291 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
6292 place to use for propagating external mounts into the
6293 container. Similarly /run/host/notify is now used as the socket path
6294 for container payloads to communicate with the container manager
6295 using sd_notify(). The container manager now uses the
6296 /run/host/inaccessible/ directory to place "inaccessible" file nodes
6297 of all relevant types which may be used by the container payload as
6298 bind mount source to over-mount inodes to make them inaccessible.
6299 /run/host/container-manager will now be initialized with the same
6300 string as the $container environment variable passed to the
6301 container's PID 1. /run/host/container-uuid will be initialized with
6302 the same string as $container_uuid. This means the /run/host/
6303 hierarchy is now the primary way to make host resources available to
6304 the container. The Container Interface documents these new files and
6307 https://systemd.io/CONTAINER_INTERFACE
6309 * Support for the "ConditionNull=" unit file condition has been
6310 deprecated and undocumented for 6 years. systemd started to warn
6311 about its use 1.5 years ago. It has now been removed entirely.
6313 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
6314 a sd_bus_error struct and a list of error names, and checks if the
6315 error matches one of these names. It's a convenience wrapper that is
6316 useful in cases where multiple errors shall be handled the same way.
6318 * A new system call filter list "@known" has been added, that contains
6319 all system calls known at the time systemd was built.
6321 * Behaviour of system call filter allow lists has changed slightly:
6322 system calls that are contained in @known will result in EPERM by
6323 default, while those not contained in it result in ENOSYS. This
6324 should improve compatibility because known system calls will thus be
6325 communicated as prohibited, while unknown (and thus newer ones) will
6326 be communicated as not implemented, which hopefully has the greatest
6327 chance of triggering the right fallback code paths in client
6330 * "systemd-analyze syscall-filter" will now show two separate sections
6331 at the bottom of the output: system calls known during systemd build
6332 time but not included in any of the filter groups shown above, and
6333 system calls defined on the local kernel but known during systemd
6336 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
6337 systemd-nspawn all system call filter violations will be logged by
6338 the kernel (audit). This is useful for tracking down system calls
6339 invoked by container payloads that are prohibited by the container's
6340 system call filter policy.
6342 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
6343 systemd-nspawn (and other programs that use seccomp) all seccomp
6344 filtering is turned off.
6346 * Two new unit file settings ProtectProc= and ProcSubset= have been
6347 added that expose the hidepid= and subset= mount options of procfs.
6348 All processes of the unit will only see processes in /proc that are
6349 are owned by the unit's user. This is an important new sandboxing
6350 option that is recommended to be set on all system services. All
6351 long-running system services that are included in systemd itself set
6352 this option now. This option is only supported on kernel 5.8 and
6353 above, since the hidepid= option supported on older kernels was not a
6354 per-mount option but actually applied to the whole PID namespace.
6356 * Socket units gained a new boolean setting FlushPending=. If enabled
6357 all pending socket data/connections are flushed whenever the socket
6358 unit enters the "listening" state, i.e. after the associated service
6361 * The unit file setting NUMAMask= gained a new "all" value: when used,
6362 all existing NUMA nodes are added to the NUMA mask.
6364 * A new "credentials" logic has been added to system services. This is
6365 a simple mechanism to pass privileged data to services in a safe and
6366 secure way. It's supposed to be used to pass per-service secret data
6367 such as passwords or cryptographic keys but also associated less
6368 private information such as user names, certificates, and similar to
6369 system services. Each credential is identified by a short user-chosen
6370 name and may contain arbitrary binary data. Two new unit file
6371 settings have been added: SetCredential= and LoadCredential=. The
6372 former allows setting a credential to a literal string, the latter
6373 sets a credential to the contents of a file (or data read from a
6374 user-chosen AF_UNIX stream socket). Credentials are passed to the
6375 service via a special credentials directory, one file for each
6376 credential. The path to the credentials directory is passed in a new
6377 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
6378 are passed in the file system they may be easily referenced in
6379 ExecStart= command lines too, thus no explicit support for the
6380 credentials logic in daemons is required (though ideally daemons
6381 would look for the bits they need in $CREDENTIALS_DIRECTORY
6382 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
6383 backed by unswappable memory if privileges allow it, immutable if
6384 privileges allow it, is accessible only to the service's UID, and is
6385 automatically destroyed when the service stops.
6387 * systemd-nspawn supports the same credentials logic. It can both
6388 consume credentials passed to it via the aforementioned
6389 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
6390 to its payload. The service manager/PID 1 has been updated to match
6391 this: it can also accept credentials from the container manager that
6392 invokes it (in fact: any process that invokes it), and passes them on
6393 to its services. Thus, credentials can be propagated recursively down
6394 the tree: from a system's service manager to a systemd-nspawn
6395 service, to the service manager that runs as container payload and to
6396 the service it runs below. Credentials may also be added on the
6397 systemd-nspawn command line, using new --set-credential= and
6398 --load-credential= command line switches that match the
6399 aforementioned service settings.
6401 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
6402 the partition drop-ins which may be used to format/LUKS
6403 encrypt/populate any created partitions. The partitions are
6404 encrypted/formatted/populated before they are registered in the
6405 partition table, so that they appear atomically: either the
6406 partitions do not exist yet or they exist fully encrypted, formatted,
6407 and populated — there is no time window where they are
6408 "half-initialized". Thus the system is robust to abrupt shutdown: if
6409 the tool is terminated half-way during its operations on next boot it
6410 will start from the beginning.
6412 * systemd-repart's --size= operation gained a new "auto" value. If
6413 specified, and operating on a loopback file it is automatically sized
6414 to the minimal size the size constraints permit. This is useful to
6415 use "systemd-repart" as an image builder for minimally sized images.
6417 * systemd-resolved now gained a third IPC interface for requesting name
6418 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
6419 interface is now supported. The nss-resolve NSS module has been
6420 modified to use this new interface instead of D-Bus. Using Varlink
6421 has a major benefit over D-Bus: it works without a broker service,
6422 and thus already during earliest boot, before the dbus daemon has
6423 been started. This means name resolution via systemd-resolved now
6424 works at the same time systemd-networkd operates: from earliest boot
6425 on, including in the initrd.
6427 * systemd-resolved gained support for a new DNSStubListenerExtra=
6428 configuration file setting which may be used to specify additional IP
6429 addresses the built-in DNS stub shall listen on, in addition to the
6430 main one on 127.0.0.53:53.
6432 * Name lookups issued via systemd-resolved's D-Bus and Varlink
6433 interfaces (and thus also via glibc NSS if nss-resolve is used) will
6434 now honour a trailing dot in the hostname: if specified the search
6435 path logic is turned off. Thus "resolvectl query foo." is now
6436 equivalent to "resolvectl query --search=off foo.".
6438 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
6439 exposes how /etc/resolv.conf is currently managed: by resolved (and
6440 in which mode if so) or another subsystem. "resolvctl" will display
6441 this property in its status output.
6443 * The resolv.conf snippets systemd-resolved provides will now set "."
6444 as the search domain if no other search domain is known. This turns
6445 off the derivation of an implicit search domain by nss-dns for the
6446 hostname, when the hostname is set to an FQDN. This change is done to
6447 make nss-dns using resolv.conf provided by systemd-resolved behave
6448 more similarly to nss-resolve.
6450 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
6451 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
6452 "birth" time (btime) of a file in addition to the atime, mtime, and
6455 * systemd-analyze gained a new verb "capability" that lists all known
6456 capabilities by the systemd build and by the kernel.
6458 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
6459 advance the system clock to it at boot if it is noticed to be before
6460 that time. Previously, PID 1 would only advance the time to an epoch
6461 time that is set during build-time. With this new file OS builders
6462 can change this epoch timestamp on individual OS images without
6463 having to rebuild systemd.
6465 * systemd-logind will now listen to the KEY_RESTART key from the Linux
6466 input layer and reboot the system if it is pressed, similarly to how
6467 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
6468 was originally defined in the Multimedia context (to restart playback
6469 of a song or film), but is now primarily used in various embedded
6470 devices for "Reboot" buttons. Accordingly, systemd-logind will now
6471 honour it as such. This may configured in more detail via the new
6472 HandleRebootKey= and RebootKeyIgnoreInhibited=.
6474 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
6475 copying OS trees, for example in "systemd-nspawn --ephemeral",
6476 "systemd-nspawn --template=", "machinectl clone" and similar. This is
6477 useful when operating with OSTree images, which use hardlinks heavily
6478 throughout, and where such copies previously resulting in "exploding"
6481 * systemd-nspawn's --console= setting gained support for a new
6482 "autopipe" value, which is identical to "interactive" when invoked on
6483 a TTY, and "pipe" otherwise.
6485 * systemd-networkd's .network files gained support for explicitly
6486 configuring the multicast membership entries of bridge devices in the
6487 [BridgeMDB] section. It also gained support for the PIE queuing
6488 discipline in the [FlowQueuePIE] sections.
6490 * systemd-networkd's .netdev files may now be used to create "BareUDP"
6491 tunnels, configured in the new [BareUDP] setting.
6493 * systemd-networkd's Gateway= setting in .network files now accepts the
6494 special values "_dhcp4" and "_ipv6ra" to configure additional,
6495 locally defined, explicit routes to the gateway acquired via DHCP or
6496 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
6497 but still accepted for backwards compatibility.
6499 * systemd-networkd's [IPv6PrefixDelegation] section and
6500 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
6501 IPv6SendRA= (the old names are still accepted for backwards
6504 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
6505 boolean setting in [Network] section. If enabled, the delegated prefix
6506 gained by another link will be configured, and an address within the
6507 prefix will be assigned.
6509 * systemd-networkd's .network files gained the Announce= boolean setting
6510 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
6511 prefix will be announced through IPv6 router advertisement (IPv6 RA).
6512 The setting is enabled by default.
6514 * VXLAN tunnels may now be marked as independent of any underlying
6515 network interface via the new Independent= boolean setting.
6517 * systemctl gained support for two new verbs: "service-log-level" and
6518 "service-log-target" may be used on services that implement the
6519 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
6520 adjust the log level and target. All of systemd's long-running
6521 services support this now, but ideally all system services would
6522 implement this interface to make the system more uniformly
6525 * The SystemCallErrorNumber= unit file setting now accepts the new
6526 "kill" and "log" actions, in addition to arbitrary error number
6527 specifications as before. If "kill" the processes are killed on the
6528 event, if "log" the offending system call is audit logged.
6530 * A new SystemCallLog= unit file setting has been added that accepts a
6531 list of system calls that shall be logged about (audit).
6533 * The OS image dissection logic (as used by RootImage= in unit files or
6534 systemd-nspawn's --image= switch) has gained support for identifying
6535 and mounting explicit /usr/ partitions, which are now defined in the
6536 discoverable partition specification. This should be useful for
6537 environments where the root file system is
6538 generated/formatted/populated dynamically on first boot and combined
6539 with an immutable /usr/ tree that is supplied by the vendor.
6541 * In the final phase of shutdown, within the systemd-shutdown binary
6542 we'll now try to detach MD devices (i.e software RAID) in addition to
6543 loopback block devices and DM devices as before. This is supposed to
6544 be a safety net only, in order to increase robustness if things go
6545 wrong. Storage subsystems are expected to properly detach their
6546 storage volumes during regular shutdown already (or in case of
6547 storage backing the root file system: in the initrd hook we return to
6550 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
6551 will now log the thread ID in their log output. This is useful when
6552 working with heavily threaded programs.
6554 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
6555 not use the RDRAND CPU instruction. This is useful in environments
6556 such as replay debuggers where non-deterministic behaviour is not
6559 * The autopaging logic in systemd's various tools (such as systemctl)
6560 has been updated to turn on "secure" mode in "less"
6561 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
6562 detected. This disables invoking external programs from the pager,
6563 via the pipe logic. This behaviour may be overridden via the new
6564 $SYSTEMD_PAGERSECURE environment variable.
6566 * Units which have resource limits (.service, .mount, .swap, .slice,
6567 .socket, and .slice) gained new configuration settings
6568 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
6569 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
6570 limits and optional action taken by systemd-oomd.
6572 * A new service systemd-oomd has been added. It monitors resource
6573 contention for selected parts of the unit hierarchy using the PSI
6574 information reported by the kernel, and kills processes when memory
6575 or swap pressure is above configured limits. This service is only
6576 enabled by default in developer mode (see below) and should be
6577 considered a preview in this release. Behaviour details and option
6578 names are subject to change without the usual backwards-compatibility
6581 * A new helper oomctl has been added to introspect systemd-oomd state.
6582 It is only enabled by default in developer mode and should be
6583 considered a preview without the usual backwards-compatibility
6586 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
6587 enabled, systemd reads the system UID boundaries from /etc/login.defs
6588 at runtime, instead of using the built-in values selected during
6589 build. This is an option to improve compatibility for upgrades from
6590 old systems. It's strongly recommended not to make use of this
6591 functionality on new systems (or even enable it during build), as it
6592 makes something runtime-configurable that is mostly an implementation
6593 detail of the OS, and permits avoidable differences in deployments
6594 that create all kinds of problems in the long run.
6596 * New meson option '-Dmode=developer|release' has been added. When
6597 'developer', additional checks and features are enabled that are
6598 relevant during upstream development, e.g. verification that
6599 semi-automatically-generated documentation has been properly updated
6600 following API changes. Those checks are considered hints for
6601 developers and are not actionable in downstream builds. In addition,
6602 extra features that are not ready for general consumption may be
6603 enabled in developer mode. It is thus recommended to set
6604 '-Dmode=release' in end-user and distro builds.
6606 * systemd-cryptsetup gained support for processing detached LUKS
6607 headers specified on the kernel command line via the header=
6608 parameter of the luks.options= kernel command line option. The same
6609 device/path syntax as for key files is supported for header files
6612 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
6613 slot index data for devices that are connected through a PCI bridge
6614 where the _SUN index is associated with the bridge instead of the
6615 network device itself. Previously this would create ambiguous device
6616 naming if multiple network interfaces were connected to the same PCI
6617 bridge. Since this is a naming scheme incompatibility on systems that
6618 possess hardware like this it has been introduced as new naming
6619 scheme "v247". The previous scheme can be selected via the
6620 "net.naming_scheme=v245" kernel command line parameter.
6622 * ConditionFirstBoot= semantics have been modified to be safe towards
6623 abnormal system power-off during first boot. Specifically, the
6624 "systemd-machine-id-commit.service" service now acts as boot
6625 milestone indicating when the first boot process is sufficiently
6626 complete in order to not consider the next following boot also a
6627 first boot. If the system is reset before this unit is reached the
6628 first time, the next boot will still be considered a first boot; once
6629 it has been reached, no further boots will be considered a first
6630 boot. The "first-boot-complete.target" unit now acts as official hook
6631 point to order against this. If a service shall be run on every boot
6632 until the first boot fully succeeds it may thus be ordered before
6633 this target unit (and pull it in) and carry ConditionFirstBoot=
6636 * bootctl's set-default and set-oneshot commands now accept the three
6637 special strings "@default", "@oneshot", "@current" in place of a boot
6638 entry id. These strings are resolved to the current default and
6639 oneshot boot loader entry, as well as the currently booted one. Thus
6640 a command "bootctl set-default @current" may be used to make the
6641 currently boot menu item the new default for all subsequent boots.
6643 * "systemctl edit" has been updated to show the original effective unit
6644 contents in commented form in the text editor.
6646 * Units in user mode are now segregated into three new slices:
6647 session.slice (units that form the core of graphical session),
6648 app.slice ("normal" user applications), and background.slice
6649 (low-priority tasks). Unless otherwise configured, user units are
6650 placed in app.slice. The plan is to add resource limits and
6651 protections for the different slices in the future.
6653 * New GPT partition types for RISCV32/64 for the root and /usr
6654 partitions, and their associated Verity partitions have been defined,
6655 and are now understood by systemd-gpt-auto-generator, and the OS
6656 image dissection logic.
6658 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
6659 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
6660 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
6661 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
6662 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
6663 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
6664 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
6665 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
6666 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
6667 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
6668 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
6669 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
6670 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
6671 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
6672 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
6673 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
6674 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
6675 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
6676 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
6677 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
6678 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
6679 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
6680 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
6681 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
6682 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
6683 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
6684 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
6685 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
6686 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
6687 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
6688 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
6689 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
6690 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
6692 – Warsaw, 2020-11-26
6696 * The service manager gained basic support for cgroup v2 freezer. Units
6697 can now be suspended or resumed either using new systemctl verbs,
6698 freeze and thaw respectively, or via D-Bus.
6700 * PID 1 may now automatically load pre-compiled AppArmor policies from
6701 /etc/apparmor/earlypolicy during early boot.
6703 * The CPUAffinity= setting in service unit files now supports a new
6704 special value "numa" that causes the CPU affinity masked to be set
6705 based on the NUMA mask.
6707 * systemd will now log about all left-over processes remaining in a
6708 unit when the unit is stopped. It will now warn about services using
6709 KillMode=none, as this is generally an unsafe thing to make use of.
6711 * Two new unit file settings
6712 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
6713 added. They may be used to check whether a specific file system path
6714 resides on a block device that is encrypted on the block level
6715 (i.e. using dm-crypt/LUKS).
6717 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
6718 has been added that may be used for simple environment checks. This
6719 is particularly useful when passing in environment variables from a
6720 container manager (or from PAM in case of the systemd --user
6723 * .service unit files now accept a new setting CoredumpFilter= which
6724 allows configuration of the memory sections coredumps of the
6725 service's processes shall include.
6727 * .mount units gained a new ReadWriteOnly= boolean option. If set
6728 it will not be attempted to mount a file system read-only if mounting
6729 in read-write mode doesn't succeed. An option x-systemd.rw-only is
6730 available in /etc/fstab to control the same.
6732 * .socket units gained a new boolean setting PassPacketInfo=. If
6733 enabled, the kernel will attach additional per-packet metadata to all
6734 packets read from the socket, as an ancillary message. This controls
6735 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
6736 depending on socket type.
6738 * .service units gained a new setting RootHash= which may be used to
6739 specify the root hash for verity enabled disk images which are
6740 specified in RootImage=. RootVerity= may be used to specify a path to
6741 the Verity data matching a RootImage= file system. (The latter is
6742 only useful for images that do not contain the Verity data embedded
6743 into the same image that carries a GPT partition table following the
6744 Discoverable Partition Specification). Similarly, systemd-nspawn
6745 gained a new switch --verity-data= that takes a path to a file with
6746 the verity data of the disk image supplied in --image=, if the image
6747 doesn't contain the verity data itself.
6749 * .service units gained a new setting RootHashSignature= which takes
6750 either a base64 encoded PKCS#7 signature of the root hash specified
6751 with RootHash=, or a path to a file to read the signature from. This
6752 allows validation of the root hash against public keys available in
6753 the kernel keyring, and is only supported on recent kernels
6754 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
6755 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
6756 this mechanism has also been added to systemd-veritysetup.
6758 * .service unit files gained two new options
6759 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
6760 tune behaviour if a start or stop timeout is hit, i.e. whether to
6761 terminate the service with SIGTERM, SIGABRT or SIGKILL.
6763 * Most options in systemd that accept hexadecimal values prefixed with
6764 0x in additional to the usual decimal notation now also support octal
6765 notation when the 0o prefix is used and binary notation if the 0b
6768 * Various command line parameters and configuration file settings that
6769 configure key or certificate files now optionally take paths to
6770 AF_UNIX sockets in the file system. If configured that way a stream
6771 connection is made to the socket and the required data read from
6772 it. This is a simple and natural extension to the existing regular
6773 file logic, and permits other software to provide keys or
6774 certificates via simple IPC services, for example when unencrypted
6775 storage on disk is not desired. Specifically, systemd-networkd's
6776 Wireguard and MACSEC key file settings as well as
6777 systemd-journal-gatewayd's and systemd-journal-remote's PEM
6778 key/certificate parameters support this now.
6780 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
6781 configuration files that support specifier expansion learnt six new
6782 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
6783 resolve to the various ID fields from /etc/os-release, %l resolves to
6784 the "short" hostname of the system, i.e. the hostname configured in
6785 the kernel truncated at the first dot.
6787 * Support for the .include syntax in unit files has been removed. The
6788 concept has been obsolete for 6 years and we started warning about
6789 its pending removal 2 years ago (also see NEWS file below). It's
6792 * StandardError= and StandardOutput= in unit files no longer support
6793 the "syslog" and "syslog-console" switches. They were long removed
6794 from the documentation, but will now result in warnings when used,
6795 and be converted to "journal" and "journal+console" automatically.
6797 * If the service setting User= is set to the "nobody" user, a warning
6798 message is now written to the logs (but the value is nonetheless
6799 accepted). Setting User=nobody is unsafe, since the primary purpose
6800 of the "nobody" user is to own all files whose owner cannot be mapped
6801 locally. It's in particular used by the NFS subsystem and in user
6802 namespacing. By running a service under this user's UID it might get
6803 read and even write access to all these otherwise unmappable files,
6804 which is quite likely a major security problem.
6806 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
6807 and others) now have a size and inode limits applied (50% of RAM for
6808 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
6809 that the implicit kernel default is 50% too, so there is no change
6810 in the size limit for /tmp and /dev/shm.
6812 * nss-mymachines lost support for resolution of users and groups, and
6813 now only does resolution of hostnames. This functionality is now
6814 provided by nss-systemd. Thus, the 'mymachines' entry should be
6815 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
6816 (and 'systemd' added if it is not already there).
6818 * A new kernel command line option systemd.hostname= has been added
6819 that allows controlling the hostname that is initialized early during
6822 * A kernel command line option "udev.blockdev_read_only" has been
6823 added. If specified all hardware block devices that show up are
6824 immediately marked as read-only by udev. This option is useful for
6825 making sure that a specific boot under no circumstances modifies data
6826 on disk. Use "blockdev --setrw" to undo the effect of this, per
6829 * A new boolean kernel command line option systemd.swap= has been
6830 added, which may be used to turn off automatic activation of swap
6831 devices listed in /etc/fstab.
6833 * New kernel command line options systemd.condition_needs_update= and
6834 systemd.condition_first_boot= have been added, which override the
6835 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
6838 * A new kernel command line option systemd.clock_usec= has been added
6839 that allows setting the system clock to the specified time in µs
6840 since Jan 1st, 1970 early during boot. This is in particular useful
6841 in order to make test cases more reliable.
6843 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
6844 systemd-coredump to save core files for suid processes. When saving
6845 the core file, systemd-coredump will use the effective uid and gid of
6846 the process that faulted.
6848 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
6849 now automatically set to "Y" at boot, in order to enable pstore
6850 generation for collection with systemd-pstore.
6852 * We provide a set of udev rules to enable auto-suspend on PCI and USB
6853 devices that were tested to correctly support it. Previously, this
6854 was distributed as a set of udev rules, but has now been replaced by
6855 by a set of hwdb entries (and a much shorter udev rule to take action
6856 if the device modalias matches one of the new hwdb entries).
6858 As before, entries are periodically imported from the database
6859 maintained by the ChromiumOS project. If you have a device that
6860 supports auto-suspend correctly and where it should be enabled by
6861 default, please submit a patch that adds it to the database (see
6862 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
6864 * systemd-udevd gained the new configuration option timeout_signal= as well
6865 as a corresponding kernel command line option udev.timeout_signal=.
6866 The option can be used to configure the UNIX signal that the main
6867 daemon sends to the worker processes on timeout. Setting the signal
6868 to SIGABRT is useful for debugging.
6870 * .link files managed by systemd-udevd gained options RxFlowControl=,
6871 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
6872 order to configure various flow control parameters. They also gained
6873 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
6874 frame ring buffer sizes.
6876 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
6877 enabled systemd-networkd manages all routes configured by other tools.
6879 * .network files managed by systemd-networkd gained a new section
6880 [SR-IOV], in order to configure SR-IOV capable network devices.
6882 * systemd-networkd's [IPv6Prefix] section in .network files gained a
6883 new boolean setting Assign=. If enabled an address from the prefix is
6884 automatically assigned to the interface.
6886 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
6887 controls delegated prefixes assigned by DHCPv6 client. The section
6888 has three settings: SubnetID=, Assign=, and Token=. The setting
6889 SubnetID= allows explicit configuration of the preferred subnet that
6890 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
6891 Assign= is enabled (which is the default) an address from any acquired
6892 delegated prefix is automatically chosen and assigned to the
6893 interface. The setting Token= specifies an optional address generation
6896 * systemd-networkd's [Network] section gained a new setting
6897 IPv4AcceptLocal=. If enabled the interface accepts packets with local
6900 * systemd-networkd gained support for configuring the HTB queuing
6901 discipline in the [HierarchyTokenBucket] and
6902 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
6903 be configured in the [PFIFO] section, "GRED" in
6904 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
6905 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
6906 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
6907 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
6908 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
6909 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
6911 * systemd-networkd gained support for a new Termination= setting in the
6912 [CAN] section for configuring the termination resistor. It also
6913 gained a new ListenOnly= setting for controlling whether to only
6914 listen on CAN interfaces, without interfering with traffic otherwise
6915 (which is useful for debugging/monitoring CAN network
6916 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
6917 been added to configure various CAN-FD aspects.
6919 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
6920 When enabled, DHCPv6 will be attempted right-away without requiring an
6921 Router Advertisement packet suggesting it first (i.e. without the 'M'
6922 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
6923 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
6924 the RA packets suggest it.
6926 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
6927 which may be used to turn off use of the gateway information provided
6928 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
6929 used to configure how to process leases that lack a lifetime option.
6931 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
6932 setting SendVendorOption= allowing configuration of additional vendor
6933 options to send in the DHCP requests/responses. The [DHCPv6] section
6934 gained a new SendOption= setting for sending arbitrary DHCP
6935 options. RequestOptions= has been added to request arbitrary options
6936 from the server. UserClass= has been added to set the DHCP user class
6939 * systemd-networkd's [DHCPServer] section gained a new set of options
6940 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
6941 information about these three protocols in the DHCP lease. It also
6942 gained support for including "MUD" URLs ("Manufacturer Usage
6943 Description"). Support for "MUD" URLs was also added to the LLDP
6944 stack, configurable in the [LLDP] section in .network files.
6946 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
6947 mode. Also, the sections now support a new setting SourceMACAddress=.
6949 * systemd-networkd's .netdev files now support a new setting
6950 VLANProtocol= in the [Bridge] section that allows configuration of
6951 the VLAN protocol to use.
6953 * systemd-networkd supports a new Group= setting in the [Link] section
6954 of the .network files, to control the link group.
6956 * systemd-networkd's [Network] section gained a new
6957 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
6958 link local address is generated.
6960 * A new default .network file is now shipped that matches TUN/TAP
6961 devices that begin with "vt-" in their name. Such interfaces will
6962 have IP routing onto the host links set up automatically. This is
6963 supposed to be used by VM managers to trivially acquire a network
6964 interface which is fully set up for host communication, simply by
6965 carefully picking an interface name to use.
6967 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
6968 which sets the route priority for routes specified by the DHCP server.
6970 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
6971 which configures the vendor class information sent to DHCP server.
6973 * The BlackList= settings in .network files' [DHCPv4] and
6974 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
6975 are still understood to provide compatibility.
6977 * networkctl gained the new "forcerenew" command for forcing all DHCP
6978 server clients to renew their lease. The interface "status" output
6979 will now show numerous additional fields of information about an
6980 interface. There are new "up" and "down" commands to bring specific
6981 interfaces up or down.
6983 * systemd-resolved's DNS= configuration option now optionally accepts a
6984 port number (after ":") and a host name (after "#"). When the host
6985 name is specified, the DNS-over-TLS certificate is validated to match
6986 the specified hostname. Additionally, in case of IPv6 addresses, an
6987 interface may be specified (after "%").
6989 * systemd-resolved may be configured to forward single-label DNS names.
6990 This is not standard-conformant, but may make sense in setups where
6991 public DNS servers are not used.
6993 * systemd-resolved's DNS-over-TLS support gained SNI validation.
6995 * systemd-nspawn's --resolv-conf= switch gained a number of new
6996 supported values. Specifically, options starting with "replace-" are
6997 like those prefixed "copy-" but replace any existing resolv.conf
6998 file. And options ending in "-uplink" and "-stub" can now be used to
6999 propagate other flavours of resolv.conf into the container (as
7000 defined by systemd-resolved).
7002 * The various programs included in systemd can now optionally output
7003 their log messages on stderr prefixed with a timestamp, controlled by
7004 the $SYSTEMD_LOG_TIME environment variable.
7006 * systemctl gained a new "-P" switch that is a shortcut for "--value
7009 * "systemctl list-units" and "systemctl list-machines" no longer hide
7010 their first output column with --no-legend. To hide the first column,
7013 * "systemctl reboot" takes the option "--reboot-argument=".
7014 The optional positional argument to "systemctl reboot" is now
7015 being deprecated in favor of this option.
7017 * systemd-run gained a new switch --slice-inherit. If specified the
7018 unit it generates is placed in the same slice as the systemd-run
7021 * systemd-journald gained support for zstd compression of large fields
7022 in journal files. The hash tables in journal files have been hardened
7023 against hash collisions. This is an incompatible change and means
7024 that journal files created with new systemd versions are not readable
7025 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
7026 environment variable for systemd-journald.service is set to 0 this
7027 new hardening functionality may be turned off, so that generated
7028 journal files remain compatible with older journalctl
7031 * journalctl will now include a clickable link in the default output for
7032 each log message for which a URL with further documentation is
7033 known. This is only supported on terminal emulators that support
7034 clickable hyperlinks, and is turned off if a pager is used (since
7035 "less" still doesn't support hyperlinks,
7036 unfortunately). Documentation URLs may be included in log messages
7037 either by including a DOCUMENTATION= journal field in it, or by
7038 associating a journal message catalog entry with the log message's
7039 MESSAGE_ID, which then carries a "Documentation:" tag.
7041 * journald.conf gained a new boolean setting Audit= that may be used to
7042 control whether systemd-journald will enable audit during
7045 * when systemd-journald's log stream is broken up into multiple lines
7046 because the PID of the sender changed this is indicated in the
7047 generated log records via the _LINE_BREAK=pid-change field.
7049 * journalctl's "-o cat" output mode will now show one or more journal
7050 fields specified with --output-fields= instead of unconditionally
7051 MESSAGE=. This is useful to retrieve a very specific set of fields
7052 without any decoration.
7054 * The sd-journal.h API gained two new functions:
7055 sd_journal_enumerate_available_unique() and
7056 sd_journal_enumerate_available_data() that operate like their
7057 counterparts that lack the _available_ in the name, but skip items
7058 that cannot be read and processed by the local implementation
7059 (i.e. are compressed in an unsupported format or such),
7061 * coredumpctl gained a new --file= switch, matching the same one in
7062 journalctl: a specific journal file may be specified to read the
7065 * coredumps collected by systemd-coredump may now be compressed using
7068 * systemd-binfmt gained a new switch --unregister for unregistering all
7069 registered entries at once. This is now invoked automatically at
7070 shutdown, so that binary formats registered with the "F" flag will
7071 not block clean file system unmounting.
7073 * systemd-notify's --pid= switch gained new values: "parent", "self",
7074 "auto" for controlling which PID to send to the service manager: the
7075 systemd-notify process' PID, or the one of the process invoking it.
7077 * systemd-logind's Session bus object learnt a new method call
7078 SetType() for temporarily updating the session type of an already
7079 allocated session. This is useful for upgrading tty sessions to
7080 graphical ones once a compositor is invoked.
7082 * systemd-socket-proxy gained a new switch --exit-idle-time= for
7083 configuring an exit-on-idle time.
7085 * systemd-repart's --empty= setting gained a new value "create". If
7086 specified a new empty regular disk image file is created under the
7087 specified name. Its size may be specified with the new --size=
7088 option. The latter is also supported without the "create" mode, in
7089 order to grow existing disk image files to the specified size. These
7090 two new options are useful when creating or manipulating disk images
7091 instead of operating on actual block devices.
7093 * systemd-repart drop-ins now support a new UUID= setting to control
7094 the UUID to assign to a newly created partition.
7096 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
7099 * systemd-repart's Label= setting now support the usual, simple
7100 specifier expansion.
7102 * systemd-homed's LUKS backend gained the ability to discard empty file
7103 system blocks automatically when the user logs out. This is enabled
7104 by default to ensure that home directories take minimal space when
7105 logged out but get full size guarantees when logged in. This may be
7106 controlled with the new --luks-offline-discard= switch to homectl.
7108 * If systemd-homed detects that /home/ is encrypted as a whole it will
7109 now default to the directory or subvolume backends instead of the
7110 LUKS backend, in order to avoid double encryption. The default
7111 storage and file system may now be configured explicitly, too, via
7112 the new /etc/systemd/homed.conf configuration file.
7114 * systemd-homed now supports unlocking home directories with FIDO2
7115 security tokens that support the 'hmac-secret' extension, in addition
7116 to the existing support for PKCS#11 security token unlocking
7117 support. Note that many recent hardware security tokens support both
7118 interfaces. The FIDO2 support is accessible via homectl's
7119 --fido2-device= option.
7121 * homectl's --pkcs11-uri= setting now accepts two special parameters:
7122 if "auto" is specified and only one suitable PKCS#11 security token
7123 is plugged in, its URL is automatically determined and enrolled for
7124 unlocking the home directory. If "list" is specified a brief table of
7125 suitable PKCS#11 security tokens is shown. Similar, the new
7126 --fido2-device= option also supports these two special values, for
7127 automatically selecting and listing suitable FIDO2 devices.
7129 * The /etc/crypttab tmp option now optionally takes an argument
7130 selecting the file system to use. Moreover, the default is now
7131 changed from ext2 to ext4.
7133 * There's a new /etc/crypttab option "keyfile-erase". If specified the
7134 key file listed in the same line is removed after use, regardless if
7135 volume activation was successful or not. This is useful if the key
7136 file is only acquired transiently at runtime and shall be erased
7137 before the system continues to boot.
7139 * There's also a new /etc/crypttab option "try-empty-password". If
7140 specified, before asking the user for a password it is attempted to
7141 unlock the volume with an empty password. This is useful for
7142 installing encrypted images whose password shall be set on first boot
7143 instead of at installation time.
7145 * systemd-cryptsetup will now attempt to load the keys to unlock
7146 volumes with automatically from files in
7147 /etc/cryptsetup-keys.d/<volume>.key and
7148 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
7150 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
7151 /etc/crypttab, during boot.
7153 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
7154 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
7157 * A new generator systemd-xdg-autostart-generator has been added. It
7158 generates systemd unit files from XDG autostart .desktop files, and
7159 may be used to let the systemd user instance manage services that are
7160 started automatically as part of the desktop session.
7162 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
7163 to query and change the firmware's 'Reboot Into Firmware Interface'
7166 * systemd-firstboot gained a new switch --kernel-command-line= that may
7167 be used to initialize the /etc/kernel/cmdline file of the image. It
7168 also gained a new switch --root-password-hashed= which is like
7169 --root-password= but accepts a pre-hashed UNIX password as
7170 argument. The new option --delete-root-password may be used to unset
7171 any password for the root user (dangerous!). The --root-shell= switch
7172 may be used to control the shell to use for the root account. A new
7173 --force option may be used to override any already set settings with
7174 the parameters specified on the command line (by default, the tool
7175 will not override what has already been set before, i.e. is purely
7178 * systemd-firstboot gained support for a new --image= switch, which is
7179 similar to --root= but accepts the path to a disk image file, on
7180 which it then operates.
7182 * A new sd-path.h API has been added to libsystemd. It provides a
7183 simple API for retrieving various search paths and primary
7184 directories for various resources.
7186 * A new call sd_notify_barrier() has been added to the sd-daemon.h
7187 API. The call will block until all previously sent sd_notify()
7188 messages have been processed by the service manager. This is useful
7189 to remove races caused by a process already having disappeared at the
7190 time a notification message is processed by the service manager,
7191 making correct attribution impossible. The systemd-notify tool will
7192 now make use of this call implicitly, but this can be turned off again
7193 via the new --no-block switch.
7195 * When sending a file descriptor (fd) to the service manager to keep
7196 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
7197 may be specified. If passed the service manager will refrain from
7198 poll()ing on the file descriptor. Traditionally (and when the
7199 parameter is not specified), the service manager will poll it for
7200 POLLHUP or POLLERR events, and immediately close the fds in that
7203 * The service manager (PID1) gained a new D-Bus method call
7204 SetShowStatus() which may be used to control whether it shall show
7205 boot-time status output on the console. This method has a similar
7206 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
7208 * The sd-bus API gained a number of convenience functions that take
7209 va_list arguments rather than "...". For example, there's now
7210 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
7211 it easier to build wrappers that accept variadic arguments and want
7212 to pass a ready va_list structure to sd-bus.
7214 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
7215 flag which alters how the userdata pointer to pass to the callbacks
7216 is determined. When the flag is set, the offset field is converted
7217 as-is into a pointer, without adding it to the object pointer the
7218 vtable is associated with.
7220 * sd-bus now exposes four new functions:
7221 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
7222 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
7223 validate strings to check if they qualify as various D-Bus concepts.
7225 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
7226 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
7227 that simplify adding argument names to D-Bus methods and signals.
7229 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
7231 * Various D-Bus APIs of systemd daemons now have man pages that
7232 document the methods, signals and properties.
7234 * The expectations on user/group name syntax are now documented in
7235 detail; documentation on how classic home directories may be
7236 converted into home directories managed by homed has been added;
7237 documentation regarding integration of homed/userdb functionality in
7238 desktops has been added:
7240 https://systemd.io/USER_NAMES
7241 https://systemd.io/CONVERTING_TO_HOMED
7242 https://systemd.io/USERDB_AND_DESKTOPS
7244 * Documentation for the on-disk Journal file format has been updated
7245 and has now moved to:
7247 https://systemd.io/JOURNAL_FILE_FORMAT
7249 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
7250 has been extended by a set of environment variables that expose
7251 select fields from the host's os-release file to the container
7252 payload. Similarly, host's os-release files can be mounted into the
7253 container underneath /run/host. Together, those mechanisms provide a
7254 standardized way to expose information about the host to the
7255 container payload. Both interfaces are implemented in systemd-nspawn.
7257 * All D-Bus services shipped in systemd now implement the generic
7258 LogControl1 D-Bus API which allows clients to change log level +
7259 target of the service during runtime.
7261 * Only relevant for developers: the mkosi.default symlink has been
7262 dropped from version control. Please create a symlink to one of the
7263 distribution-specific defaults in .mkosi/ based on your preference.
7265 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
7266 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
7267 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
7268 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
7269 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
7270 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
7271 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
7272 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
7273 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
7274 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
7275 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
7276 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
7277 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
7278 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
7279 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
7280 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
7281 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
7282 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
7283 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
7284 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
7285 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
7286 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
7287 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
7288 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
7289 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
7290 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
7291 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
7292 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
7293 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
7294 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
7295 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
7296 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
7297 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
7298 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
7299 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
7300 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
7301 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
7302 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
7304 – Warsaw, 2020-07-30
7308 * A new tool "systemd-repart" has been added, that operates as an
7309 idempotent declarative repartitioner for GPT partition tables.
7310 Specifically, a set of partitions that must or may exist can be
7311 configured via drop-in files, and during every boot the partition
7312 table on disk is compared with these files, creating missing
7313 partitions or growing existing ones based on configurable relative
7314 and absolute size constraints. The tool is strictly incremental,
7315 i.e. does not delete, shrink or move partitions, but only adds and
7316 grows them. The primary use-case is OS images that ship in minimized
7317 form, that on first boot are grown to the size of the underlying
7318 block device or augmented with additional partitions. For example,
7319 the root partition could be extended to cover the whole disk, or a
7320 swap or /home partitions could be added on first boot. It can also be
7321 used for systems that use an A/B update scheme but ship images with
7322 just the A partition, with B added on first boot. The tool is
7323 primarily intended to be run in the initrd, shortly before
7324 transitioning into the host OS, but can also be run after the
7325 transition took place. It automatically discovers the disk backing
7326 the root file system, and should hence not require any additional
7327 configuration besides the partition definition drop-ins. If no
7328 configuration drop-ins are present, no action is taken.
7330 * A new component "userdb" has been added, along with a small daemon
7331 "systemd-userdbd.service" and a client tool "userdbctl". The framework
7332 allows defining rich user and group records in a JSON format,
7333 extending on the classic "struct passwd" and "struct group"
7334 structures. Various components in systemd have been updated to
7335 process records in this format, including systemd-logind and
7336 pam-systemd. The user records are intended to be extensible, and
7337 allow setting various resource management, security and runtime
7338 parameters that shall be applied to processes and sessions of the
7339 user as they log in. This facility is intended to allow associating
7340 such metadata directly with user/group records so that they can be
7341 produced, extended and consumed in unified form. We hope that
7342 eventually frameworks such as sssd will generate records this way, so
7343 that for the first time resource management and various other
7344 per-user settings can be configured in LDAP directories and then
7345 provided to systemd (specifically to systemd-logind and pam-system)
7346 to apply on login. For further details see:
7348 https://systemd.io/USER_RECORD
7349 https://systemd.io/GROUP_RECORD
7350 https://systemd.io/USER_GROUP_API
7352 * A small new service systemd-homed.service has been added, that may be
7353 used to securely manage home directories with built-in encryption.
7354 The complete user record data is unified with the home directory,
7355 thus making home directories naturally migratable. Its primary
7356 back-end is based on LUKS volumes, but fscrypt, plain directories,
7357 and other storage schemes are also supported. This solves a couple of
7358 problems we saw with traditional ways to manage home directories, in
7359 particular when it comes to encryption. For further discussion of
7360 this, see the video of Lennart's talk at AllSystemsGo! 2019:
7362 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
7364 For further details about the format and expectations on home
7365 directories this new daemon makes, see:
7367 https://systemd.io/HOME_DIRECTORY
7369 * systemd-journald is now multi-instantiable. In addition to the main
7370 instance systemd-journald.service there's now a template unit
7371 systemd-journald@.service, with each instance defining a new named
7372 log 'namespace' (whose name is specified via the instance part of the
7373 unit name). A new unit file setting LogNamespace= has been added,
7374 taking such a namespace name, that assigns services to the specified
7375 log namespaces. As each log namespace is serviced by its own
7376 independent journal daemon, this functionality may be used to improve
7377 performance and increase isolation of applications, at the price of
7378 losing global message ordering. Each instance of journald has a
7379 separate set of configuration files, with possibly different disk
7380 usage limitations and other settings.
7382 journalctl now takes a new option --namespace= to show logs from a
7383 specific log namespace. The sd-journal.h API gained
7384 sd_journal_open_namespace() for opening the log stream of a specific
7385 log namespace. systemd-journald also gained the ability to exit on
7386 idle, which is useful in the context of log namespaces, as this means
7387 log daemons for log namespaces can be activated automatically on
7388 demand and will stop automatically when no longer used, minimizing
7391 * When systemd-tmpfiles copies a file tree using the 'C' line type it
7392 will now label every copied file according to the SELinux database.
7394 * When systemd/PID 1 detects it is used in the initrd it will now boot
7395 into initrd.target rather than default.target by default. This should
7396 make it simpler to build initrds with systemd as for many cases the
7397 only difference between a host OS image and an initrd image now is
7398 the presence of the /etc/initrd-release file.
7400 * A new kernel command line option systemd.cpu_affinity= is now
7401 understood. It's equivalent to the CPUAffinity= option in
7402 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
7403 itself and the default for all other processes.
7405 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
7406 equivalent), the SELinux database is now reloaded, ensuring that
7407 sockets and other file system objects are generated taking the new
7408 database into account.
7410 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
7411 "quiet" has been changed to imply that instead of
7412 "systemd.show-status=auto". In this mode, only messages about errors
7413 and significant delays in boot are shown on the console.
7415 * The sd-event.h API gained native support for the new Linux "pidfd"
7416 concept. This permits watching processes using file descriptors
7417 instead of PID numbers, which fixes a number of races and makes
7418 process supervision more robust and efficient. All of systemd's
7419 components will now use pidfds if the kernel supports it for process
7420 watching, with the exception of PID 1 itself, unfortunately. We hope
7421 to move PID 1 to exclusively using pidfds too eventually, but this
7422 requires some more kernel work first. (Background: PID 1 watches
7423 processes using waitid() with the P_ALL flag, and that does not play
7424 together nicely with pidfds yet.)
7426 * Closely related to this, the sd-event.h API gained two new calls
7427 sd_event_source_send_child_signal() (for sending a signal to a
7428 watched process) and sd_event_source_get_child_process_own() (for
7429 marking a process so that it is killed automatically whenever the
7430 event source watching it is freed).
7432 * systemd-networkd gained support for configuring Token Bucket Filter
7433 (TBF) parameters in its qdisc configuration support. Similarly,
7434 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
7435 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
7437 * systemd-networkd gained support for Intermediate Functional Block
7438 (IFB) network devices.
7440 * systemd-networkd gained support for configuring multi-path IP routes,
7441 using the new MultiPathRoute= setting in the [Route] section.
7443 * systemd-networkd's DHCPv4 client has been updated to support a new
7444 SendDecline= option. If enabled, duplicate address detection is done
7445 after a DHCP offer is received from the server. If a conflict is
7446 detected, the address is declined. The DHCPv4 client also gained
7447 support for a new RouteMTUBytes= setting that allows to configure the
7448 MTU size to be used for routes generated from DHCPv4 leases.
7450 * The PrefixRoute= setting in systemd-networkd's [Address] section of
7451 .network files has been deprecated, and replaced by AddPrefixRoute=,
7452 with its sense inverted.
7454 * The Gateway= setting of [Route] sections of .network files gained
7455 support for a special new value "_dhcp". If set, the configured
7456 static route uses the gateway host configured via DHCP.
7458 * New User= and SuppressPrefixLength= settings have been implemented
7459 for the [RoutingPolicyRule] section of .network files to configure
7460 source routing based on UID ranges and prefix length, respectively.
7462 * The Type= match property of .link files has been generalized to
7463 always match the device type shown by 'networkctl status', even for
7464 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
7467 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
7468 D-Bus message object as "sensitive". Those objects are erased from
7469 memory when they are freed. This concept is intended to be used for
7470 messages that contain security sensitive data. A new flag
7471 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
7472 in sd-bus vtables, causing any incoming and outgoing messages of
7473 those methods to be implicitly marked as "sensitive".
7475 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
7476 contents of a message (or parts thereof) to standard output for
7479 * systemd-sysusers gained support for creating users with the primary
7480 group named differently than the user.
7482 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
7483 gained support for growing XFS partitions. Previously it supported
7484 only ext4 and btrfs partitions.
7486 * The support for /etc/crypttab gained a new x-initrd.attach option. If
7487 set, the specified encrypted volume is unlocked already in the
7488 initrd. This concept corresponds to the x-initrd.mount option in
7491 * systemd-cryptsetup gained native support for unlocking encrypted
7492 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
7493 encryption of volumes to YubiKeys. This is exposed in the new
7494 pkcs11-uri= option in /etc/crypttab.
7496 * The /etc/fstab support in systemd now supports two new mount options
7497 x-systemd.{required,wanted}-by=, for explicitly configuring the units
7498 that the specified mount shall be pulled in by, in place of
7499 the usual local-fs.target/remote-fs.target.
7501 * The https://systemd.io/ web site has been relaunched, directly
7502 populated with most of the documentation included in the systemd
7503 repository. systemd also acquired a new logo, thanks to Tobias
7506 * systemd-udevd gained support for managing "alternative" network
7507 interface names, as supported by new Linux kernels. For the first
7508 time this permits assigning multiple (and longer!) names to a network
7509 interface. systemd-udevd will now by default assign the names
7510 generated via all supported naming schemes to each interface. This
7511 may be further tweaked with .link files and the AlternativeName= and
7512 AlternativeNamesPolicy= settings. Other components of systemd have
7513 been updated to support the new alternative names wherever
7514 appropriate. For example, systemd-nspawn will now generate
7515 alternative interface names for the host-facing side of container
7516 veth links based on the full container name without truncation.
7518 * systemd-nspawn interface naming logic has been updated in another way
7519 too: if the main interface name (i.e. as opposed to new-style
7520 "alternative" names) based on the container name is truncated, a
7521 simple hashing scheme is used to give different interface names to
7522 multiple containers whose names all begin with the same prefix. Since
7523 this changes the primary interface names pointing to containers if
7524 truncation happens, the old scheme may still be requested by
7525 selecting an older naming scheme, via the net.naming_scheme= kernel
7526 command line option.
7528 * PrivateUsers= in service files now works in services run by the
7529 systemd --user per-user instance of the service manager.
7531 * A new per-service sandboxing option ProtectClock= has been added that
7532 locks down write access to the system clock. It takes away device
7533 node access to /dev/rtc as well as the system calls that set the
7534 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
7535 Note that this option does not affect access to auxiliary services
7536 that allow changing the clock, for example access to
7539 * The systemd-id128 tool gained a new "show" verb for listing or
7540 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
7541 GPT partition table types.
7543 * The Discoverable Partitions Specification has been updated to support
7544 /var and /var/tmp partition discovery. Support for this has been
7545 added to systemd-gpt-auto-generator. For details see:
7547 https://systemd.io/DISCOVERABLE_PARTITIONS
7549 * "systemctl list-unit-files" has been updated to show a new column
7550 with the suggested enablement state based on the vendor preset files
7551 for the respective units.
7553 * "systemctl" gained a new option "--with-dependencies". If specified
7554 commands such as "systemctl status" or "systemctl cat" will now show
7555 all specified units along with all units they depend on.
7557 * networkctl gained support for showing per-interface logs in its
7560 * systemd-networkd-wait-online gained support for specifying the maximum
7561 operational state to wait for, and to wait for interfaces to
7564 * The [Match] section of .link and .network files now supports a new
7565 option PermanentMACAddress= which may be used to check against the
7566 permanent MAC address of a network device even if a randomized MAC
7569 * The [TrafficControlQueueingDiscipline] section in .network files has
7570 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
7571 dropped from the individual setting names.
7573 * Any .link and .network files that have an empty [Match] section (this
7574 also includes empty and commented-out files) will now be
7575 rejected. systemd-udev and systemd-networkd started warning about
7576 such files in version 243.
7578 * systemd-logind will now validate access to the operation of changing
7579 the virtual terminal via a polkit action. By default, only users
7580 with at least one session on a local VT are granted permission.
7582 * When systemd sets up PAM sessions that invoked service processes
7583 shall run in, the pam_setcred() API is now invoked, thus permitting
7584 PAM modules to set additional credentials for the processes.
7586 * portablectl attach/detach verbs now accept --now and --enable options
7587 to combine attachment with enablement and invocation, or detachment
7588 with stopping and disablement.
7590 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
7591 fixed, which in turn exposed bugs in unit configuration of services
7592 which have Type=oneshot and should only run once, but do not have
7593 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
7594 service may be started again after exiting successfully, for example
7595 as a dependency in another transaction. Affected services included
7596 some internal systemd services (most notably
7597 systemd-vconsole-setup.service, which was updated to have
7598 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
7599 plymouth has been suitably updated or patched before upgrading to
7600 this systemd release. See
7601 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
7602 additional discussion.
7604 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
7605 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
7606 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
7607 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
7608 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
7609 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
7610 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
7611 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
7612 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
7613 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
7614 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
7615 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
7616 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
7617 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
7618 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
7619 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
7620 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
7621 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
7622 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
7623 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
7624 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
7625 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
7626 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
7627 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
7630 – Warsaw, 2020-03-06
7634 * Support for the cpuset cgroups v2 controller has been added.
7635 Processes may be restricted to specific CPUs using the new
7636 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
7637 AllowedMemoryNodes= setting.
7639 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
7640 now be configured using a new RestartKillSignal= setting. This
7641 allows units which signals to request termination to implement
7642 different behaviour when stopping in preparation for a restart.
7644 * "systemctl clean" may now be used also for socket, mount, and swap
7647 * systemd will also read configuration options from the EFI variable
7648 SystemdOptions. This may be used to configure systemd behaviour when
7649 modifying the kernel command line is inconvenient, but configuration
7650 on disk is read too late, for example for the options related to
7651 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
7652 set the EFI variable.
7654 * systemd will now disable printk ratelimits in early boot. This should
7655 allow us to capture more logs from the early boot phase where normal
7656 storage is not available and the kernel ring buffer is used for
7657 logging. Configuration on the kernel command line has higher priority
7658 and overrides the systemd setting.
7660 systemd programs which log to /dev/kmsg directly use internal
7661 ratelimits to prevent runaway logging. (Normally this is only used
7662 during early boot, so in practice this change has very little
7665 * Unit files now support top level dropin directories of the form
7666 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
7667 that affects all corresponding unit files.
7669 * systemctl gained support for 'stop --job-mode=triggering' which will
7670 stop the specified unit and any units which could trigger it.
7672 * Unit status display now includes units triggering and triggered by
7673 the unit being shown.
7675 * The RuntimeMaxSec= setting is now supported by scopes, not just
7676 .service units. This is particularly useful for PAM sessions which
7677 create a scope unit for the user login. systemd.runtime_max_sec=
7678 setting may used with the pam_systemd module to limit the duration
7679 of the PAM session, for example for time-limited logins.
7681 * A new @pkey system call group is now defined to make it easier to
7682 allow-list memory protection syscalls for containers and services
7683 which need to use them.
7685 * systemd-udevd: removed the 30s timeout for killing stale workers on
7686 exit. systemd-udevd now waits for workers to finish. The hard-coded
7687 exit timeout of 30s was too short for some large installations, where
7688 driver initialization could be prematurely interrupted during initrd
7689 processing if the root file system had been mounted and init was
7690 preparing to switch root. If udevd is run without systemd and workers
7691 are hanging while udevd receives an exit signal, udevd will now exit
7692 when udev.event_timeout is reached for the last hanging worker. With
7693 systemd, the exit timeout can additionally be configured using
7694 TimeoutStopSec= in systemd-udevd.service.
7696 * udev now provides a program (fido_id) that identifies FIDO CTAP1
7697 ("U2F")/CTAP2 security tokens based on the usage declared in their
7698 report and descriptor and outputs suitable environment variables.
7699 This replaces the externally maintained allow lists of all known
7700 security tokens that were used previously.
7702 * Automatically generated autosuspend udev rules for allow-listed
7703 devices have been imported from the Chromium OS project. This should
7704 improve power saving with many more devices.
7706 * udev gained a new "CONST{key}=value" setting that allows matching
7707 against system-wide constants without forking a helper binary.
7708 Currently "arch" and "virt" keys are supported.
7710 * udev now opens CDROMs in non-exclusive mode when querying their
7711 capabilities. This should fix issues where other programs trying to
7712 use the CDROM cannot gain access to it, but carries a risk of
7713 interfering with programs writing to the disk, if they did not open
7714 the device in exclusive mode as they should.
7716 * systemd-networkd does not create a default route for IPv4 link local
7717 addressing anymore. The creation of the route was unexpected and was
7718 breaking routing in various cases, but people who rely on it being
7719 created implicitly will need to adjust. Such a route may be requested
7720 with DefaultRouteOnDevice=yes.
7722 Similarly, systemd-networkd will not assign a link-local IPv6 address
7723 when IPv6 link-local routing is not enabled.
7725 * Receive and transmit buffers may now be configured on links with
7726 the new RxBufferSize= and TxBufferSize= settings.
7728 * systemd-networkd may now advertise additional IPv6 routes. A new
7729 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
7732 * systemd-networkd may now configure "next hop" routes using the
7733 [NextHop] section and Gateway= and Id= settings.
7735 * systemd-networkd will now retain DHCP config on restarts by default
7736 (but this may be overridden using the KeepConfiguration= setting).
7737 The default for SendRelease= has been changed to true.
7739 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
7740 received from the server.
7742 The client will use the received SIP server list if UseSIP=yes is
7745 The client may be configured to request specific options from the
7746 server using a new RequestOptions= setting.
7748 The client may be configured to send arbitrary options to the server
7749 using a new SendOption= setting.
7751 A new IPServiceType= setting has been added to configure the "IP
7752 service type" value used by the client.
7754 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
7755 request prefix hints in the DHCPv6 solicitation.
7757 * The DHCPv4 server may be configured to send arbitrary options using
7758 a new SendOption= setting.
7760 * The DHCPv4 server may now be configured to emit SIP server list using
7761 the new EmitSIP= and SIP= settings.
7763 * systemd-networkd and networkctl may now renew DHCP leases on demand.
7764 networkctl has a new 'networkctl renew' verb.
7766 * systemd-networkd may now reconfigure links on demand. networkctl
7767 gained two new verbs: "reload" will reload the configuration, and
7768 "reconfigure DEVICE…" will reconfigure one or more devices.
7770 * .network files may now match on SSID and BSSID of a wireless network,
7771 i.e. the access point name and hardware address using the new SSID=
7772 and BSSID= options. networkctl will display the current SSID and
7773 BSSID for wireless links.
7775 .network files may also match on the wireless network type using the
7776 new WLANInterfaceType= option.
7778 * systemd-networkd now includes default configuration that enables
7779 link-local addressing when connected to an ad-hoc wireless network.
7781 * systemd-networkd may configure the Traffic Control queueing
7782 disciplines in the kernel using the new
7783 [TrafficControlQueueingDiscipline] section and Parent=,
7784 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
7785 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
7786 NetworkEmulatorDuplicateRate= settings.
7788 * systemd-tmpfiles gained a new w+ setting to append to files.
7790 * systemd-analyze dump will now report when the memory configuration in
7791 the kernel does not match what systemd has configured (usually,
7792 because some external program has modified the kernel configuration
7795 * systemd-analyze gained a new --base-time= switch instructs the
7796 'calendar' verb to resolve times relative to that timestamp instead
7797 of the present time.
7799 * journalctl --update-catalog now produces deterministic output (making
7800 reproducible image builds easier).
7802 * A new devicetree-overlay setting is now documented in the Boot Loader
7805 * The default value of the WatchdogSec= setting used in systemd
7806 services (the ones bundled with the project itself) may be set at
7807 configuration time using the -Dservice-watchdog= setting. If set to
7808 empty, the watchdogs will be disabled.
7810 * systemd-resolved validates IP addresses in certificates now when GnuTLS
7813 * libcryptsetup >= 2.0.1 is now required.
7815 * A configuration option -Duser-path= may be used to override the $PATH
7816 used by the user service manager. The default is again to use the same
7817 path as the system manager.
7819 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
7820 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
7823 * Service units gained a new sandboxing option ProtectKernelLogs= which
7824 makes sure the program cannot get direct access to the kernel log
7825 buffer anymore, i.e. the syslog() system call (not to be confused
7826 with the API of the same name in libc, which is not affected), the
7827 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
7828 inaccessible to the service. It's recommended to enable this setting
7829 for all services that should not be able to read from or write to the
7830 kernel log buffer, which are probably almost all.
7832 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
7833 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
7834 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
7835 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
7836 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
7837 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
7838 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
7839 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
7840 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
7841 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
7842 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
7843 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
7844 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
7845 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
7846 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
7847 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
7848 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
7849 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
7850 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
7851 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
7852 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
7853 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
7854 Zach Smith, Zbigniew Jędrzejewski-Szmek
7856 – Warsaw, 2019-11-29
7860 * This release enables unprivileged programs (i.e. requiring neither
7861 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
7862 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
7863 kernel for the whole UNIX group range, i.e. all processes. This
7864 change should be reasonably safe, as the kernel support for it was
7865 specifically implemented to allow safe access to ICMP Echo for
7866 processes lacking any privileges. If this is not desirable, it can be
7867 disabled again by setting the parameter to "1 0".
7869 * Previously, filters defined with SystemCallFilter= would have the
7870 effect that any calling of an offending system call would terminate
7871 the calling thread. This behaviour never made much sense, since
7872 killing individual threads of unsuspecting processes is likely to
7873 create more problems than it solves. With this release the default
7874 action changed from killing the thread to killing the whole
7875 process. For this to work correctly both a kernel version (>= 4.14)
7876 and a libseccomp version (>= 2.4.0) supporting this new seccomp
7877 action is required. If an older kernel or libseccomp is used the old
7878 behaviour continues to be used. This change does not affect any
7879 services that have no system call filters defined, or that use
7880 SystemCallErrorNumber= (and thus see EPERM or another error instead
7881 of being killed when calling an offending system call). Note that
7882 systemd documentation always claimed that the whole process is
7883 killed. With this change behaviour is thus adjusted to match the
7886 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
7887 4194304 by default, i.e. the full 22bit range the kernel allows, up
7888 from the old 16-bit range. This should improve security and
7889 robustness, as PID collisions are made less likely (though certainly
7890 still possible). There are rumours this might create compatibility
7891 problems, though at this moment no practical ones are known to
7892 us. Downstream distributions are hence advised to undo this change in
7893 their builds if they are concerned about maximum compatibility, but
7894 for everybody else we recommend leaving the value bumped. Besides
7895 improving security and robustness this should also simplify things as
7896 the maximum number of allowed concurrent tasks was previously bounded
7897 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
7898 only a single knob is left ("kernel.threads-max"). There have been
7899 concerns that usability is affected by this change because larger PID
7900 numbers are harder to type, but we believe the change from 5 digits
7901 to 7 digits doesn't hamper usability.
7903 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
7904 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
7905 hierarchically set default memory protection values for a particular
7906 subtree of the unit hierarchy.
7908 * Memory protection directives can now take a value of zero, allowing
7909 explicit opting out of a default value propagated by an ancestor.
7911 * systemd now defaults to the "unified" cgroup hierarchy setup during
7912 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
7913 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
7914 change reflects the fact that cgroupsv2 support has matured
7915 substantially in both systemd and in the kernel, and is clearly the
7916 way forward. Downstream production distributions might want to
7917 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
7918 their builds as unfortunately the popular container managers have not
7919 caught up with the kernel API changes.
7921 * Man pages are not built by default anymore (html pages were already
7922 disabled by default), to make development builds quicker. When
7923 building systemd for a full installation with documentation, meson
7924 should be called with -Dman=true and/or -Dhtml=true as appropriate.
7925 The default was changed based on the assumption that quick one-off or
7926 repeated development builds are much more common than full optimized
7927 builds for installation, and people need to pass various other
7928 options to when doing "proper" builds anyway, so the gain from making
7929 development builds quicker is bigger than the one time disruption for
7932 Two scripts are created in the *build* directory to generate and
7933 preview man and html pages on demand, e.g.:
7935 build/man/man systemctl
7936 build/man/html systemd.index
7938 * libidn2 is used by default if both libidn2 and libidn are installed.
7939 Please use -Dlibidn=true if libidn is preferred.
7941 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
7942 big-endian machines. Before, bytes were written and read in native
7943 machine order as exposed by the native libc __cpu_mask interface.
7944 Now, little-endian order is always used (CPUs 0–7 are described by
7945 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
7946 This change fixes D-Bus calls that cross endianness boundary.
7948 The presentation format used for CPUAffinity= by "systemctl show" and
7949 "systemd-analyze dump" is changed to present CPU indices instead of
7950 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
7951 shown as CPUAffinity=03000000000000000000000000000… (on
7952 little-endian) or CPUAffinity=00000000000000300000000000000… (on
7953 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
7954 input format. The maximum integer that will be printed in the new
7955 format is 8191 (four digits), while the old format always used a very
7956 long number (with the length varying by architecture), so they can be
7957 unambiguously distinguished.
7959 * /usr/sbin/halt.local is no longer supported. Implementation in
7960 distributions was inconsistent and it seems this functionality was
7963 To replace this functionality, users should:
7964 - either define a new unit and make it a dependency of final.target
7965 (systemctl add-wants final.target my-halt-local.service)
7966 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
7967 and ensure that it accepts "halt", "poweroff", "reboot", and
7968 "kexec" as an argument, see the description in systemd-shutdown(8).
7970 * When a [Match] section in .link or .network file is empty (contains
7971 no match patterns), a warning will be emitted. Please add any "match
7972 all" pattern instead, e.g. OriginalName=* or Name=* in case all
7973 interfaces should really be matched.
7975 * A new setting NUMAPolicy= may be used to set process memory
7976 allocation policy. This setting can be specified in
7977 /etc/systemd/system.conf and hence will set the default policy for
7978 PID1. The default policy can be overridden on a per-service
7979 basis. The related setting NUMAMask= is used to specify NUMA node
7980 mask that should be associated with the selected policy.
7982 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
7983 generates when processes it manages are reaching their memory limits,
7984 and will place their units in a special state, and optionally kill or
7985 stop the whole unit.
7987 * The service manager will now expose bus properties for the IO
7988 resources used by units. This information is also shown in "systemctl
7989 status" now (for services that have IOAccounting=yes set). Moreover,
7990 the IO accounting data is included in the resource log message
7991 generated whenever a unit stops.
7993 * Units may now configure an explicit timeout to wait for when killed
7994 with SIGABRT, for example when a service watchdog is hit. Previously,
7995 the regular TimeoutStopSec= timeout was applied in this case too —
7996 now a separate timeout may be set using TimeoutAbortSec=.
7998 * Services may now send a special WATCHDOG=trigger message with
7999 sd_notify() to trigger an immediate "watchdog missed" event, and thus
8000 trigger service termination. This is useful both for testing watchdog
8001 handling, but also for defining error paths in services, that shall
8002 be handled the same way as watchdog events.
8004 * There are two new per-unit settings IPIngressFilterPath= and
8005 IPEgressFilterPath= which allow configuration of a BPF program
8006 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
8007 to apply to the IP packet ingress/egress path of all processes of a
8008 unit. This is useful to allow running systemd services with BPF
8009 programs set up externally.
8011 * systemctl gained a new "clean" verb for removing the state, cache,
8012 runtime or logs directories of a service while it is terminated. The
8013 new verb may also be used to remove the state maintained on disk for
8014 timer units that have Persistent= configured.
8016 * During the last phase of shutdown systemd will now automatically
8017 increase the log level configured in the "kernel.printk" sysctl so
8018 that any relevant loggable events happening during late shutdown are
8019 made visible. Previously, loggable events happening so late during
8020 shutdown were generally lost if the "kernel.printk" sysctl was set to
8021 high thresholds, as regular logging daemons are terminated at that
8022 time and thus nothing is written to disk.
8024 * If processes terminated during the last phase of shutdown do not exit
8025 quickly systemd will now show their names after a short time, to make
8026 debugging easier. After a longer timeout they are forcibly killed,
8029 * journalctl (and the other tools that display logs) will now highlight
8030 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
8031 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
8032 are now shown in blue color, to separate them visually from regular
8033 logs. References to configuration files are now turned into clickable
8034 links on terminals that support that.
8036 * systemd-journald will now stop logging to /var/log/journal during
8037 shutdown when /var/ is on a separate mount, so that it can be
8038 unmounted safely during shutdown.
8040 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
8042 * systemd-resolved "Cache=" configuration option in resolved.conf has
8043 been extended to also accept the 'no-negative' value. Previously,
8044 only a boolean option was allowed (yes/no), having yes as the
8045 default. If this option is set to 'no-negative', negative answers are
8046 not cached while the old cache heuristics are used positive answers.
8047 The default remains unchanged.
8049 * The predictable naming scheme for network devices now supports
8050 generating predictable names for "netdevsim" devices.
8052 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
8055 Those two changes form a new net.naming_scheme= entry. Distributions
8056 which want to preserve naming stability may want to set the
8057 -Ddefault-net-naming-scheme= configuration option.
8059 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
8060 interfaces natively.
8062 * systemd-networkd's bridge FDB support now allows configuration of a
8063 destination address for each entry (Destination=), as well as the
8064 VXLAN VNI (VNI=), as well as an option to declare what an entry is
8065 associated with (AssociatedWith=).
8067 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
8068 option for configuring the maximum number of DHCP lease requests. It
8069 also learnt a new BlackList= option for deny-listing DHCP servers (a
8070 similar setting has also been added to the IPv6 RA client), as well
8071 as a SendRelease= option for configuring whether to send a DHCP
8072 RELEASE message when terminating.
8074 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
8075 separately in the [DHCPv4] and [DHCPv6] sections.
8077 * systemd-networkd's DHCP support will now optionally create an
8078 implicit host route to the DNS server specified in the DHCP lease, in
8079 addition to the routes listed explicitly in the lease. This should
8080 ensure that in multi-homed systems DNS traffic leaves the systems on
8081 the interface that acquired the DNS server information even if other
8082 routes such as default routes exist. This behaviour may be turned on
8083 with the new RoutesToDNS= option.
8085 * systemd-networkd's VXLAN support gained a new option
8086 GenericProtocolExtension= for enabling VXLAN Generic Protocol
8087 Extension support, as well as IPDoNotFragment= for setting the IP
8088 "Don't fragment" bit on outgoing packets. A similar option has been
8089 added to the GENEVE support.
8091 * In systemd-networkd's [Route] section you may now configure
8092 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
8093 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
8094 propagation. The Type= setting now supports local, broadcast,
8095 anycast, multicast, any, xresolve routes, too.
8097 * systemd-networkd's [Network] section learnt a new option
8098 DefaultRouteOnDevice= for automatically configuring a default route
8099 onto the network device.
8101 * systemd-networkd's bridging support gained two new options ProxyARP=
8102 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
8103 MulticastRouter= for configuring multicast routing behaviour. A new
8104 option MulticastIGMPVersion= may be used to change bridge's multicast
8105 Internet Group Management Protocol (IGMP) version.
8107 * systemd-networkd's FooOverUDP support gained the ability to configure
8108 local and peer IP addresses via Local= and Peer=. A new option
8109 PeerPort= may be used to configure the peer's IP port.
8111 * systemd-networkd's TUN support gained a new setting VnetHeader= for
8112 tweaking Generic Segment Offload support.
8114 * The address family for policy rules may be specified using the new
8115 Family= option in the [RoutingPolicyRule] section.
8117 * networkctl gained a new "delete" command for removing virtual network
8118 devices, as well as a new "--stats" switch for showing device
8121 * networkd.conf gained a new setting SpeedMeter= and
8122 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
8123 measured speed may be shown by 'networkctl status'.
8125 * "networkctl status" now displays MTU and queue lengths, and more
8126 detailed information about VXLAN and bridge devices.
8128 * systemd-networkd's .network and .link files gained a new Property=
8129 setting in the [Match] section, to match against devices with
8130 specific udev properties.
8132 * systemd-networkd's tunnel support gained a new option
8133 AssignToLoopback= for selecting whether to use the loopback device
8134 "lo" as underlying device.
8136 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
8137 been renamed to LinkLayerAddress=, and it now allows configuration of
8140 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
8141 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
8142 IPv6 configuration (static or DHCPv6) was found for a given
8143 interface. It will not touch the sysctl otherwise.
8145 * The order of entries is $PATH used by the user manager instance was
8146 changed to put bin/ entries before the corresponding sbin/ entries.
8147 It is recommended to not rely on this order, and only ever have one
8148 binary with a given name in the system paths under /usr.
8150 * A new tool systemd-network-generator has been added that may generate
8151 .network, .netdev and .link files from IP configuration specified on
8152 the kernel command line in the format used by Dracut.
8154 * The CriticalConnection= setting in .network files is now deprecated,
8155 and replaced by a new KeepConfiguration= setting which allows more
8156 detailed configuration of the IP configuration to keep in place.
8158 * systemd-analyze gained a few new verbs:
8160 - "systemd-analyze timestamp" parses and converts timestamps. This is
8161 similar to the existing "systemd-analyze calendar" command which
8162 does the same for recurring calendar events.
8164 - "systemd-analyze timespan" parses and converts timespans (i.e.
8165 durations as opposed to points in time).
8167 - "systemd-analyze condition" will parse and test ConditionXYZ=
8170 - "systemd-analyze exit-status" will parse and convert exit status
8171 codes to their names and back.
8173 - "systemd-analyze unit-files" will print a list of all unit
8174 file paths and unit aliases.
8176 * SuccessExitStatus=, RestartPreventExitStatus=, and
8177 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
8178 is equivalent to "65"). Those exit status name mappings may be
8179 displayed with the systemd-analyze exit-status verb describe above.
8181 * systemd-logind now exposes a per-session SetBrightness() bus call,
8182 which may be used to securely change the brightness of a kernel
8183 brightness device, if it belongs to the session's seat. By using this
8184 call unprivileged clients can make changes to "backlight" and "leds"
8185 devices securely with strict requirements on session membership.
8186 Desktop environments may use this to generically make brightness
8187 changes to such devices without shipping private SUID binaries or
8188 udev rules for that purpose.
8190 * "udevadm info" gained a --wait-for-initialization switch to wait for
8191 a device to be initialized.
8193 * systemd-hibernate-resume-generator will now look for resumeflags= on
8194 the kernel command line, which is similar to rootflags= and may be
8195 used to configure device timeout for the hibernation device.
8197 * sd-event learnt a new API call sd_event_source_disable_unref() for
8198 disabling and unref'ing an event source in a single function. A
8199 related call sd_event_source_disable_unrefp() has been added for use
8200 with gcc's cleanup extension.
8202 * The sd-id128.h public API gained a new definition
8203 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
8206 * "busctl introspect" gained a new switch --xml-interface for dumping
8207 XML introspection data unmodified.
8209 * PID 1 may now show the unit name instead of the unit description
8210 string in its status output during boot. This may be configured in
8211 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
8212 kernel command line option systemd.status_unit_format=.
8214 * PID 1 now understands a new option KExecWatchdogSec= in
8215 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
8216 Previously watchdog functionality was only available for regular
8217 reboots. The new setting defaults to off, because we don't know in
8218 the general case if the watchdog will be reset after kexec (some
8219 drivers do reset it, but not all), and the new userspace might not be
8220 configured to handle the watchdog.
8222 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
8223 RebootWatchdogSec= to more clearly communicate what it is about. The
8224 old name is still accepted for compatibility.
8226 * The systemd.debug_shell kernel command line option now optionally
8227 takes a tty name to spawn the debug shell on, which allows a
8228 different tty to be selected than the built-in default.
8230 * Service units gained a new ExecCondition= setting which will run
8231 before ExecStartPre= and either continue execution of the unit (for
8232 clean exit codes), stop execution without marking the unit failed
8233 (for exit codes 1 through 254), or stop execution and fail the unit
8234 (for exit code 255 or abnormal termination).
8236 * A new service systemd-pstore.service has been added that pulls data
8237 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
8240 * timedatectl gained new verbs for configuring per-interface NTP
8241 service configuration for systemd-timesyncd.
8243 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
8244 2019. (You can set non-UTF-8 locales though, if you know their name.)
8246 * If variable assignments in sysctl.d/ files are prefixed with "-" any
8247 failures to apply them are now ignored.
8249 * systemd-random-seed.service now optionally credits entropy when
8250 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
8251 true for the service to enable this behaviour, but please consult the
8252 documentation first, since this comes with a couple of caveats.
8254 * systemd-random-seed.service is now a synchronization point for full
8255 initialization of the kernel's entropy pool. Services that require
8256 /dev/urandom to be correctly initialized should be ordered after this
8259 * The systemd-boot boot loader has been updated to optionally maintain
8260 a random seed file in the EFI System Partition (ESP). During the boot
8261 phase, this random seed is read and updated with a new seed
8262 cryptographically derived from it. Another derived seed is passed to
8263 the OS. The latter seed is then credited to the kernel's entropy pool
8264 very early during userspace initialization (from PID 1). This allows
8265 systems to boot up with a fully initialized kernel entropy pool from
8266 earliest boot on, and thus entirely removes all entropy pool
8267 initialization delays from systems using systemd-boot. Special care
8268 is taken to ensure different seeds are derived on system images
8269 replicated to multiple systems. "bootctl status" will show whether
8270 a seed was received from the boot loader.
8272 * bootctl gained two new verbs:
8274 - "bootctl random-seed" will generate the file in ESP and an EFI
8275 variable to allow a random seed to be passed to the OS as described
8278 - "bootctl is-installed" checks whether systemd-boot is currently
8281 * bootctl will warn if it detects that boot entries are misconfigured
8282 (for example if the kernel image was removed without purging the
8285 * A new document has been added describing systemd's use and support
8286 for the kernel's entropy pool subsystem:
8288 https://systemd.io/RANDOM_SEEDS
8290 * When the system is hibernated the swap device to write the
8291 hibernation image to is now automatically picked from all available
8292 swap devices, preferring the swap device with the highest configured
8293 priority over all others, and picking the device with the most free
8294 space if there are multiple devices with the highest priority.
8296 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
8297 option that permits selecting the timeout how long to wait for a
8298 device with an encryption key before asking for the password.
8300 * IOWeight= has learnt to properly set the IO weight when using the
8301 BFQ scheduler officially found in kernels 5.0+.
8303 * A new mailing list has been created for reporting of security issues:
8304 systemd-security@redhat.com. For mode details, see
8305 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
8307 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
8308 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
8309 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
8310 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
8311 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
8312 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
8313 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
8314 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
8315 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
8316 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
8317 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
8318 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
8319 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
8320 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
8321 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
8322 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
8323 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
8324 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
8325 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
8326 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
8327 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
8328 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
8329 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
8330 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
8331 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
8332 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
8333 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
8334 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
8335 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
8336 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
8338 – Camerino, 2019-09-03
8342 * In .link files, MACAddressPolicy=persistent (the default) is changed
8343 to cover more devices. For devices like bridges, tun, tap, bond, and
8344 similar interfaces that do not have other identifying information,
8345 the interface name is used as the basis for persistent seed for MAC
8346 and IPv4LL addresses. The way that devices that were handled
8347 previously is not changed, and this change is about covering more
8348 devices then previously by the "persistent" policy.
8350 MACAddressPolicy=random may be used to force randomized MACs and
8351 IPv4LL addresses for a device if desired.
8353 Hint: the log output from udev (at debug level) was enhanced to
8354 clarify what policy is followed and which attributes are used.
8355 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
8356 may be used to view this.
8358 Hint: if a bridge interface is created without any slaves, and gains
8359 a slave later, then now the bridge does not inherit slave's MAC.
8360 To inherit slave's MAC, for example, create the following file:
8362 # /etc/systemd/network/98-bridge-inherit-mac.link
8367 MACAddressPolicy=none
8370 * The .device units generated by systemd-fstab-generator and other
8371 generators do not automatically pull in the corresponding .mount unit
8372 as a Wants= dependency. This means that simply plugging in the device
8373 will not cause the mount unit to be started automatically. But please
8374 note that the mount unit may be started for other reasons, in
8375 particular if it is part of local-fs.target, and any unit which
8376 (transitively) depends on local-fs.target is started.
8378 * networkctl list/status/lldp now accept globbing wildcards for network
8379 interface names to match against all existing interfaces.
8381 * The $PIDFILE environment variable is set to point the absolute path
8382 configured with PIDFile= for processes of that service.
8384 * The fallback DNS server list was augmented with Cloudflare public DNS
8385 servers. Use `-Ddns-servers=` to set a different fallback.
8387 * A new special target usb-gadget.target will be started automatically
8388 when a USB Device Controller is detected (which means that the system
8389 is a USB peripheral).
8391 * A new unit setting CPUQuotaPeriodSec= assigns the time period
8392 relatively to which the CPU time quota specified by CPUQuota= is
8395 * A new unit setting ProtectHostname= may be used to prevent services
8396 from modifying hostname information (even if they otherwise would
8397 have privileges to do so).
8399 * A new unit setting NetworkNamespacePath= may be used to specify a
8400 namespace for service or socket units through a path referring to a
8401 Linux network namespace pseudo-file.
8403 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
8404 have an effect on .socket units: when used the listening socket is
8405 created within the configured network namespace instead of the host
8408 * ExecStart= command lines in unit files may now be prefixed with ':'
8409 in which case environment variable substitution is
8410 disabled. (Supported for the other ExecXYZ= settings, too.)
8412 * .timer units gained two new boolean settings OnClockChange= and
8413 OnTimezoneChange= which may be used to also trigger a unit when the
8414 system clock is changed or the local timezone is
8415 modified. systemd-run has been updated to make these options easily
8416 accessible from the command line for transient timers.
8418 * Two new conditions for units have been added: ConditionMemory= may be
8419 used to conditionalize a unit based on installed system
8420 RAM. ConditionCPUs= may be used to conditionalize a unit based on
8421 installed CPU cores.
8423 * The @default system call filter group understood by SystemCallFilter=
8424 has been updated to include the new rseq() system call introduced in
8427 * A new time-set.target has been added that indicates that the system
8428 time has been set from a local source (possibly imprecise). The
8429 existing time-sync.target is stronger and indicates that the time has
8430 been synchronized with a precise external source. Services where
8431 approximate time is sufficient should use the new target.
8433 * "systemctl start" (and related commands) learnt a new
8434 --show-transaction option. If specified brief information about all
8435 jobs queued because of the requested operation is shown.
8437 * systemd-networkd recognizes a new operation state 'enslaved', used
8438 (instead of 'degraded' or 'carrier') for interfaces which form a
8439 bridge, bond, or similar, and an new 'degraded-carrier' operational
8440 state used for the bond or bridge master interface when one of the
8441 enslaved devices is not operational.
8443 * .network files learnt the new IgnoreCarrierLoss= option for leaving
8444 networks configured even if the carrier is lost.
8446 * The RequiredForOnline= setting in .network files may now specify a
8447 minimum operational state required for the interface to be considered
8448 "online" by systemd-networkd-wait-online. Related to this
8449 systemd-networkd-wait-online gained a new option --operational-state=
8450 to configure the same, and its --interface= option was updated to
8451 optionally also take an operational state specific for an interface.
8453 * systemd-networkd-wait-online gained a new setting --any for waiting
8454 for only one of the requested interfaces instead of all of them.
8456 * systemd-networkd now implements L2TP tunnels.
8458 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
8459 may be used to cause autonomous and onlink prefixes received in IPv6
8460 Router Advertisements to be ignored.
8462 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
8463 file settings may be used to tweak bridge behaviour.
8465 * The new TripleSampling= option in .network files may be used to
8466 configure CAN triple sampling.
8468 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
8469 used to point to private or preshared key for a WireGuard interface.
8471 * /etc/crypttab now supports the same-cpu-crypt and
8472 submit-from-crypt-cpus options to tweak encryption work scheduling
8475 * systemd-tmpfiles will now take a BSD file lock before operating on a
8476 contents of directory. This may be used to temporarily exclude
8477 directories from aging by taking the same lock (useful for example
8478 when extracting a tarball into /tmp or /var/tmp as a privileged user,
8479 which might create files with really old timestamps, which
8480 nevertheless should not be deleted). For further details, see:
8482 https://systemd.io/TEMPORARY_DIRECTORIES
8484 * systemd-tmpfiles' h line type gained support for the
8485 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
8486 controlling project quota inheritance.
8488 * sd-boot and bootctl now implement support for an Extended Boot Loader
8489 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
8490 addition to the ESP partition mounted to /efi or /boot/efi.
8491 Configuration file fragments, kernels, initrds and other EFI images
8492 to boot will be loaded from both the ESP and XBOOTLDR partitions.
8493 The XBOOTLDR partition was previously described by the Boot Loader
8494 Specification, but implementation was missing in sd-boot. Support for
8495 this concept allows using the sd-boot boot loader in more
8496 conservative scenarios where the boot loader itself is placed in the
8497 ESP but the kernels to boot (and their metadata) in a separate
8500 * A system may now be booted with systemd.volatile=overlay on the
8501 kernel command line, which causes the root file system to be set up
8502 an overlayfs mount combining the root-only root directory with a
8503 writable tmpfs. In this setup, the underlying root device is not
8504 modified, and any changes are lost at reboot.
8506 * Similar, systemd-nspawn can now boot containers with a volatile
8507 overlayfs root with the new --volatile=overlay switch.
8509 * systemd-nspawn can now consume OCI runtime bundles using a new
8510 --oci-bundle= option. This implementation is fully usable, with most
8511 features in the specification implemented, but since this a lot of
8512 new code and functionality, this feature should most likely not
8513 be used in production yet.
8515 * systemd-nspawn now supports various options described by the OCI
8516 runtime specification on the command-line and in .nspawn files:
8517 --inaccessible=/Inaccessible= may be used to mask parts of the file
8518 system tree, --console=/--pipe may be used to configure how standard
8519 input, output, and error are set up.
8521 * busctl learned the `emit` verb to generate D-Bus signals.
8523 * systemd-analyze cat-config may be used to gather and display
8524 configuration spread over multiple files, for example system and user
8525 presets, tmpfiles.d, sysusers.d, udev rules, etc.
8527 * systemd-analyze calendar now takes an optional new parameter
8528 --iterations= which may be used to show a maximum number of iterations
8529 the specified expression will elapse next.
8531 * The sd-bus C API gained support for naming method parameters in the
8534 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
8535 the reboot() system call expects.
8537 * journalctl learnt a new --cursor-file= option that points to a file
8538 from which a cursor should be loaded in the beginning and to which
8539 the updated cursor should be stored at the end.
8541 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
8542 detected by systemd-detect-virt (and may also be used in
8543 ConditionVirtualization=).
8545 * The behaviour of systemd-logind may now be modified with environment
8546 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
8547 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
8548 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
8549 skip the relevant operation completely (when set to false), or to
8550 create a flag file in /run/systemd (when set to true), instead of
8551 actually commencing the real operation when requested. The presence
8552 of /run/systemd/reboot-to-firmware-setup,
8553 /run/systemd/reboot-to-boot-loader-menu, and
8554 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
8555 boot loader implementations to replace some steps logind performs
8556 during reboot with their own operations.
8558 * systemctl can be used to request a reboot into the boot loader menu
8559 or a specific boot loader entry with the new --boot-load-menu= and
8560 --boot-loader-entry= options to a reboot command. (This requires a
8561 boot loader that supports this, for example sd-boot.)
8563 * kernel-install will no longer unconditionally create the output
8564 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
8565 snippets, but will do only if the machine-specific parent directory
8566 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
8567 to create this parent directory during sd-boot installation.
8569 This makes it easier to use kernel-install with plugins which support
8570 a different layout of the bootloader partitions (for example grub2).
8572 * During package installation (with `ninja install`), we would create
8573 symlinks for getty@tty1.service, systemd-networkd.service,
8574 systemd-networkd.socket, systemd-resolved.service,
8575 remote-cryptsetup.target, remote-fs.target,
8576 systemd-networkd-wait-online.service, and systemd-timesyncd.service
8577 in /etc, as if `systemctl enable` was called for those units, to make
8578 the system usable immediately after installation. Now this is not
8579 done anymore, and instead calling `systemctl preset-all` is
8580 recommended after the first installation of systemd.
8582 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
8583 is built on seccomp. When turned on creation of SUID/SGID files is
8586 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
8587 implied if DynamicUser= is turned on for a service. This hardens
8588 these services, so that they neither can benefit from nor create
8589 SUID/SGID executables. This is a minor compatibility breakage, given
8590 that when DynamicUser= was first introduced SUID/SGID behaviour was
8591 unaffected. However, the security benefit of these two options is
8592 substantial, and the setting is still relatively new, hence we opted
8593 to make it mandatory for services with dynamic users.
8595 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
8596 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
8597 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
8598 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
8599 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
8600 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
8601 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
8602 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
8603 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
8604 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
8605 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
8606 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
8607 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
8608 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
8609 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
8610 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
8611 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
8612 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8614 — Warsaw, 2019-04-11
8618 * The default locale can now be configured at compile time. Otherwise,
8619 a suitable default will be selected automatically (one of C.UTF-8,
8620 en_US.UTF-8, and C).
8622 * The version string shown by systemd and other tools now includes the
8623 git commit hash when built from git. An override may be specified
8624 during compilation, which is intended to be used by distributions to
8625 include the package release information.
8627 * systemd-cat can now filter standard input and standard error streams
8628 for different syslog priorities using the new --stderr-priority=
8631 * systemd-journald and systemd-journal-remote reject entries which
8632 contain too many fields (CVE-2018-16865) and set limits on the
8633 process' command line length (CVE-2018-16864).
8635 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
8638 * A new network device NamePolicy "keep" is implemented for link files,
8639 and used by default in 99-default.link (the fallback configuration
8640 provided by systemd). With this policy, if the network device name
8641 was already set by userspace, the device will not be renamed again.
8642 This matches the naming scheme that was implemented before
8643 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
8644 is also enabled by default, even if not specified. Effectively, this
8645 means that if naming-scheme >= 240 is specified, network devices will
8646 be renamed according to the configuration, even if they have been
8647 renamed already, if "keep" is not specified as the naming policy in
8648 the .link file. The 99-default.link file provided by systemd includes
8649 "keep" for backwards compatibility, but it is recommended for user
8650 installed .link files to *not* include it.
8652 The "kernel" policy, which keeps kernel names declared to be
8653 "persistent", now works again as documented.
8655 * kernel-install script now optionally takes the paths to one or more
8656 initrd files, and passes them to all plugins.
8658 * The mincore() system call has been dropped from the @system-service
8659 system call filter group, as it is pretty exotic and may potentially
8660 used for side-channel attacks.
8662 * -fPIE is dropped from compiler and linker options. Please specify
8663 -Db_pie=true option to meson to build position-independent
8664 executables. Note that the meson option is supported since meson-0.49.
8666 * The fs.protected_regular and fs.protected_fifos sysctls, which were
8667 added in Linux 4.19 to make some data spoofing attacks harder, are
8668 now enabled by default. While this will hopefully improve the
8669 security of most installations, it is technically a backwards
8670 incompatible change; to disable these sysctls again, place the
8671 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
8673 fs.protected_regular = 0
8674 fs.protected_fifos = 0
8676 Note that the similar hardlink and symlink protection has been
8677 enabled since v199, and may be disabled likewise.
8679 * The files read from the EnvironmentFile= setting in unit files now
8680 parse backslashes inside quotes literally, matching the behaviour of
8683 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
8684 now automatically become NOPs when run in a chroot() environment.
8686 * The tmpfiles.d/ "C" line type will now copy directory trees not only
8687 when the destination is so far missing, but also if it already exists
8688 as a directory and is empty. This is useful to cater for systems
8689 where directory trees are put together from multiple separate mount
8690 points but otherwise empty.
8692 * A new function sd_bus_close_unref() (and the associated
8693 sd_bus_close_unrefp()) has been added to libsystemd, that combines
8694 sd_bus_close() and sd_bus_unref() in one.
8696 * udevadm control learnt a new option for --ping for testing whether a
8697 systemd-udevd instance is running and reacting.
8699 * udevadm trigger learnt a new option for --wait-daemon for waiting
8700 systemd-udevd daemon to be initialized.
8702 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
8703 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
8704 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
8705 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
8706 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
8707 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
8708 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
8709 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
8710 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
8711 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
8712 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
8713 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
8714 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
8715 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
8716 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
8717 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
8718 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
8720 — Berlin, 2019-02-14
8724 * NoNewPrivileges=yes has been set for all long-running services
8725 implemented by systemd. Previously, this was problematic due to
8726 SELinux (as this would also prohibit the transition from PID1's label
8727 to the service's label). This restriction has since been lifted, but
8728 an SELinux policy update is required.
8729 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
8731 * DynamicUser=yes is dropped from systemd-networkd.service,
8732 systemd-resolved.service and systemd-timesyncd.service, which was
8733 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
8734 and since v236 for systemd-timesyncd.service. The users and groups
8735 systemd-network, systemd-resolve and systemd-timesync are created
8736 by systemd-sysusers again. Distributors or system administrators
8737 may need to create these users and groups if they not exist (or need
8738 to re-enable DynamicUser= for those units) while upgrading systemd.
8739 Also, the clock file for systemd-timesyncd may need to move from
8740 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
8742 * When unit files are loaded from disk, previously systemd would
8743 sometimes (depending on the unit loading order) load units from the
8744 target path of symlinks in .wants/ or .requires/ directories of other
8745 units. This meant that unit could be loaded from different paths
8746 depending on whether the unit was requested explicitly or as a
8747 dependency of another unit, not honouring the priority of directories
8748 in search path. It also meant that it was possible to successfully
8749 load and start units which are not found in the unit search path, as
8750 long as they were requested as a dependency and linked to from
8751 .wants/ or .requires/. The target paths of those symlinks are not
8752 used for loading units anymore and the unit file must be found in
8755 * A new service type has been added: Type=exec. It's very similar to
8756 Type=simple but ensures the service manager will wait for both fork()
8757 and execve() of the main service binary to complete before proceeding
8758 with follow-up units. This is primarily useful so that the manager
8759 propagates any errors in the preparation phase of service execution
8760 back to the job that requested the unit to be started. For example,
8761 consider a service that has ExecStart= set to a file system binary
8762 that doesn't exist. With Type=simple starting the unit would be
8763 considered instantly successful, as only fork() has to complete
8764 successfully and the manager does not wait for execve(), and hence
8765 its failure is seen "too late". With the new Type=exec service type
8766 starting the unit will fail, as the manager will wait for the
8767 execve() and notice its failure, which is then propagated back to the
8770 NOTE: with the next release 241 of systemd we intend to change the
8771 systemd-run tool to default to Type=exec for transient services
8772 started by it. This should be mostly safe, but in specific corner
8773 cases might result in problems, as the systemd-run tool will then
8774 block on NSS calls (such as user name look-ups due to User=) done
8775 between the fork() and execve(), which under specific circumstances
8776 might cause problems. It is recommended to specify "-p Type=simple"
8777 explicitly in the few cases where this applies. For regular,
8778 non-transient services (i.e. those defined with unit files on disk)
8779 we will continue to default to Type=simple.
8781 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
8782 userspace processes is set to 1024 (soft) and 4096
8783 (hard). Previously, systemd passed this on unmodified to all
8784 processes it forked off. With this systemd release the hard limit
8785 systemd passes on is increased to 512K, overriding the kernel's
8786 defaults and substantially increasing the number of simultaneous file
8787 descriptors unprivileged userspace processes can allocate. Note that
8788 the soft limit remains at 1024 for compatibility reasons: the
8789 traditional UNIX select() call cannot deal with file descriptors >=
8790 1024 and increasing the soft limit globally might thus result in
8791 programs unexpectedly allocating a high file descriptor and thus
8792 failing abnormally when attempting to use it with select() (of
8793 course, programs shouldn't use select() anymore, and prefer
8794 poll()/epoll, but the call unfortunately remains undeservedly popular
8795 at this time). This change reflects the fact that file descriptor
8796 handling in the Linux kernel has been optimized in more recent
8797 kernels and allocating large numbers of them should be much cheaper
8798 both in memory and in performance than it used to be. Programs that
8799 want to take benefit of the increased limit have to "opt-in" into
8800 high file descriptors explicitly by raising their soft limit. Of
8801 course, when they do that they must acknowledge that they cannot use
8802 select() anymore (and neither can any shared library they use — or
8803 any shared library used by any shared library they use and so on).
8804 Which default hard limit is most appropriate is of course hard to
8805 decide. However, given reports that ~300K file descriptors are used
8806 in real-life applications we believe 512K is sufficiently high as new
8807 default for now. Note that there are also reports that using very
8808 high hard limits (e.g. 1G) is problematic: some software allocates
8809 large arrays with one element for each potential file descriptor
8810 (Java, …) — a high hard limit thus triggers excessively large memory
8811 allocations in these applications. Hopefully, the new default of 512K
8812 is a good middle ground: higher than what real-life applications
8813 currently need, and low enough for avoid triggering excessively large
8814 allocations in problematic software. (And yes, somebody should fix
8817 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
8818 to the highest possible values, as separate accounting of file
8819 descriptors is no longer necessary, as memcg tracks them correctly as
8820 part of the memory accounting anyway. Thus, from the four limits on
8821 file descriptors currently enforced (fs.file-max, fs.nr_open,
8822 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
8823 and keep only the latter two. A set of build-time options
8824 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
8825 has been added to revert this change in behaviour, which might be
8826 an option for systems that turn off memcg in the kernel.
8828 * When no /etc/locale.conf file exists (and hence no locale settings
8829 are in place), systemd will now use the "C.UTF-8" locale by default,
8830 and set LANG= to it. This locale is supported by various
8831 distributions including Fedora, with clear indications that upstream
8832 glibc is going to make it available too. This locale enables UTF-8
8833 mode by default, which appears appropriate for 2018.
8835 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
8836 default. This effectively switches the RFC3704 Reverse Path filtering
8837 from Strict mode to Loose mode. This is more appropriate for hosts
8838 that have multiple links with routes to the same networks (e.g.
8839 a client with a Wi-Fi and Ethernet both connected to the internet).
8841 Consult the kernel documentation for details on this sysctl:
8842 https://docs.kernel.org/networking/ip-sysctl.html
8844 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
8847 * CPUAccounting=yes no longer enables the CPU controller when using
8848 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
8849 statistics are now provided independently from the CPU controller.
8851 * Support for disabling a particular cgroup controller within a sub-tree
8852 has been added through the DisableControllers= directive.
8854 * cgroup_no_v1=all on the kernel command line now also implies
8855 using the unified cgroup hierarchy, unless one explicitly passes
8856 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
8858 * The new "MemoryMin=" unit file property may now be used to set the
8859 memory usage protection limit of processes invoked by the unit. This
8860 controls the cgroup v2 memory.min attribute. Similarly, the new
8861 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
8862 cgroup v2 io.latency cgroup property for configuring per-service I/O
8865 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
8866 to the cgroup v1 "devices" cgroup controller.
8868 * systemd-escape now is able to combine --unescape with --template. It
8869 also learnt a new option --instance for extracting and unescaping the
8870 instance part of a unit name.
8872 * sd-bus now provides the sd_bus_message_readv() which is similar to
8873 sd_bus_message_read() but takes a va_list object. The pair
8874 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
8875 has been added for configuring the default method call timeout to
8876 use. sd_bus_error_move() may be used to efficiently move the contents
8877 from one sd_bus_error structure to another, invalidating the
8878 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
8879 be used to control whether a bus connection object is automatically
8880 flushed when an sd-event loop is exited.
8882 * When processing classic BSD syslog log messages, journald will now
8883 save the original time-stamp string supplied in the new
8884 SYSLOG_TIMESTAMP= journal field. This permits consumers to
8885 reconstruct the original BSD syslog message more correctly.
8887 * StandardOutput=/StandardError= in service files gained support for
8888 new "append:…" parameters, for connecting STDOUT/STDERR of a service
8889 to a file, and appending to it.
8891 * The signal to use as last step of killing of unit processes is now
8892 configurable. Previously it was hard-coded to SIGKILL, which may now
8893 be overridden with the new KillSignal= setting. Note that this is the
8894 signal used when regular termination (i.e. SIGTERM) does not suffice.
8895 Similarly, the signal used when aborting a program in case of a
8896 watchdog timeout may now be configured too (WatchdogSignal=).
8898 * The XDG_SESSION_DESKTOP environment variable may now be configured in
8899 the pam_systemd argument line, using the new desktop= switch. This is
8900 useful to initialize it properly from a display manager without
8901 having to touch C code.
8903 * Most configuration options that previously accepted percentage values
8904 now also accept permille values with the '‰' suffix (instead of '%').
8906 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
8909 * systemd-resolved's configuration file resolved.conf gained a new
8910 option ReadEtcHosts= which may be used to turn off processing and
8911 honoring /etc/hosts entries.
8913 * The "--wait" switch may now be passed to "systemctl
8914 is-system-running", in which case the tool will synchronously wait
8915 until the system finished start-up.
8917 * hostnamed gained a new bus call to determine the DMI product UUID.
8919 * On x86-64 systemd will now prefer using the RDRAND processor
8920 instruction over /dev/urandom whenever it requires randomness that
8921 neither has to be crypto-grade nor should be reproducible. This
8922 should substantially reduce the amount of entropy systemd requests
8923 from the kernel during initialization on such systems, though not
8924 reduce it to zero. (Why not zero? systemd still needs to allocate
8925 UUIDs and such uniquely, which require high-quality randomness.)
8927 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
8928 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
8929 for forcing the "Other Information" bit in IPv6 RA messages. The
8930 bonding logic gained four new options AdActorSystemPriority=,
8931 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
8932 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
8933 shuffling of flows. The tunnel logic gained a new
8934 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
8935 Deployment. The policy rule logic gained four new options IPProtocol=,
8936 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
8937 support for the MulticastToUnicast= option. networkd also gained
8938 support for configuring static IPv4 ARP or IPv6 neighbor entries.
8940 * .preset files (as read by 'systemctl preset') may now be used to
8941 instantiate services.
8943 * /etc/crypttab now understands the sector-size= option to configure
8944 the sector size for an encrypted partition.
8946 * Key material for encrypted disks may now be placed on a formatted
8947 medium, and referenced from /etc/crypttab by the UUID of the file
8948 system, followed by "=" suffixed by the path to the key file.
8950 * The "collect" udev component has been removed without replacement, as
8951 it is neither used nor maintained.
8953 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
8954 LogsDirectory=, ConfigurationDirectory= settings are used in a
8955 service the executed processes will now receive a set of environment
8956 variables containing the full paths of these directories.
8957 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
8958 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
8959 are used. Note that these options may be used multiple times per
8960 service in which case the resulting paths will be concatenated and
8961 separated by colons.
8963 * Predictable interface naming has been extended to cover InfiniBand
8964 NICs. They will be exposed with an "ib" prefix.
8966 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
8967 which case the respective line failing is ignored.
8969 * .link files may now be used to configure the equivalent to the
8970 "ethtool advertise" commands.
8972 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
8973 alternative to libudev.h. Previously, the latter was just an internal
8974 wrapper around the former, but now these two APIs are exposed
8977 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
8978 which calculates an app-specific boot ID similar to how
8979 sd_id128_get_machine_app_specific() generates an app-specific machine
8982 * A new tool systemd-id128 has been added that can be used to determine
8983 and generate various 128-bit IDs.
8985 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
8988 * systemd-hibernate-resume-generator will now honor the "noresume"
8989 kernel command line option, in which case it will bypass resuming
8990 from any hibernated image.
8992 * The systemd-sleep.conf configuration file gained new options
8993 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
8994 AllowHybridSleep= for prohibiting specific sleep modes even if the
8995 kernel exports them.
8997 * portablectl is now officially supported and has thus moved to
9000 * bootctl learnt the two new commands "set-default" and "set-oneshot"
9001 for setting the default boot loader item to boot to (either
9002 persistently or only for the next boot). This is currently only
9003 compatible with sd-boot, but may be implemented on other boot loaders
9004 too, that follow the boot loader interface. The updated interface is
9005 now documented here:
9007 https://systemd.io/BOOT_LOADER_INTERFACE
9009 * A new kernel command line option systemd.early_core_pattern= is now
9010 understood which may be used to influence the core_pattern PID 1
9011 installs during early boot.
9013 * busctl learnt two new options -j and --json= for outputting method
9014 call replies, properties and monitoring output in JSON.
9016 * journalctl's JSON output now supports simple ANSI coloring as well as
9017 a new "json-seq" mode for generating RFC7464 output.
9019 * Unit files now support the %g/%G specifiers that resolve to the UNIX
9020 group/GID of the service manager runs as, similar to the existing
9021 %u/%U specifiers that resolve to the UNIX user/UID.
9023 * systemd-logind learnt a new global configuration option
9024 UserStopDelaySec= that may be set in logind.conf. It specifies how
9025 long the systemd --user instance shall remain started after a user
9026 logs out. This is useful to speed up repetitive re-connections of the
9027 same user, as it means the user's service manager doesn't have to be
9028 stopped/restarted on each iteration, but can be reused between
9029 subsequent options. This setting defaults to 10s. systemd-logind also
9030 exports two new properties on its Manager D-Bus objects indicating
9031 whether the system's lid is currently closed, and whether the system
9034 * systemd gained support for a generic boot counting logic, which
9035 generically permits automatic reverting to older boot loader entries
9036 if newer updated ones don't work. The boot loader side is implemented
9037 in sd-boot, but is kept open for other boot loaders too. For details
9040 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
9042 * The SuccessAction=/FailureAction= unit file settings now learnt two
9043 new parameters: "exit" and "exit-force", which result in immediate
9044 exiting of the service manager, and are only useful in systemd --user
9045 and container environments.
9047 * Unit files gained support for a pair of options
9048 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
9049 exit status to use as service manager exit status when
9050 SuccessAction=/FailureAction= is set to exit or exit-force.
9052 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
9053 options may now be used to configure the log rate limiting applied by
9054 journald per-service.
9056 * systemd-analyze gained a new verb "timespan" for parsing and
9057 normalizing time span values (i.e. strings like "5min 7s 8us").
9059 * systemd-analyze also gained a new verb "security" for analyzing the
9060 security and sand-boxing settings of services in order to determine an
9061 "exposure level" for them, indicating whether a service would benefit
9062 from more sand-boxing options turned on for them.
9064 * "systemd-analyze syscall-filter" will now also show system calls
9065 supported by the local kernel but not included in any of the defined
9068 * .nspawn files now understand the Ephemeral= setting, matching the
9069 --ephemeral command line switch.
9071 * sd-event gained the new APIs sd_event_source_get_floating() and
9072 sd_event_source_set_floating() for controlling whether a specific
9073 event source is "floating", i.e. destroyed along with the even loop
9076 * Unit objects on D-Bus gained a new "Refs" property that lists all
9077 clients that currently have a reference on the unit (to ensure it is
9080 * The JoinControllers= option in system.conf is no longer supported, as
9081 it didn't work correctly, is hard to support properly, is legacy (as
9082 the concept only exists on cgroup v1) and apparently wasn't used.
9084 * Journal messages that are generated whenever a unit enters the failed
9085 state are now tagged with a unique MESSAGE_ID. Similarly, messages
9086 generated whenever a service process exits are now made recognizable,
9087 too. A tagged message is also emitted whenever a unit enters the
9088 "dead" state on success.
9090 * systemd-run gained a new switch --working-directory= for configuring
9091 the working directory of the service to start. A shortcut -d is
9092 equivalent, setting the working directory of the service to the
9093 current working directory of the invoking program. The new --shell
9094 (or just -S) option has been added for invoking the $SHELL of the
9095 caller as a service, and implies --pty --same-dir --wait --collect
9096 --service-type=exec. Or in other words, "systemd-run -S" is now the
9097 quickest way to quickly get an interactive in a fully clean and
9098 well-defined system service context.
9100 * machinectl gained a new verb "import-fs" for importing an OS tree
9101 from a directory. Moreover, when a directory or tarball is imported
9102 and single top-level directory found with the OS itself below the OS
9103 tree is automatically mangled and moved one level up.
9105 * systemd-importd will no longer set up an implicit btrfs loop-back
9106 file system on /var/lib/machines. If one is already set up, it will
9107 continue to be used.
9109 * A new generator "systemd-run-generator" has been added. It will
9110 synthesize a unit from one or more program command lines included in
9111 the kernel command line. This is very useful in container managers
9114 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
9116 This will run "systemd-nspawn" on an image, invoke the specified
9117 command line and immediately shut down the container again, returning
9118 the command line's exit code.
9120 * The block device locking logic is now documented:
9122 https://systemd.io/BLOCK_DEVICE_LOCKING
9124 * loginctl and machinectl now optionally output the various tables in
9125 JSON using the --output= switch. It is our intention to add similar
9126 support to systemctl and all other commands.
9128 * udevadm's query and trigger verb now optionally take a .device unit
9131 * systemd-udevd's network naming logic now understands a new
9132 net.naming_scheme= kernel command line switch, which may be used to
9133 pick a specific version of the naming scheme. This helps stabilizing
9134 interface names even as systemd/udev are updated and the naming logic
9137 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
9138 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
9139 initialize one to all 0xFF.
9141 * After loading the SELinux policy systemd will now recursively relabel
9142 all files and directories listed in
9143 /run/systemd/relabel-extra.d/*.relabel (which should be simple
9144 newline separated lists of paths) in addition to the ones it already
9145 implicitly relabels in /run, /dev and /sys. After the relabelling is
9146 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
9147 removed. This is useful to permit initrds (i.e. code running before
9148 the SELinux policy is in effect) to generate files in the host
9149 filesystem safely and ensure that the correct label is applied during
9150 the transition to the host OS.
9152 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
9153 mknod() handling in user namespaces. Previously mknod() would always
9154 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
9155 but device nodes generated that way cannot be opened, and attempts to
9156 open them result in EPERM. This breaks the "graceful fallback" logic
9157 in systemd's PrivateDevices= sand-boxing option. This option is
9158 implemented defensively, so that when systemd detects it runs in a
9159 restricted environment (such as a user namespace, or an environment
9160 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
9161 where device nodes cannot be created the effect of PrivateDevices= is
9162 bypassed (following the logic that 2nd-level sand-boxing is not
9163 essential if the system systemd runs in is itself already sand-boxed
9164 as a whole). This logic breaks with 4.18 in container managers where
9165 user namespacing is used: suddenly PrivateDevices= succeeds setting
9166 up a private /dev/ file system containing devices nodes — but when
9167 these are opened they don't work.
9169 At this point it is recommended that container managers utilizing
9170 user namespaces that intend to run systemd in the payload explicitly
9171 block mknod() with seccomp or similar, so that the graceful fallback
9174 We are very sorry for the breakage and the requirement to change
9175 container configurations for newer kernels. It's purely caused by an
9176 incompatible kernel change. The relevant kernel developers have been
9177 notified about this userspace breakage quickly, but they chose to
9180 * PermissionsStartOnly= setting is deprecated (but is still supported
9181 for backwards compatibility). The same functionality is provided by
9182 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
9185 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
9186 pam_systemd anymore.
9188 * The naming scheme for network devices was changed to always rename
9189 devices, even if they were already renamed by userspace. The "kernel"
9190 policy was changed to only apply as a fallback, if no other naming
9193 * The requirements to build systemd is bumped to meson-0.46 and
9196 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
9197 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
9198 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
9199 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
9200 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
9201 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
9202 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
9203 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
9204 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
9205 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
9206 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
9207 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
9208 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
9209 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
9210 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
9211 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
9212 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
9213 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
9214 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
9215 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
9216 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
9217 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
9218 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
9219 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
9220 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
9221 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
9222 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
9223 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
9224 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
9225 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
9226 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
9227 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
9228 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
9229 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
9230 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
9231 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
9232 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
9233 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
9234 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
9235 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
9236 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
9237 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
9238 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
9239 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
9240 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
9242 — Warsaw, 2018-12-21
9246 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
9247 builtin will name network interfaces differently than in previous
9248 versions for virtual network interfaces created with SR-IOV and NPAR
9249 and for devices where the PCI network controller device does not have
9250 a slot number associated.
9252 SR-IOV virtual devices are now named based on the name of the parent
9253 interface, with a suffix of "v<N>", where <N> is the virtual device
9254 number. Previously those virtual devices were named as if completely
9257 The ninth and later NPAR virtual devices will be named following the
9258 scheme used for the first eight NPAR partitions. Previously those
9259 devices were not renamed and the kernel default (eth<n>) was used.
9261 "net_id" will also generate names for PCI devices where the PCI
9262 network controller device does not have an associated slot number
9263 itself, but one of its parents does. Previously those devices were
9264 not renamed and the kernel default (eth<n>) was used.
9266 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
9267 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
9268 the unit. So, it is expected that the default behavior of
9269 systemd-logind is not changed. However, if distribution packagers or
9270 administrators disabled or modified IPAddressDeny= setting by a
9271 drop-in config file, then it may be necessary to update the file to
9272 re-enable AF_INET and AF_INET6 to support network user name services,
9275 * When the RestrictNamespaces= unit property is specified multiple
9276 times, then the specified types are merged now. Previously, only the
9277 last assignment was used. So, if distribution packagers or
9278 administrators modified the setting by a drop-in config file, then it
9279 may be necessary to update the file.
9281 * When OnFailure= is used in combination with Restart= on a service
9282 unit, then the specified units will no longer be triggered on
9283 failures that result in restarting. Previously, the specified units
9284 would be activated each time the unit failed, even when the unit was
9285 going to be restarted automatically. This behaviour contradicted the
9286 documentation. With this release the code is adjusted to match the
9289 * systemd-tmpfiles will now print a notice whenever it encounters
9290 tmpfiles.d/ lines referencing the /var/run/ directory. It will
9291 recommend reworking them to use the /run/ directory instead (for
9292 which /var/run/ is simply a symlinked compatibility alias). This way
9293 systemd-tmpfiles can properly detect line conflicts and merge lines
9294 referencing the same file by two paths, without having to access
9297 * systemctl disable/unmask/preset/preset-all cannot be used with
9298 --runtime. Previously this was allowed, but resulted in unintuitive
9299 behaviour that wasn't useful. systemctl disable/unmask will now undo
9300 both runtime and persistent enablement/masking, i.e. it will remove
9301 any relevant symlinks both in /run and /etc.
9303 * Note that all long-running system services shipped with systemd will
9304 now default to a system call allow list (rather than a deny list, as
9305 before). In particular, systemd-udevd will now enforce one too. For
9306 most cases this should be safe, however downstream distributions
9307 which disabled sandboxing of systemd-udevd (specifically the
9308 MountFlags= setting), might want to disable this security feature
9309 too, as the default allow-listing will prohibit all mount, swap,
9310 reboot and clock changing operations from udev rules.
9312 * sd-boot acquired new loader configuration settings to optionally turn
9313 off Windows and MacOS boot partition discovery as well as
9314 reboot-into-firmware menu items. It is also able to pick a better
9315 screen resolution for HiDPI systems, and now provides loader
9316 configuration settings to change the resolution explicitly.
9318 * systemd-resolved now supports DNS-over-TLS. It's still
9319 turned off by default, use DNSOverTLS=opportunistic to turn it on in
9320 resolved.conf. We intend to make this the default as soon as couple
9321 of additional techniques for optimizing the initial latency caused by
9322 establishing a TLS/TCP connection are implemented.
9324 * systemd-resolved.service and systemd-networkd.service now set
9325 DynamicUser=yes. The users systemd-resolve and systemd-network are
9326 not created by systemd-sysusers anymore.
9328 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
9329 that embed a network facing module into any process using getpwuid()
9330 or related call: the dynamic allocation of the user ID for
9331 systemd-resolved.service means the service manager has to check NSS
9332 if the user name is already taken when forking off the service. Since
9333 the user in the common case won't be defined in /etc/passwd the
9334 lookup is likely to trigger nss-ldap which in turn might use NSS to
9335 ask systemd-resolved for hostname lookups. This will hence result in
9336 a deadlock: a user name lookup in order to start
9337 systemd-resolved.service will result in a hostname lookup for which
9338 systemd-resolved.service needs to be started already. There are
9339 multiple ways to work around this problem: pre-allocate the
9340 "systemd-resolve" user on such systems, so that nss-ldap won't be
9341 triggered; or use a different NSS package that doesn't do networking
9342 in-process but provides a local asynchronous name cache; or configure
9343 the NSS package to avoid lookups for UIDs in the range `pkg-config
9344 systemd --variable=dynamicuidmin` … `pkg-config systemd
9345 --variable=dynamicuidmax`, so that it does not consider itself
9346 authoritative for the same UID range systemd allocates dynamic users
9349 * The systemd-resolve tool has been renamed to resolvectl (it also
9350 remains available under the old name, for compatibility), and its
9351 interface is now verb-based, similar in style to the other <xyz>ctl
9352 tools, such as systemctl or loginctl.
9354 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
9355 compatibility. It may be symlinked under the 'resolvconf' name, in
9356 which case it will take arguments and input compatible with the
9357 Debian and FreeBSD resolvconf tool.
9359 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
9360 where the system initially suspends, and after a timeout resumes and
9363 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
9364 set the client will only send a DUID as client identifier. (EDIT: the
9365 option was broken, and was dropped in v255.)
9367 * The nss-systemd glibc NSS module will now enumerate dynamic users and
9368 groups in effect. Previously, it could resolve UIDs/GIDs to user
9369 names/groups and vice versa, but did not support enumeration.
9371 * journald's Compress= configuration setting now optionally accepts a
9372 byte threshold value. All journal objects larger than this threshold
9373 will be compressed, smaller ones will not. Previously this threshold
9374 was not configurable and set to 512.
9376 * A new system.conf setting NoNewPrivileges= is now available which may
9377 be used to turn off acquisition of new privileges system-wide
9378 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
9379 for all its children). Note that turning this option on means setuid
9380 binaries and file system capabilities lose their special powers.
9381 While turning on this option is a big step towards a more secure
9382 system, doing so is likely to break numerous pre-existing UNIX tools,
9383 in particular su and sudo.
9385 * A new service systemd-time-sync-wait.service has been added. If
9386 enabled it will delay the time-sync.target unit at boot until time
9387 synchronization has been received from the network. This
9388 functionality is useful on systems lacking a local RTC or where it is
9389 acceptable that the boot process shall be delayed by external network
9392 * When hibernating, systemd will now inform the kernel of the image
9393 write offset, on kernels new enough to support this. This means swap
9394 files should work for hibernation now.
9396 * When loading unit files, systemd will now look for drop-in unit files
9397 extensions in additional places. Previously, for a unit file name
9398 "foo-bar-baz.service" it would look for dropin files in
9399 "foo-bar-baz.service.d/*.conf". Now, it will also look in
9400 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
9401 service name truncated after all inner dashes. This scheme allows
9402 writing drop-ins easily that apply to a whole set of unit files at
9403 once. It's particularly useful for mount and slice units (as their
9404 naming is prefix based), but is also useful for service and other
9405 units, for packages that install multiple unit files at once,
9406 following a strict naming regime of beginning the unit file name with
9407 the package's name. Two new specifiers are now supported in unit
9408 files to match this: %j and %J are replaced by the part of the unit
9409 name following the last dash.
9411 * Unit files and other configuration files that support specifier
9412 expansion now understand another three new specifiers: %T and %V will
9413 resolve to /tmp and /var/tmp respectively, or whatever temporary
9414 directory has been set for the calling user. %E will expand to either
9415 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
9417 * The ExecStart= lines of unit files are no longer required to
9418 reference absolute paths. If non-absolute paths are specified the
9419 specified binary name is searched within the service manager's
9420 built-in $PATH, which may be queried with 'systemd-path
9421 search-binaries-default'. It's generally recommended to continue to
9422 use absolute paths for all binaries specified in unit files.
9424 * Units gained a new load state "bad-setting", which is used when a
9425 unit file was loaded, but contained fatal errors which prevent it
9426 from being started (for example, a service unit has been defined
9427 lacking both ExecStart= and ExecStop= lines).
9429 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
9430 support alternative debuggers, for example lldb. The old name
9431 continues to be available however, for compatibility reasons. Use the
9432 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
9433 to pick an alternative debugger instead of the default gdb.
9435 * systemctl and the other tools will now output escape sequences that
9436 generate proper clickable hyperlinks in various terminal emulators
9437 where useful (for example, in the "systemctl status" output you can
9438 now click on the unit file name to quickly open it in the
9439 editor/viewer of your choice). Note that not all terminal emulators
9440 support this functionality yet, but many do. Unfortunately, the
9441 "less" pager doesn't support this yet, hence this functionality is
9442 currently automatically turned off when a pager is started (which
9443 happens quite often due to auto-paging). We hope to remove this
9444 limitation as soon as "less" learns these escape sequences. This new
9445 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
9446 environment variable. For details on these escape sequences see:
9447 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
9449 * networkd's .network files now support a new IPv6MTUBytes= option for
9450 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
9451 option in the [Route] section to configure the MTU to use for
9452 specific routes. It also gained support for configuration of the DHCP
9453 "UserClass" option through the new UserClass= setting. It gained
9454 three new options in the new [CAN] section for configuring CAN
9455 networks. The MULTICAST and ALLMULTI interface flags may now be
9456 controlled explicitly with the new Multicast= and AllMulticast=
9459 * networkd will now automatically make use of the kernel's route
9460 expiration feature, if it is available.
9462 * udevd's .link files now support setting the number of receive and
9463 transmit channels, using the RxChannels=, TxChannels=,
9464 OtherChannels=, CombinedChannels= settings.
9466 * Support for UDPSegmentationOffload= has been removed, given its
9467 limited support in hardware, and waning software support.
9469 * networkd's .netdev files now support creating "netdevsim" interfaces.
9471 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
9472 to query the unit belonging to a specific kernel control group.
9474 * systemd-analyze gained a new verb "cat-config", which may be used to
9475 dump the contents of any configuration file, with all its matching
9476 drop-in files added in, and honouring the usual search and masking
9477 logic applied to systemd configuration files. For example use
9478 "systemd-analyze cat-config systemd/system.conf" to get the complete
9479 system configuration file of systemd how it would be loaded by PID 1
9480 itself. Similar to this, various tools such as systemd-tmpfiles or
9481 systemd-sysusers, gained a new option "--cat-config", which does the
9482 corresponding operation for their own configuration settings. For
9483 example, "systemd-tmpfiles --cat-config" will now output the full
9484 list of tmpfiles.d/ lines in place.
9486 * timedatectl gained three new verbs: "show" shows bus properties of
9487 systemd-timedated, "timesync-status" shows the current NTP
9488 synchronization state of systemd-timesyncd, and "show-timesync"
9489 shows bus properties of systemd-timesyncd.
9491 * systemd-timesyncd gained a bus interface on which it exposes details
9494 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
9495 understood by systemd-timedated. It takes a colon-separated list of
9496 unit names of NTP client services. The list is used by
9497 "timedatectl set-ntp".
9499 * systemd-nspawn gained a new --rlimit= switch for setting initial
9500 resource limits for the container payload. There's a new switch
9501 --hostname= to explicitly override the container's hostname. A new
9502 --no-new-privileges= switch may be used to control the
9503 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
9504 --oom-score-adjust= switch controls the OOM scoring adjustment value
9505 for the payload. The new --cpu-affinity= switch controls the CPU
9506 affinity of the container payload. The new --resolv-conf= switch
9507 allows more detailed control of /etc/resolv.conf handling of the
9508 container. Similarly, the new --timezone= switch allows more detailed
9509 control of /etc/localtime handling of the container.
9511 * systemd-detect-virt gained a new --list switch, which will print a
9512 list of all currently known VM and container environments.
9514 * Support for "Portable Services" has been added, see
9515 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
9516 experimental, but this is expected to change soon. Reflecting this
9517 experimental state, the "portablectl" binary is not installed into
9518 /usr/bin yet. The binary has to be called with the full path
9519 /usr/lib/systemd/portablectl instead.
9521 * journalctl's and systemctl's -o switch now knows a new log output
9522 mode "with-unit". The output it generates is very similar to the
9523 regular "short" mode, but displays the unit name instead of the
9524 syslog tag for each log line. Also, the date is shown with timezone
9525 information. This mode is probably more useful than the classic
9526 "short" output mode for most purposes, except where pixel-perfect
9527 compatibility with classic /var/log/messages formatting is required.
9529 * A new --dump-bus-properties switch has been added to the systemd
9530 binary, which may be used to dump all supported D-Bus properties.
9531 (Options which are still supported, but are deprecated, are *not*
9534 * sd-bus gained a set of new calls:
9535 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
9536 enable/disable the "floating" state of a bus slot object,
9537 i.e. whether the slot object pins the bus it is allocated for into
9538 memory or if the bus slot object gets disconnected when the bus goes
9539 away. sd_bus_open_with_description(),
9540 sd_bus_open_user_with_description(),
9541 sd_bus_open_system_with_description() may be used to allocate bus
9542 objects and set their description string already during allocation.
9544 * sd-event gained support for watching inotify events from the event
9545 loop, in an efficient way, sharing inotify handles between multiple
9546 users. For this a new function sd_event_add_inotify() has been added.
9548 * sd-event and sd-bus gained support for calling special user-supplied
9549 destructor functions for userdata pointers associated with
9550 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
9551 functions sd_bus_slot_set_destroy_callback,
9552 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
9553 sd_bus_track_get_destroy_callback,
9554 sd_event_source_set_destroy_callback,
9555 sd_event_source_get_destroy_callback have been added.
9557 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
9559 * PID 1 will now automatically reschedule .timer units whenever the
9560 local timezone changes. (They previously got rescheduled
9561 automatically when the system clock changed.)
9563 * New documentation has been added to document cgroups delegation,
9564 portable services and the various code quality tools we have set up:
9566 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
9567 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
9568 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
9570 * The Boot Loader Specification has been added to the source tree.
9572 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
9574 While moving it into our source tree we have updated it and further
9575 changes are now accepted through the usual github PR workflow.
9577 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
9578 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
9579 earlier PAM modules. The data in these fields is used to initialize
9580 the session scope's resource properties. Thus external PAM modules
9581 may now configure per-session limits, for example sourced from
9582 external user databases.
9584 * socket units with Accept=yes will now maintain a "refused" counter in
9585 addition to the existing "accepted" counter, counting connections
9586 refused due to the enforced limits.
9588 * The "systemd-path search-binaries-default" command may now be use to
9589 query the default, built-in $PATH PID 1 will pass to the services it
9592 * A new unit file setting PrivateMounts= has been added. It's a boolean
9593 option. If enabled the unit's processes are invoked in their own file
9594 system namespace. Note that this behaviour is also implied if any
9595 other file system namespacing options (such as PrivateTmp=,
9596 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
9597 primarily useful for services that do not use any of the other file
9598 system namespacing options. One such service is systemd-udevd.service
9599 where this is now used by default.
9601 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
9602 when the system is booted in UEFI "secure mode".
9604 * A new unit "system-update-pre.target" is added, which defines an
9605 optional synchronization point for offline system updates, as
9606 implemented by the pre-existing "system-update.target" unit. It
9607 allows ordering services before the service that executes the actual
9608 update process in a generic way.
9610 * Systemd now emits warnings whenever .include syntax is used.
9612 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
9613 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
9614 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
9615 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
9616 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
9617 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
9618 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
9619 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
9620 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
9621 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
9622 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
9623 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
9624 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
9625 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
9626 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
9627 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
9628 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
9629 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
9630 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
9631 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
9632 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
9633 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
9634 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
9635 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
9636 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
9637 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
9638 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
9639 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
9640 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9642 — Berlin, 2018-06-22
9646 * The MemoryAccounting= unit property now defaults to on. After
9647 discussions with the upstream control group maintainers we learnt
9648 that the negative impact of cgroup memory accounting on current
9649 kernels is finally relatively minimal, so that it should be safe to
9650 enable this by default without affecting system performance. Besides
9651 memory accounting only task accounting is turned on by default, all
9652 other forms of resource accounting (CPU, IO, IP) remain off for now,
9653 because it's not clear yet that their impact is small enough to move
9654 from opt-in to opt-out. We recommend downstreams to leave memory
9655 accounting on by default if kernel 4.14 or higher is primarily
9656 used. On very resource constrained systems or when support for old
9657 kernels is a necessity, -Dmemory-accounting-default=false can be used
9658 to revert this change.
9660 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
9661 %udev_rules_update) and the journal catalog (%journal_catalog_update)
9662 from the upgrade scriptlets of individual packages now do nothing.
9663 Transfiletriggers have been added which will perform those updates
9664 once at the end of the transaction.
9666 Similar transfiletriggers have been added to execute any sysctl.d
9667 and binfmt.d rules. Thus, it should be unnecessary to provide any
9668 scriptlets to execute this configuration from package installation
9671 * systemd-sysusers gained a mode where the configuration to execute is
9672 specified on the command line, but this configuration is not executed
9673 directly, but instead it is merged with the configuration on disk,
9674 and the result is executed. This is useful for package installation
9675 scripts which want to create the user before installing any files on
9676 disk (in case some of those files are owned by that user), while
9677 still allowing local admin overrides.
9679 This functionality is exposed to rpm scriptlets through a new
9680 %sysusers_create_package macro. Old %sysusers_create and
9681 %sysusers_create_inline macros are deprecated.
9683 A transfiletrigger for sysusers.d configuration is now installed,
9684 which means that it should be unnecessary to call systemd-sysusers from
9685 package installation scripts, unless the package installs any files
9686 owned by those newly-created users, in which case
9687 %sysusers_create_package should be used.
9689 * Analogous change has been done for systemd-tmpfiles: it gained a mode
9690 where the command-line configuration is merged with the configuration
9691 on disk. This is exposed as the new %tmpfiles_create_package macro,
9692 and %tmpfiles_create is deprecated. A transfiletrigger is installed
9693 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
9694 from package installation scripts.
9696 * sysusers.d configuration for a user may now also specify the group
9697 number, in addition to the user number ("u username 123:456"), or
9698 without the user number ("u username -:456").
9700 * Configution items for systemd-sysusers can now be specified as
9701 positional arguments when the new --inline switch is used.
9703 * The login shell of users created through sysusers.d may now be
9704 specified (previously, it was always /bin/sh for root and
9705 /sbin/nologin for other users).
9707 * systemd-analyze gained a new --global switch to look at global user
9708 configuration. It also gained a unit-paths verb to list the unit load
9709 paths that are compiled into systemd (which can be used with
9710 --systemd, --user, or --global).
9712 * udevadm trigger gained a new --settle/-w option to wait for any
9713 triggered events to finish (but just those, and not any other events
9714 which are triggered meanwhile).
9716 * The action that systemd-logind takes when the lid is closed and the
9717 machine is connected to external power can now be configured using
9718 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
9719 was determined by HandleLidSwitch=, and, for backwards compatibility,
9720 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
9722 * journalctl will periodically call sd_journal_process() to make it
9723 resilient against inotify queue overruns when journal files are
9724 rotated very quickly.
9726 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
9727 sd_bus_get_n_queued_write — may be used to check the number of
9728 pending bus messages.
9730 * systemd gained a new
9731 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
9732 which can be used to migrate foreign processes to scope and service
9733 units. The primary user for this new API is systemd itself: the
9734 systemd --user instance uses this call of the systemd --system
9735 instance to migrate processes if it itself gets the request to
9736 migrate processes and the kernel refuses this due to access
9737 restrictions. Thanks to this "systemd-run --scope --user …" works
9738 again in pure cgroup v2 environments when invoked from the user
9741 * A new TemporaryFileSystem= setting can be used to mask out part of
9742 the real file system tree with tmpfs mounts. This may be combined
9743 with BindPaths= and BindReadOnlyPaths= to hide files or directories
9744 not relevant to the unit, while still allowing some paths lower in
9745 the tree to be accessed.
9747 ProtectHome=tmpfs may now be used to hide user home and runtime
9748 directories from units, in a way that is mostly equivalent to
9749 "TemporaryFileSystem=/home /run/user /root".
9751 * Non-service units are now started with KeyringMode=shared by default.
9752 This means that mount and swapon and other mount tools have access
9753 to keys in the main keyring.
9755 * /sys/fs/bpf is now mounted automatically.
9757 * QNX virtualization is now detected by systemd-detect-virt and may
9758 be used in ConditionVirtualization=.
9760 * IPAccounting= may now be enabled also for slice units.
9762 * A new -Dsplit-bin= build configuration switch may be used to specify
9763 whether bin and sbin directories are merged, or if they should be
9764 included separately in $PATH and various listings of executable
9765 directories. The build configuration scripts will try to autodetect
9766 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
9767 system, but distributions are encouraged to configure this
9770 * A new -Dok-color= build configuration switch may be used to change
9771 the colour of "OK" status messages.
9773 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
9774 PrivateNetwork=yes was buggy in previous versions of systemd. This
9775 means that after the upgrade and daemon-reexec, any such units must
9778 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
9779 will not exclude read-only files owned by root from cleanup.
9781 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
9782 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
9783 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
9784 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
9785 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
9786 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
9787 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
9788 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
9789 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
9790 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
9791 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
9792 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
9793 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
9794 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
9795 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
9796 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
9798 — Warsaw, 2018-03-05
9802 * Some keyboards come with a zoom see-saw or rocker which until now got
9803 mapped to the Linux "zoomin/out" keys in hwdb. However, these
9804 keycodes are not recognized by any major desktop. They now produce
9805 Up/Down key events so that they can be used for scrolling.
9807 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
9808 slightly: previously, if an argument was specified for lines of this
9809 type (i.e. the right-most column was set) this string was appended to
9810 existing files each time systemd-tmpfiles was run. This behaviour was
9811 different from what the documentation said, and not particularly
9812 useful, as repeated systemd-tmpfiles invocations would not be
9813 idempotent and grow such files without bounds. With this release
9814 behaviour has been altered to match what the documentation says:
9815 lines of this type only have an effect if the indicated files don't
9816 exist yet, and only then the argument string is written to the file.
9818 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
9819 systemd-tmpfiles behaviour: previously, read-only files owned by root
9820 were always excluded from the file "aging" algorithm (i.e. the
9821 automatic clean-up of directories like /tmp based on
9822 atime/mtime/ctime). We intend to drop this restriction, and age files
9823 by default even when owned by root and read-only. This behaviour was
9824 inherited from older tools, but there have been requests to remove
9825 it, and it's not obvious why this restriction was made in the first
9826 place. Please speak up now, if you are aware of software that requires
9827 this behaviour, otherwise we'll remove the restriction in v238.
9829 * A new environment variable $SYSTEMD_OFFLINE is now understood by
9830 systemctl. It takes a boolean argument. If on, systemctl assumes it
9831 operates on an "offline" OS tree, and will not attempt to talk to the
9832 service manager. Previously, this mode was implicitly enabled if a
9833 chroot() environment was detected, and this new environment variable
9834 now provides explicit control.
9836 * .path and .socket units may now be created transiently, too.
9837 Previously only service, mount, automount and timer units were
9838 supported as transient units. The systemd-run tool has been updated
9839 to expose this new functionality, you may hence use it now to bind
9840 arbitrary commands to path or socket activation on-the-fly from the
9841 command line. Moreover, almost all properties are now exposed for the
9842 unit types that already supported transient operation.
9844 * The systemd-mount command gained support for a new --owner= parameter
9845 which takes a user name, which is then resolved and included in uid=
9846 and gid= mount options string of the file system to mount.
9848 * A new unit condition ConditionControlGroupController= has been added
9849 that checks whether a specific cgroup controller is available.
9851 * Unit files, udev's .link files, and systemd-networkd's .netdev and
9852 .network files all gained support for a new condition
9853 ConditionKernelVersion= for checking against specific kernel
9856 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
9857 support for configuring device flags in the Flags= setting. In the
9858 same files, the [Tunnel] section gained support for configuring
9859 AllowLocalRemote=. The [Route] section in .network files gained
9860 support for configuring InitialCongestionWindow=,
9861 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
9862 understands RapidCommit=.
9864 * systemd-networkd's DHCPv6 support gained support for Prefix
9867 * sd-bus gained support for a new "watch-bind" feature. When this
9868 feature is enabled, an sd_bus connection may be set up to connect to
9869 an AF_UNIX socket in the file system as soon as it is created. This
9870 functionality is useful for writing early-boot services that
9871 automatically connect to the system bus as soon as it is started,
9872 without ugly time-based polling. systemd-networkd and
9873 systemd-resolved have been updated to make use of this
9874 functionality. busctl exposes this functionality in a new
9875 --watch-bind= command line switch.
9877 * sd-bus will now optionally synthesize a local "Connected" signal as
9878 soon as a D-Bus connection is set up fully. This message mirrors the
9879 already existing "Disconnected" signal which is synthesized when the
9880 connection is terminated. This signal is generally useful but
9881 particularly handy in combination with the "watch-bind" feature
9882 described above. Synthesizing of this message has to be requested
9883 explicitly through the new API call sd_bus_set_connected_signal(). In
9884 addition a new call sd_bus_is_ready() has been added that checks
9885 whether a connection is fully set up (i.e. between the "Connected" and
9886 "Disconnected" signals).
9888 * sd-bus gained two new calls sd_bus_request_name_async() and
9889 sd_bus_release_name_async() for asynchronously registering bus
9890 names. Similar, there is now sd_bus_add_match_async() for installing
9891 a signal match asynchronously. All of systemd's own services have
9892 been updated to make use of these calls. Doing these operations
9893 asynchronously has two benefits: it reduces the risk of deadlocks in
9894 case of cyclic dependencies between bus services, and it speeds up
9895 service initialization since synchronization points for bus
9896 round-trips are removed.
9898 * sd-bus gained two new calls sd_bus_match_signal() and
9899 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
9900 and sd_bus_add_match_async() but instead of taking a D-Bus match
9901 string take match fields as normal function parameters.
9903 * sd-bus gained two new calls sd_bus_set_sender() and
9904 sd_bus_message_set_sender() for setting the sender name of outgoing
9905 messages (either for all outgoing messages or for just one specific
9906 one). These calls are only useful in direct connections as on
9907 brokered connections the broker fills in the sender anyway,
9908 overwriting whatever the client filled in.
9910 * sd-event gained a new pseudo-handle that may be specified on all API
9911 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
9912 used this refers to the default event loop object of the calling
9913 thread. Note however that this does not implicitly allocate one —
9914 which has to be done prior by using sd_event_default(). Similarly
9915 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
9916 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
9917 to the default bus of the specified type of the calling thread. Here
9918 too this does not implicitly allocate bus connection objects, this
9919 has to be done prior with sd_bus_default() and friends.
9921 * sd-event gained a new call pair
9922 sd_event_source_{get|set}_io_fd_own(). This may be used to request
9923 automatic closure of the file descriptor an IO event source watches
9924 when the event source is destroyed.
9926 * systemd-networkd gained support for natively configuring WireGuard
9929 * In previous versions systemd synthesized user records both for the
9930 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
9931 internally. In order to simplify distribution-wide renames of the
9932 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
9933 new transitional flag file has been added: if
9934 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
9935 user and group record within the systemd codebase is disabled.
9937 * systemd-notify gained a new --uid= option for selecting the source
9938 user/UID to use for notification messages sent to the service
9941 * journalctl gained a new --grep= option to list only entries in which
9942 the message matches a certain pattern. By default matching is case
9943 insensitive if the pattern is lowercase, and case sensitive
9944 otherwise. Option --case-sensitive=yes|no can be used to override
9945 this an specify case sensitivity or case insensitivity.
9947 * There's now a "systemd-analyze service-watchdogs" command for printing
9948 the current state of the service runtime watchdog, and optionally
9949 enabling or disabling the per-service watchdogs system-wide if given a
9950 boolean argument (i.e. the concept you configure in WatchdogSec=), for
9951 debugging purposes. There's also a kernel command line option
9952 systemd.service_watchdogs= for controlling the same.
9954 * Two new "log-level" and "log-target" options for systemd-analyze were
9955 added that merge the now deprecated get-log-level, set-log-level and
9956 get-log-target, set-log-target pairs. The deprecated options are still
9957 understood for backwards compatibility. The two new options print the
9958 current value when no arguments are given, and set them when a
9959 level/target is given as an argument.
9961 * sysusers.d's "u" lines now optionally accept both a UID and a GID
9962 specification, separated by a ":" character, in order to create users
9963 where UID and GID do not match.
9965 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
9966 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
9967 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
9968 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
9969 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
9970 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
9971 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
9972 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
9973 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
9974 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
9975 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
9976 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
9977 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
9978 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
9979 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
9980 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
9981 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
9982 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
9983 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
9984 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
9991 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
9992 in v235 has been extended to also set the dummy.ko module option
9993 numdummies=0, preventing the kernel from automatically creating
9994 dummy0. All dummy interfaces must now be explicitly created.
9996 * Unknown '%' specifiers in configuration files are now rejected. This
9997 applies to units and tmpfiles.d configuration. Any percent characters
9998 that are followed by a letter or digit that are not supposed to be
9999 interpreted as the beginning of a specifier should be escaped by
10000 doubling ("%%"). (So "size=5%" is still accepted, as well as
10001 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
10002 valid specifiers today.)
10004 * systemd-resolved now maintains a new dynamic
10005 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
10006 recommended to make /etc/resolv.conf a symlink to it. This file
10007 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
10008 includes dynamically acquired search domains, achieving more correct
10009 DNS resolution by software that bypasses local DNS APIs such as NSS.
10011 * The "uaccess" udev tag has been dropped from /dev/kvm and
10012 /dev/dri/renderD*. These devices now have the 0666 permissions by
10013 default (but this may be changed at build-time). /dev/dri/renderD*
10014 will now be owned by the "render" group along with /dev/kfd.
10016 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
10017 systemd-journal-gatewayd.service and
10018 systemd-journal-upload.service. This means "nss-systemd" must be
10019 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
10020 services are resolved properly.
10022 * In /etc/fstab two new mount options are now understood:
10023 x-systemd.makefs and x-systemd.growfs. The former has the effect that
10024 the configured file system is formatted before it is mounted, the
10025 latter that the file system is resized to the full block device size
10026 after it is mounted (i.e. if the file system is smaller than the
10027 partition it resides on, it's grown). This is similar to the fsck
10028 logic in /etc/fstab, and pulls in systemd-makefs@.service and
10029 systemd-growfs@.service as necessary, similar to
10030 systemd-fsck@.service. Resizing is currently only supported on ext4
10033 * In systemd-networkd, the IPv6 RA logic now optionally may announce
10034 DNS server and domain information.
10036 * Support for the LUKS2 on-disk format for encrypted partitions has
10037 been added. This requires libcryptsetup2 during compilation and
10040 * The systemd --user instance will now signal "readiness" when its
10041 basic.target unit has been reached, instead of when the run queue ran
10042 empty for the first time.
10044 * Tmpfiles.d with user configuration are now also supported.
10045 systemd-tmpfiles gained a new --user switch, and snippets placed in
10046 ~/.config/user-tmpfiles.d/ and corresponding directories will be
10047 executed by systemd-tmpfiles --user running in the new
10048 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
10049 running in the user session.
10051 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
10052 %S resolves to the top-level state directory (/var/lib for the system
10053 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
10054 top-level cache directory (/var/cache for the system instance,
10055 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
10056 logs directory (/var/log for the system instance,
10057 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
10058 existing %t specifier, that resolves to the top-level runtime
10059 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
10062 * journalctl learnt a new parameter --output-fields= for limiting the
10063 set of journal fields to output in verbose and JSON output modes.
10065 * systemd-timesyncd's configuration file gained a new option
10066 RootDistanceMaxSec= for setting the maximum root distance of servers
10067 it'll use, as well as the new options PollIntervalMinSec= and
10068 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
10070 * bootctl gained a new command "list" for listing all available boot
10071 menu items on systems that follow the boot loader specification.
10073 * systemctl gained a new --dry-run switch that shows what would be done
10074 instead of doing it, and is currently supported by the shutdown and
10077 * ConditionSecurity= can now detect the TOMOYO security module.
10079 * Unit file [Install] sections are now also respected in unit drop-in
10080 files. This is intended to be used by drop-ins under /usr/lib/.
10082 * systemd-firstboot may now also set the initial keyboard mapping.
10084 * Udev "changed" events for devices which are exposed as systemd
10085 .device units are now propagated to units specified in
10086 ReloadPropagatedFrom= as reload requests.
10088 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
10089 unit template name (i.e. a name in the form of 'foobar@.service',
10090 without the instance component between the '@' and - the '.'), then
10091 the escaped sysfs path of the device is automatically used as the
10094 * SystemCallFilter= in unit files has been extended so that an "errno"
10095 can be specified individually for each system call. Example:
10096 SystemCallFilter=~uname:EILSEQ.
10098 * The cgroup delegation logic has been substantially updated. Delegate=
10099 now optionally takes a list of controllers (instead of a boolean, as
10100 before), which lists the controllers to delegate at least.
10102 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
10104 * A new LogLevelMax= setting configures the maximum log level any
10105 process of the service may log at (i.e. anything with a lesser
10106 priority than what is specified is automatically dropped). A new
10107 LogExtraFields= setting allows configuration of additional journal
10108 fields to attach to all log records generated by any of the unit's
10111 * New StandardInputData= and StandardInputText= settings along with the
10112 new option StandardInput=data may be used to configure textual or
10113 binary data that shall be passed to the executed service process via
10114 standard input, encoded in-line in the unit file.
10116 * StandardInput=, StandardOutput= and StandardError= may now be used to
10117 connect stdin/stdout/stderr of executed processes directly with a
10118 file or AF_UNIX socket in the file system, using the new "file:" option.
10120 * A new unit file option CollectMode= has been added, that allows
10121 tweaking the garbage collection logic for units. It may be used to
10122 tell systemd to garbage collect units that have failed automatically
10123 (normally it only GCs units that exited successfully). systemd-run
10124 and systemd-mount expose this new functionality with a new -G option.
10126 * "machinectl bind" may now be used to bind mount non-directories
10127 (i.e. regularfiles, devices, fifos, sockets).
10129 * systemd-analyze gained a new verb "calendar" for validating and
10130 testing calendar time specifications to use for OnCalendar= in timer
10131 units. Besides validating the expression it will calculate the next
10132 time the specified expression would elapse.
10134 * In addition to the pre-existing FailureAction= unit file setting
10135 there's now SuccessAction=, for configuring a shutdown action to
10136 execute when a unit completes successfully. This is useful in
10137 particular inside containers that shall terminate after some workload
10138 has been completed. Also, both options are now supported for all unit
10139 types, not just services.
10141 * networkds's IP rule support gained two new options
10142 IncomingInterface= and OutgoingInterface= for configuring the incoming
10143 and outgoing interfaces of configured rules. systemd-networkd also
10144 gained support for "vxcan" network devices.
10146 * networkd gained a new setting RequiredForOnline=, taking a
10147 boolean. If set, systemd-wait-online will take it into consideration
10148 when determining that the system is up, otherwise it will ignore the
10149 interface for this purpose.
10151 * The sd_notify() protocol gained support for a new operation: with
10152 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
10153 store again, ahead of POLLHUP or POLLERR when they are removed
10156 * A new document doc/UIDS-GIDS.md has been added to the source tree,
10157 that documents the UID/GID range and assignment assumptions and
10158 requirements of systemd.
10160 * The watchdog device PID 1 will ping may now be configured through the
10161 WatchdogDevice= configuration file setting, or by setting the
10162 systemd.watchdog_service= kernel command line option.
10164 * systemd-resolved's gained support for registering DNS-SD services on
10165 the local network using MulticastDNS. Services may either be
10166 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
10167 the same dir below /run, /usr/lib), or through its D-Bus API.
10169 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
10170 extend the effective start, runtime, and stop time. The service must
10171 continue to send EXTEND_TIMEOUT_USEC within the period specified to
10172 prevent the service manager from making the service as timedout.
10174 * systemd-resolved's DNSSEC support gained support for RFC 8080
10175 (Ed25519 keys and signatures).
10177 * The systemd-resolve command line tool gained a new set of options
10178 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
10179 --set-nta= and --revert to configure per-interface DNS configuration
10180 dynamically during runtime. It's useful for pushing DNS information
10181 into systemd-resolved from DNS hook scripts that various interface
10182 managing software supports (such as pppd).
10184 * systemd-nspawn gained a new --network-namespace-path= command line
10185 option, which may be used to make a container join an existing
10186 network namespace, by specifying a path to a "netns" file.
10188 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
10189 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
10190 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
10191 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
10192 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
10193 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
10194 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
10195 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
10196 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
10197 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
10198 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
10199 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
10200 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
10201 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
10202 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
10203 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
10204 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
10205 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
10206 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
10207 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
10208 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
10209 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
10210 Jędrzejewski-Szmek, Zeal Jagannatha
10212 — Berlin, 2017-12-14
10216 * INCOMPATIBILITY: systemd-logind.service and other long-running
10217 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
10218 communication with the outside. This generally improves security of
10219 the system, and is in almost all cases a safe and good choice, as
10220 these services do not and should not provide any network-facing
10221 functionality. However, systemd-logind uses the glibc NSS API to
10222 query the user database. This creates problems on systems where NSS
10223 is set up to directly consult network services for user database
10224 lookups. In particular, this creates incompatibilities with the
10225 "nss-nis" module, which attempts to directly contact the NIS/YP
10226 network servers it is configured for, and will now consistently
10227 fail. In such cases, it is possible to turn off IP sandboxing for
10228 systemd-logind.service (set IPAddressDeny= in its [Service] section
10229 to the empty string, via a .d/ unit file drop-in). Downstream
10230 distributions might want to update their nss-nis packaging to include
10231 such a drop-in snippet, accordingly, to hide this incompatibility
10232 from the user. Another option is to make use of glibc's nscd service
10233 to proxy such network requests through a privilege-separated, minimal
10234 local caching daemon, or to switch to more modern technologies such
10235 sssd, whose NSS hook-ups generally do not involve direct network
10236 access. In general, we think it's definitely time to question the
10237 implementation choices of nss-nis, i.e. whether it's a good idea
10238 today to embed a network-facing loadable module into all local
10239 processes that need to query the user database, including the most
10240 trivial and benign ones, such as "ls". For more details about
10241 IPAddressDeny= see below.
10243 * A new modprobe.d drop-in is now shipped by default that sets the
10244 bonding module option max_bonds=0. This overrides the kernel default,
10245 to avoid conflicts and ambiguity as to whether or not bond0 should be
10246 managed by systemd-networkd or not. This resolves multiple issues
10247 with bond0 properties not being applied, when bond0 is configured
10248 with systemd-networkd. Distributors may choose to not package this,
10249 however in that case users will be prevented from correctly managing
10250 bond0 interface using systemd-networkd.
10252 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
10253 which print the logging level and target of the system manager. They
10254 complement the existing "set-log-level" and "set-log-target" verbs
10255 used to change those values.
10257 * journald.conf gained a new boolean setting ReadKMsg= which defaults
10258 to on. If turned off kernel log messages will not be read by
10259 systemd-journald or included in the logs. It also gained a new
10260 setting LineMax= for configuring the maximum line length in
10261 STDOUT/STDERR log streams. The new default for this value is 48K, up
10262 from the previous hardcoded 2048.
10264 * A new unit setting RuntimeDirectoryPreserve= has been added, which
10265 allows more detailed control of what to do with a runtime directory
10266 configured with RuntimeDirectory= (i.e. a directory below /run or
10267 $XDG_RUNTIME_DIR) after a unit is stopped.
10269 * The RuntimeDirectory= setting for units gained support for creating
10270 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
10271 one top-level directory.
10273 * Units gained new options StateDirectory=, CacheDirectory=,
10274 LogsDirectory= and ConfigurationDirectory= which are closely related
10275 to RuntimeDirectory= but manage per-service directories below
10276 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
10277 possible to write unit files which when activated automatically gain
10278 properly owned service specific directories in these locations, thus
10279 making unit files self-contained and increasing compatibility with
10280 stateless systems and factory reset where /etc or /var are
10281 unpopulated at boot. Matching these new settings there's also
10282 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
10283 ConfigurationDirectoryMode= for configuring the access mode of these
10284 directories. These settings are particularly useful in combination
10285 with DynamicUser=yes as they provide secure, properly-owned,
10286 writable, and stateful locations for storage, excluded from the
10287 sandbox that such services live in otherwise.
10289 * Automake support has been removed from this release. systemd is now
10292 * systemd-journald will now aggressively cache client metadata during
10293 runtime, speeding up log write performance under pressure. This comes
10294 at a small price though: as much of the metadata is read
10295 asynchronously from /proc/ (and isn't implicitly attached to log
10296 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
10297 metadata stored alongside a log entry might be slightly
10298 out-of-date. Previously it could only be slightly newer than the log
10299 message. The time window is small however, and given that the kernel
10300 is unlikely to be improved anytime soon in this regard, this appears
10303 * nss-myhostname/systemd-resolved will now by default synthesize an
10304 A/AAAA resource record for the "_gateway" hostname, pointing to the
10305 current default IP gateway. Previously it did that for the "gateway"
10306 name, hampering adoption, as some distributions wanted to leave that
10307 hostname open for local use. The old behaviour may still be
10308 requested at build time.
10310 * systemd-networkd's [Address] section in .network files gained a new
10311 Scope= setting for configuring the IP address scope. The [Network]
10312 section gained a new boolean setting ConfigureWithoutCarrier= that
10313 tells systemd-networkd to ignore link sensing when configuring the
10314 device. The [DHCP] section gained a new Anonymize= boolean option for
10315 turning on a number of options suggested in RFC 7844. A new
10316 [RoutingPolicyRule] section has been added for configuring the IP
10317 routing policy. The [Route] section has gained support for a new
10318 Type= setting which permits configuring
10319 blackhole/unreachable/prohibit routes.
10321 * The [VRF] section in .netdev files gained a new Table= setting for
10322 configuring the routing table to use. The [Tunnel] section gained a
10323 new Independent= boolean field for configuring tunnels independent of
10324 an underlying network interface. The [Bridge] section gained a new
10325 GroupForwardMask= option for configuration of propagation of link
10326 local frames between bridge ports.
10328 * The WakeOnLan= setting in .link files gained support for a number of
10329 new modes. A new TCP6SegmentationOffload= setting has been added for
10330 configuring TCP/IPv6 hardware segmentation offload.
10332 * The IPv6 RA sender implementation may now optionally send out RDNSS
10333 and RDNSSL records to supply DNS configuration to peers.
10335 * systemd-nspawn gained support for a new --system-call-filter= command
10336 line option for adding and removing entries in the default system
10337 call filter it applies. Moreover systemd-nspawn has been changed to
10338 implement a system call allow list instead of a deny list.
10340 * systemd-run gained support for a new --pipe command line option. If
10341 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
10342 are directly passed on to the activated transient service
10343 executable. This allows invoking arbitrary processes as systemd
10344 services (for example to take benefit of dependency management,
10345 accounting management, resource management or log management that is
10346 done automatically for services) — while still allowing them to be
10347 integrated in a classic UNIX shell pipeline.
10349 * When a service sends RELOAD=1 via sd_notify() and reload propagation
10350 using ReloadPropagationTo= is configured, a reload is now propagated
10351 to configured units. (Previously this was only done on explicitly
10352 requested reloads, using "systemctl reload" or an equivalent
10355 * For each service unit a restart counter is now kept: it is increased
10356 each time the service is restarted due to Restart=, and may be
10357 queried using "systemctl show -p NRestarts …".
10359 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
10360 @signal and @timer have been added, for usage with SystemCallFilter=
10361 in unit files and the new --system-call-filter= command line option
10362 of systemd-nspawn (see above).
10364 * ExecStart= lines in unit files gained two new modifiers: when a
10365 command line is prefixed with "!" the command will be executed as
10366 configured, except for the credentials applied by
10367 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
10368 "+", but does still apply namespacing options unlike "+". There's
10369 also "!!" now, which is mostly identical, but becomes a NOP on
10370 systems that support ambient capabilities. This is useful to write
10371 unit files that work with ambient capabilities where possible but
10372 automatically fall back to traditional privilege dropping mechanisms
10373 on systems where this is not supported.
10375 * ListenNetlink= settings in socket units now support RDMA netlink
10378 * A new unit file setting LockPersonality= has been added which permits
10379 locking down the chosen execution domain ("personality") of a service
10382 * A new special target "getty-pre.target" has been added, which is
10383 ordered before all text logins, and may be used to order services
10384 before textual logins acquire access to the console.
10386 * systemd will now attempt to load the virtio-rng.ko kernel module very
10387 early on if a VM environment supporting this is detected. This should
10388 improve entropy during early boot in virtualized environments.
10390 * A _netdev option is now supported in /etc/crypttab that operates in a
10391 similar way as the same option in /etc/fstab: it permits configuring
10392 encrypted devices that need to be ordered after the network is up.
10393 Following this logic, two new special targets
10394 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
10395 added that are to cryptsetup.target what remote-fs.target and
10396 remote-fs-pre.target are to local-fs.target.
10398 * Service units gained a new UnsetEnvironment= setting which permits
10399 unsetting specific environment variables for services that are
10400 normally passed to it (for example in order to mask out locale
10401 settings for specific services that can't deal with it).
10403 * Units acquired a new boolean option IPAccounting=. When turned on, IP
10404 traffic accounting (packet count as well as byte count) is done for
10405 the service, and shown as part of "systemctl status" or "systemd-run
10408 * Service units acquired two new options IPAddressAllow= and
10409 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
10410 for configuring a simple IP access control list for all sockets of
10411 the unit. These options are available also on .slice and .socket
10412 units, permitting flexible access list configuration for individual
10413 services as well as groups of services (as defined by a slice unit),
10414 including system-wide. Note that IP ACLs configured this way are
10415 enforced on every single IPv4 and IPv6 socket created by any process
10416 of the service unit, and apply to ingress as well as egress traffic.
10418 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
10419 structured log message is generated each time the unit is stopped,
10420 containing information about the consumed resources of this
10423 * A new setting KeyringMode= has been added to unit files, which may be
10424 used to control how the kernel keyring is set up for executed
10427 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
10428 "systemctl kexec" and "systemctl exit" are now always asynchronous in
10429 behaviour (that is: these commands return immediately after the
10430 operation was enqueued instead of waiting for the operation to
10431 complete). Previously, "systemctl poweroff" and "systemctl reboot"
10432 were asynchronous on systems using systemd-logind (i.e. almost
10433 always, and like they were on sysvinit), and the other three commands
10434 were unconditionally synchronous. With this release this is cleaned
10435 up, and callers will see the same asynchronous behaviour on all
10436 systems for all five operations.
10438 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
10441 * .timer units now accept calendar specifications in other timezones
10442 than UTC or the local timezone.
10444 * The tmpfiles snippet var.conf has been changed to create
10445 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
10446 the "utmp" group already, and it appears to be generally understood
10447 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
10448 databases. Previously this was implemented correctly for all these
10449 databases excepts btmp, which has been opened up like this now
10450 too. Note that while the other databases are world-readable
10451 (i.e. 0644), btmp is not and remains more restrictive.
10453 * The systemd-resolve tool gained a new --reset-server-features
10454 switch. When invoked like this systemd-resolved will forget
10455 everything it learnt about the features supported by the configured
10456 upstream DNS servers, and restarts the feature probing logic on the
10457 next resolver look-up for them at the highest feature level
10460 * The status dump systemd-resolved sends to the logs upon receiving
10461 SIGUSR1 now also includes information about all DNS servers it is
10462 configured to use, and the features levels it probed for them.
10464 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
10465 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
10466 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
10467 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
10468 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
10469 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
10470 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
10471 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
10472 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
10473 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
10474 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
10475 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
10476 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
10477 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
10478 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
10479 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
10480 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
10481 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
10482 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
10483 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
10485 — Berlin, 2017-10-06
10489 * Meson is now supported as build system in addition to Automake. It is
10490 our plan to remove Automake in one of our next releases, so that
10491 Meson becomes our exclusive build system. Hence, please start using
10492 the Meson build system in your downstream packaging. There's plenty
10493 of documentation around how to use Meson, the extremely brief
10496 ./autogen.sh && ./configure && make && sudo make install
10500 meson build && ninja -C build && sudo ninja -C build install
10502 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
10503 which permits configuring a timeout on the time a job is
10504 running. This is particularly useful for setting timeouts on jobs for
10507 * Unit files gained two new options ConditionUser= and ConditionGroup=
10508 for conditionalizing units based on the identity of the user/group
10509 running a systemd user instance.
10511 * systemd-networkd now understands a new FlowLabel= setting in the
10512 [VXLAN] section of .network files, as well as a Priority= in
10513 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
10514 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
10515 gained support for configuration of GENEVE links, and IPv6 address
10516 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
10518 * .link files now understand a new Port= setting.
10520 * systemd-networkd's DHCP support gained support for DHCP option 119
10521 (domain search list).
10523 * systemd-networkd gained support for serving IPv6 address ranges using
10524 the Router Advertisement protocol. The new .network configuration
10525 section [IPv6Prefix] may be used to configure the ranges to
10526 serve. This is implemented based on a new, minimal, native server
10527 implementation of RA.
10529 * journalctl's --output= switch gained support for a new parameter
10530 "short-iso-precise" for a mode where timestamps are shown as precise
10533 * systemd-udevd's "net_id" builtin may now generate stable network
10534 interface names from IBM PowerVM VIO devices as well as ACPI platform
10537 * MulticastDNS support in systemd-resolved may now be explicitly
10538 enabled/disabled using the new MulticastDNS= configuration file
10541 * systemd-resolved may now optionally use libidn2 instead of the libidn
10542 for processing internationalized domain names. Support for libidn2
10543 should be considered experimental and should not be enabled by
10546 * "machinectl pull-tar" and related call may now do verification of
10547 downloaded images using SUSE-style .sha256 checksum files in addition
10548 to the already existing support for validating using Ubuntu-style
10551 * sd-bus gained support for a new sd_bus_message_appendv() call which
10552 is va_list equivalent of sd_bus_message_append().
10554 * sd-boot gained support for validating images using SHIM/MOK.
10556 * The SMACK code learnt support for "onlycap".
10558 * systemd-mount --umount is now much smarter in figuring out how to
10559 properly unmount a device given its mount or device path.
10561 * The code to call libnss_dns as a fallback from libnss_resolve when
10562 the communication with systemd-resolved fails was removed. This
10563 fallback was redundant and interfered with the [!UNAVAIL=return]
10564 suffix. See nss-resolve(8) for the recommended configuration.
10566 * systemd-logind may now be restarted without losing state. It stores
10567 the file descriptors for devices it manages in the system manager
10568 using the FDSTORE= mechanism. Please note that further changes in
10569 other components may be required to make use of this (for example
10570 Xorg has code to listen for stops of systemd-logind and terminate
10571 itself when logind is stopped or restarted, in order to avoid using
10572 stale file descriptors for graphical devices, which is now
10573 counterproductive and must be reverted in order for restarts of
10574 systemd-logind to be safe. See
10575 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
10577 * All kernel-install plugins are called with the environment variable
10578 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
10579 /etc/machine-id. If the machine ID could not be determined,
10580 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
10581 anything in the entry directory (passed as the second argument) if
10582 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
10583 temporary directory is passed as the entry directory and removed
10584 after all the plugins exit.
10586 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
10587 will now use its value as the machine ID instead of the machine ID
10588 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
10589 /etc/machine-info and no machine ID is set in /etc/machine-id,
10590 kernel-install will try to store the current machine ID there as
10591 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
10592 will generate a new UUID, store it in /etc/machine-info as
10593 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
10595 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
10596 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
10597 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
10598 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
10599 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
10600 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
10601 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
10602 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
10603 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
10604 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
10605 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
10606 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
10607 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
10608 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
10609 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
10610 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
10611 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
10612 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
10613 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
10614 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
10615 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
10616 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
10617 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
10618 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
10619 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
10620 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
10621 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
10622 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
10623 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
10626 — Berlin, 2017-07-12
10630 * The "hybrid" control group mode has been modified to improve
10631 compatibility with "legacy" cgroups-v1 setups. Specifically, the
10632 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
10633 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
10634 cgroups-v1 hierarchy), the only externally visible change being that
10635 the cgroups-v2 hierarchy is also mounted, to
10636 /sys/fs/cgroup/unified. This should provide a large degree of
10637 compatibility with "legacy" cgroups-v1, while taking benefit of the
10638 better management capabilities of cgroups-v2.
10640 * The default control group setup mode may be selected both a boot-time
10641 via a set of kernel command line parameters (specifically:
10642 systemd.unified_cgroup_hierarchy= and
10643 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
10644 default selected on the configure command line
10645 (--with-default-hierarchy=). The upstream default is "hybrid"
10646 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
10647 this will change in a future systemd version to be "unified" (pure
10648 cgroups-v2 mode). The third option for the compile time option is
10649 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
10650 distributions to default to "hybrid" mode for release distributions,
10651 starting with v233. We recommend "unified" for development
10652 distributions (specifically: distributions such as Fedora's rawhide)
10653 as that's where things are headed in the long run. Use "legacy" for
10654 greatest stability and compatibility only.
10656 * Note one current limitation of "unified" and "hybrid" control group
10657 setup modes: the kernel currently does not permit the systemd --user
10658 instance (i.e. unprivileged code) to migrate processes between two
10659 disconnected cgroup subtrees, even if both are managed and owned by
10660 the user. This effectively means "systemd-run --user --scope" doesn't
10661 work when invoked from outside of any "systemd --user" service or
10662 scope. Specifically, it is not supported from session scopes. We are
10663 working on fixing this in a future systemd version. (See #3388 for
10664 further details about this.)
10666 * DBus policy files are now installed into /usr rather than /etc. Make
10667 sure your system has dbus >= 1.9.18 running before upgrading to this
10668 version, or override the install path with --with-dbuspolicydir= .
10670 * All python scripts shipped with systemd (specifically: the various
10671 tests written in Python) now require Python 3.
10673 * systemd unit tests can now run standalone (without the source or
10674 build directories), and can be installed into /usr/lib/systemd/tests/
10675 with 'make install-tests'.
10677 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
10678 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
10681 * Support for the %c, %r, %R specifiers in unit files has been
10682 removed. Specifiers are not supposed to be dependent on configuration
10683 in the unit file itself (so that they resolve the same regardless
10684 where used in the unit files), but these specifiers were influenced
10685 by the Slice= option.
10687 * The shell invoked by debug-shell.service now defaults to /bin/sh in
10688 all cases. If distributions want to use a different shell for this
10689 purpose (for example Fedora's /sbin/sushell) they need to specify
10690 this explicitly at configure time using --with-debug-shell=.
10692 * The confirmation spawn prompt has been reworked to offer the
10695 (c)ontinue, proceed without asking anymore
10696 (D)ump, show the state of the unit
10697 (f)ail, don't execute the command and pretend it failed
10699 (i)nfo, show a short summary of the unit
10700 (j)obs, show jobs that are in progress
10701 (s)kip, don't execute the command and pretend it succeeded
10702 (y)es, execute the command
10704 The 'n' choice for the confirmation spawn prompt has been removed,
10705 because its meaning was confusing.
10707 The prompt may now also be redirected to an alternative console by
10708 specifying the console as parameter to systemd.confirm_spawn=.
10710 * Services of Type=notify require a READY=1 notification to be sent
10711 during startup. If no such message is sent, the service now fails,
10712 even if the main process exited with a successful exit code.
10714 * Services that fail to start up correctly now always have their
10715 ExecStopPost= commands executed. Previously, they'd enter "failed"
10716 state directly, without executing these commands.
10718 * The option MulticastDNS= of network configuration files has acquired
10719 an actual implementation. With MulticastDNS=yes a host can resolve
10720 names of remote hosts and reply to mDNS A and AAAA requests.
10722 * When units are about to be started an additional check is now done to
10723 ensure that all dependencies of type BindsTo= (when used in
10724 combination with After=) have been started.
10726 * systemd-analyze gained a new verb "syscall-filter" which shows which
10727 system call groups are defined for the SystemCallFilter= unit file
10728 setting, and which system calls they contain.
10730 * A new system call filter group "@filesystem" has been added,
10731 consisting of various file system related system calls. Group
10732 "@reboot" has been added, covering reboot, kexec and shutdown related
10733 calls. Finally, group "@swap" has been added covering swap
10734 configuration related calls.
10736 * A new unit file option RestrictNamespaces= has been added that may be
10737 used to restrict access to the various process namespace types the
10738 Linux kernel provides. Specifically, it may be used to take away the
10739 right for a service unit to create additional file system, network,
10740 user, and other namespaces. This sandboxing option is particularly
10741 relevant due to the high amount of recently discovered namespacing
10742 related vulnerabilities in the kernel.
10744 * systemd-udev's .link files gained support for a new AutoNegotiation=
10745 setting for configuring Ethernet auto-negotiation.
10747 * systemd-networkd's .network files gained support for a new
10748 ListenPort= setting in the [DHCP] section to explicitly configure the
10749 UDP client port the DHCP client shall listen on.
10751 * .network files gained a new Unmanaged= boolean setting for explicitly
10752 excluding one or more interfaces from management by systemd-networkd.
10754 * The systemd-networkd ProxyARP= option has been renamed to
10755 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
10756 renamed to ReduceARPProxy=. The old names continue to be available
10759 * systemd-networkd gained support for configuring IPv6 Proxy NDP
10760 addresses via the new IPv6ProxyNDPAddress= .network file setting.
10762 * systemd-networkd's bonding device support gained support for two new
10763 configuration options ActiveSlave= and PrimarySlave=.
10765 * The various options in the [Match] section of .network files gained
10766 support for negative matching.
10768 * New systemd-specific mount options are now understood in /etc/fstab:
10770 x-systemd.mount-timeout= may be used to configure the maximum
10771 permitted runtime of the mount command.
10773 x-systemd.device-bound may be set to bind a mount point to its
10774 backing device unit, in order to automatically remove a mount point
10775 if its backing device is unplugged. This option may also be
10776 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
10777 on the block device, which is now automatically set for all CDROM
10778 drives, so that mounted CDs are automatically unmounted when they are
10779 removed from the drive.
10781 x-systemd.after= and x-systemd.before= may be used to explicitly
10782 order a mount after or before another unit or mount point.
10784 * Enqueued start jobs for device units are now automatically garbage
10785 collected if there are no jobs waiting for them anymore.
10787 * systemctl list-jobs gained two new switches: with --after, for every
10788 queued job the jobs it's waiting for are shown; with --before the
10789 jobs which it's blocking are shown.
10791 * systemd-nspawn gained support for ephemeral boots from disk images
10792 (or in other words: --ephemeral and --image= may now be
10793 combined). Moreover, ephemeral boots are now supported for normal
10794 directories, even if the backing file system is not btrfs. Of course,
10795 if the file system does not support file system snapshots or
10796 reflinks, the initial copy operation will be relatively expensive, but
10797 this should still be suitable for many use cases.
10799 * Calendar time specifications in .timer units now support
10800 specifications relative to the end of a month by using "~" instead of
10801 "-" as separator between month and day. For example, "*-02~03" means
10802 "the third last day in February". In addition a new syntax for
10803 repeated events has been added using the "/" character. For example,
10804 "9..17/2:00" means "every two hours from 9am to 5pm".
10806 * systemd-socket-proxyd gained a new parameter --connections-max= for
10807 configuring the maximum number of concurrent connections.
10809 * sd-id128 gained a new API for generating unique IDs for the host in a
10810 way that does not leak the machine ID. Specifically,
10811 sd_id128_get_machine_app_specific() derives an ID based on the
10812 machine ID in a well-defined, non-reversible, stable way. This is
10813 useful whenever an identifier for the host is needed but where the
10814 identifier shall not be useful to identify the system beyond the
10815 scope of the application itself. (Internally this uses HMAC-SHA256 as
10816 keyed hash function using the machine ID as input.)
10818 * NotifyAccess= gained a new supported value "exec". When set
10819 notifications are accepted from all processes systemd itself invoked,
10820 including all control processes.
10822 * .nspawn files gained support for defining overlay mounts using the
10823 Overlay= and OverlayReadOnly= options. Previously this functionality
10824 was only available on the systemd-nspawn command line.
10826 * systemd-nspawn's --bind= and --overlay= options gained support for
10827 bind/overlay mounts whose source lies within the container tree by
10828 prefixing the source path with "+".
10830 * systemd-nspawn's --bind= and --overlay= options gained support for
10831 automatically allocating a temporary source directory in /var/tmp
10832 that is removed when the container dies. Specifically, if the source
10833 directory is specified as empty string this mechanism is selected. An
10834 example usage is --overlay=+/var::/var, which creates an overlay
10835 mount based on the original /var contained in the image, overlaid
10836 with a temporary directory in the host's /var/tmp. This way changes
10837 to /var are automatically flushed when the container shuts down.
10839 * systemd-nspawn --image= option does now permit raw file system block
10840 devices (in addition to images containing partition tables, as
10843 * The disk image dissection logic in systemd-nspawn gained support for
10844 automatically setting up LUKS encrypted as well as Verity protected
10845 partitions. When a container is booted from an encrypted image the
10846 passphrase is queried at start-up time. When a container with Verity
10847 data is started, the root hash is search in a ".roothash" file
10848 accompanying the disk image (alternatively, pass the root hash via
10849 the new --root-hash= command line option).
10851 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
10852 be used to dissect disk images the same way as systemd-nspawn does
10853 it, following the Bootable Partition Specification. It may even be
10854 used to mount disk images with complex partition setups (including
10855 LUKS and Verity partitions) to a local host directory, in order to
10856 inspect them. This tool is not considered public API (yet), and is
10857 thus not installed into /usr/bin. Please do not rely on its
10858 existence, since it might go away or be changed in later systemd
10861 * A new generator "systemd-verity-generator" has been added, similar in
10862 style to "systemd-cryptsetup-generator", permitting automatic setup of
10863 Verity root partitions when systemd boots up. In order to make use of
10864 this your partition setup should follow the Discoverable Partitions
10865 Specification, and the GPT partition ID of the root file system
10866 partition should be identical to the upper 128-bit of the Verity root
10867 hash. The GPT partition ID of the Verity partition protecting it
10868 should be the lower 128-bit of the Verity root hash. If the partition
10869 image follows this model it is sufficient to specify a single
10870 "roothash=" kernel command line argument to both configure which root
10871 image and verity partition to use as well as the root hash for
10872 it. Note that systemd-nspawn's Verity support follows the same
10873 semantics, meaning that disk images with proper Verity data in place
10874 may be booted in containers with systemd-nspawn as well as on
10875 physical systems via the verity generator. Also note that the "mkosi"
10876 tool available at https://github.com/systemd/mkosi has been updated
10877 to generate Verity protected disk images following this scheme. In
10878 fact, it has been updated to generate disk images that optionally
10879 implement a complete UEFI SecureBoot trust chain, involving a signed
10880 kernel and initrd image that incorporates such a root hash as well as
10881 a Verity-enabled root partition.
10883 * The hardware database (hwdb) udev supports has been updated to carry
10884 accelerometer quirks.
10886 * All system services are now run with a fresh kernel keyring set up
10887 for them. The invocation ID is stored by default in it, thus
10888 providing a safe, non-overridable way to determine the invocation
10889 ID of each service.
10891 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
10892 options for bind mounting arbitrary paths in a service-specific
10893 way. When these options are used, arbitrary host or service files and
10894 directories may be mounted to arbitrary locations in the service's
10897 * Documentation has been added that lists all of systemd's low-level
10898 environment variables:
10900 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
10902 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
10903 whether a specific socket file descriptor matches a specified socket
10906 * systemd-firstboot has been updated to check for the
10907 systemd.firstboot= kernel command line option. It accepts a boolean
10908 and when set to false the first boot questions are skipped.
10910 * systemd-fstab-generator has been updated to check for the
10911 systemd.volatile= kernel command line option, which either takes an
10912 optional boolean parameter or the special value "state". If used the
10913 system may be booted in a "volatile" boot mode. Specifically,
10914 "systemd.volatile" is used, the root directory will be mounted as
10915 tmpfs, and only /usr is mounted from the actual root file system. If
10916 "systemd.volatile=state" is used, the root directory will be mounted
10917 as usual, but /var is mounted as tmpfs. This concept provides similar
10918 functionality as systemd-nspawn's --volatile= option, but provides it
10919 on physical boots. Use this option for implementing stateless
10920 systems, or testing systems with all state and/or configuration reset
10921 to the defaults. (Note though that many distributions are not
10922 prepared to boot up without a populated /etc or /var, though.)
10924 * systemd-gpt-auto-generator gained support for LUKS encrypted root
10925 partitions. Previously it only supported LUKS encrypted partitions
10926 for all other uses, except for the root partition itself.
10928 * Socket units gained support for listening on AF_VSOCK sockets for
10929 communication in virtualized QEMU environments.
10931 * The "configure" script gained a new option --with-fallback-hostname=
10932 for specifying the fallback hostname to use if none is configured in
10933 /etc/hostname. For example, by specifying
10934 --with-fallback-hostname=fedora it is possible to default to a
10935 hostname of "fedora" on pristine installations.
10937 * systemd-cgls gained support for a new --unit= switch for listing only
10938 the control groups of a specific unit. Similar --user-unit= has been
10939 added for listing only the control groups of a specific user unit.
10941 * systemd-mount gained a new --umount switch for unmounting a mount or
10942 automount point (and all mount/automount points below it).
10944 * systemd will now refuse full configuration reloads (via systemctl
10945 daemon-reload and related calls) unless at least 16MiB of free space
10946 are available in /run. This is a safety precaution in order to ensure
10947 that generators can safely operate after the reload completed.
10949 * A new unit file option RootImage= has been added, which has a similar
10950 effect as RootDirectory= but mounts the service's root directory from
10951 a disk image instead of plain directory. This logic reuses the same
10952 image dissection and mount logic that systemd-nspawn already uses,
10953 and hence supports any disk images systemd-nspawn supports, including
10954 those following the Discoverable Partition Specification, as well as
10955 Verity enabled images. This option enables systemd to run system
10956 services directly off disk images acting as resource bundles,
10957 possibly even including full integrity data.
10959 * A new MountAPIVFS= unit file option has been added, taking a boolean
10960 argument. If enabled /proc, /sys and /dev (collectively called the
10961 "API VFS") will be mounted for the service. This is only relevant if
10962 RootDirectory= or RootImage= is used for the service, as these mounts
10963 are of course in place in the host mount namespace anyway.
10965 * systemd-nspawn gained support for a new --pivot-root= switch. If
10966 specified the root directory within the container image is pivoted to
10967 the specified mount point, while the original root disk is moved to a
10968 different place. This option enables booting of ostree images
10969 directly with systemd-nspawn.
10971 * The systemd build scripts will no longer complain if the NTP server
10972 addresses are not changed from the defaults. Google now supports
10973 these NTP servers officially. We still recommend downstreams to
10974 properly register an NTP pool with the NTP pool project though.
10976 * coredumpctl gained a new "--reverse" option for printing the list
10977 of coredumps in reverse order.
10979 * coredumpctl will now show additional information about truncated and
10980 inaccessible coredumps, as well as coredumps that are still being
10981 processed. It also gained a new --quiet switch for suppressing
10982 additional informational message in its output.
10984 * coredumpctl gained support for only showing coredumps newer and/or
10985 older than specific timestamps, using the new --since= and --until=
10986 options, reminiscent of journalctl's options by the same name.
10988 * The systemd-coredump logic has been improved so that it may be reused
10989 to collect backtraces in non-compiled languages, for example in
10990 scripting languages such as Python.
10992 * machinectl will now show the UID shift of local containers, if user
10993 namespacing is enabled for them.
10995 * systemd will now optionally run "environment generator" binaries at
10996 configuration load time. They may be used to add environment
10997 variables to the environment block passed to services invoked. One
10998 user environment generator is shipped by default that sets up
10999 environment variables based on files dropped into /etc/environment.d
11000 and ~/.config/environment.d/.
11002 * systemd-resolved now includes the new, recently published 2017 DNSSEC
11005 * hostnamed has been updated to report a new chassis type of
11006 "convertible" to cover "foldable" laptops that can both act as a
11007 tablet and as a laptop, such as various Lenovo Yoga devices.
11009 Contributions from: Adrián López, Alexander Galanin, Alexander
11010 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
11011 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
11012 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
11013 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
11014 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
11015 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
11016 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
11017 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
11018 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
11019 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
11020 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
11021 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
11022 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
11023 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
11024 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
11025 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
11026 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
11027 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
11028 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
11029 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
11030 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
11031 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
11032 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
11033 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
11034 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
11035 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
11036 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
11039 — Berlin, 2017-03-01
11043 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
11044 RestrictAddressFamilies= enabled. These sandboxing options should
11045 generally be compatible with the various external udev call-out
11046 binaries we are aware of, however there may be exceptions, in
11047 particular when exotic languages for these call-outs are used. In
11048 this case, consider turning off these settings locally.
11050 * The new RemoveIPC= option can be used to remove IPC objects owned by
11051 the user or group of a service when that service exits.
11053 * The new ProtectKernelModules= option can be used to disable explicit
11054 load and unload operations of kernel modules by a service. In
11055 addition access to /usr/lib/modules is removed if this option is set.
11057 * ProtectSystem= option gained a new value "strict", which causes the
11058 whole file system tree with the exception of /dev, /proc, and /sys,
11059 to be remounted read-only for a service.
11061 * The new ProtectKernelTunables= option can be used to disable
11062 modification of configuration files in /sys and /proc by a service.
11063 Various directories and files are remounted read-only, so access is
11064 restricted even if the file permissions would allow it.
11066 * The new ProtectControlGroups= option can be used to disable write
11067 access by a service to /sys/fs/cgroup.
11069 * Various systemd services have been hardened with
11070 ProtectKernelTunables=yes, ProtectControlGroups=yes,
11071 RestrictAddressFamilies=.
11073 * Support for dynamically creating users for the lifetime of a service
11074 has been added. If DynamicUser=yes is specified, user and group IDs
11075 will be allocated from the range 61184…65519 for the lifetime of the
11076 service. They can be resolved using the new nss-systemd.so NSS
11077 module. The module must be enabled in /etc/nsswitch.conf. Services
11078 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
11079 any resources allocated by the service will be cleaned up when the
11080 service exits. They also have ProtectHome=read-only and
11081 ProtectSystem=strict enabled, so they are not able to make any
11082 permanent modifications to the system.
11084 * The nss-systemd module also always resolves root and nobody, making
11085 it possible to have no /etc/passwd or /etc/group files in minimal
11086 container or chroot environments.
11088 * Services may be started with their own user namespace using the new
11089 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
11090 under which the service is running are mapped. All other users are
11093 * Support for the cgroup namespace has been added to systemd-nspawn. If
11094 supported by kernel, the container system started by systemd-nspawn
11095 will have its own view of the cgroup hierarchy. This new behaviour
11096 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
11098 * The new MemorySwapMax= option can be used to limit the maximum swap
11099 usage under the unified cgroup hierarchy.
11101 * Support for the CPU controller in the unified cgroup hierarchy has
11102 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
11103 options. This controller requires out-of-tree patches for the kernel
11104 and the support is provisional.
11106 * Mount and automount units may now be created transiently
11107 (i.e. dynamically at runtime via the bus API, instead of requiring
11108 unit files in the file system).
11110 * systemd-mount is a new tool which may mount file systems – much like
11111 mount(8), optionally pulling in additional dependencies through
11112 transient .mount and .automount units. For example, this tool
11113 automatically runs fsck on a backing block device before mounting,
11114 and allows the automount logic to be used dynamically from the
11115 command line for establishing mount points. This tool is particularly
11116 useful when dealing with removable media, as it will ensure fsck is
11117 run – if necessary – before the first access and that the file system
11118 is quickly unmounted after each access by utilizing the automount
11119 logic. This maximizes the chance that the file system on the
11120 removable media stays in a clean state, and if it isn't in a clean
11121 state is fixed automatically.
11123 * LazyUnmount=yes option for mount units has been added to expose the
11124 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
11127 * /efi will be used as the mount point of the EFI boot partition, if
11128 the directory is present, and the mount point was not configured
11129 through other means (e.g. fstab). If /efi directory does not exist,
11130 /boot will be used as before. This makes it easier to automatically
11131 mount the EFI partition on systems where /boot is used for something
11134 * When operating on GPT disk images for containers, systemd-nspawn will
11135 now mount the ESP to /boot or /efi according to the same rules as PID
11136 1 running on a host. This allows tools like "bootctl" to operate
11137 correctly within such containers, in order to make container images
11138 bootable on physical systems.
11140 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
11142 * Two new user session targets have been added to support running
11143 graphical sessions under the systemd --user instance:
11144 graphical-session.target and graphical-session-pre.target. See
11145 systemd.special(7) for a description of how those targets should be
11148 * The vconsole initialization code has been significantly reworked to
11149 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
11150 support unicode keymaps. Font and keymap configuration will now be
11151 copied to all allocated virtual consoles.
11153 * FreeBSD's bhyve virtualization is now detected.
11155 * Information recorded in the journal for core dumps now includes the
11156 contents of /proc/mountinfo and the command line of the process at
11157 the top of the process hierarchy (which is usually the init process
11160 * systemd-journal-gatewayd learned the --directory= option to serve
11161 files from the specified location.
11163 * journalctl --root=… can be used to peruse the journal in the
11164 /var/log/ directories inside of a container tree. This is similar to
11165 the existing --machine= option, but does not require the container to
11168 * The hardware database has been extended to support
11169 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
11172 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
11173 specify the click rate for mice which include a horizontal wheel with
11174 a click rate that is different than the one for the vertical wheel.
11176 * systemd-run gained a new --wait option that makes service execution
11177 synchronous. (Specifically, the command will not return until the
11178 specified service binary exited.)
11180 * systemctl gained a new --wait option that causes the start command to
11181 wait until the units being started have terminated again.
11183 * A new journal output mode "short-full" has been added which displays
11184 timestamps with abbreviated English day names and adds a timezone
11185 suffix. Those timestamps include more information than the default
11186 "short" output mode, and can be passed directly to journalctl's
11187 --since= and --until= options.
11189 * /etc/resolv.conf will be bind-mounted into containers started by
11190 systemd-nspawn, if possible, so any changes to resolv.conf contents
11191 are automatically propagated to the container.
11193 * The number of instances for socket-activated services originating
11194 from a single IP address can be limited with
11195 MaxConnectionsPerSource=, extending the existing setting of
11198 * systemd-networkd gained support for vcan ("Virtual CAN") interface
11201 * .netdev and .network configuration can now be extended through
11204 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
11205 Segmentation Offload, Generic Receive Offload, Large Receive Offload
11206 can be enabled and disabled using the new UDPSegmentationOffload=,
11207 TCPSegmentationOffload=, GenericSegmentationOffload=,
11208 GenericReceiveOffload=, LargeReceiveOffload= options in the
11209 [Link] section of .link files.
11211 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
11212 Port VLAN ID can be configured for bridge devices using the new STP=,
11213 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
11214 section of .netdev files.
11216 * The route table to which routes received over DHCP or RA should be
11217 added can be configured with the new RouteTable= option in the [DHCP]
11218 and [IPv6AcceptRA] sections of .network files.
11220 * The Address Resolution Protocol can be disabled on links managed by
11221 systemd-networkd using the ARP=no setting in the [Link] section of
11224 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
11225 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
11226 encode information about the result and exit codes of the current
11227 service runtime cycle.
11229 * systemd-sysctl will now configure kernel parameters in the order
11230 they occur in the configuration files. This matches what sysctl
11231 has been traditionally doing.
11233 * kernel-install "plugins" that are executed to perform various
11234 tasks after a new kernel is added and before an old one is removed
11235 can now return a special value to terminate the procedure and
11236 prevent any later plugins from running.
11238 * Journald's SplitMode=login setting has been deprecated. It has been
11239 removed from documentation, and its use is discouraged. In a future
11240 release it will be completely removed, and made equivalent to current
11241 default of SplitMode=uid.
11243 * Storage=both option setting in /etc/systemd/coredump.conf has been
11244 removed. With fast LZ4 compression storing the core dump twice is not
11247 * The --share-system systemd-nspawn option has been replaced with an
11248 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
11249 this functionality is discouraged. In addition the variables
11250 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
11251 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
11252 individual namespaces.
11254 * "machinectl list" now shows the IP address of running containers in
11255 the output, as well as OS release information.
11257 * "loginctl list" now shows the TTY of each session in the output.
11259 * sd-bus gained new API calls sd_bus_track_set_recursive(),
11260 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
11261 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
11262 tracking objects in a "recursive" mode, where a single client can be
11263 counted multiple times, if it takes multiple references.
11265 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
11266 sd_bus_get_exit_on_disconnect(). They may be used to make a
11267 process using sd-bus automatically exit if the bus connection is
11270 * Bus clients of the service manager may now "pin" loaded units into
11271 memory, by taking an explicit reference on them. This is useful to
11272 ensure the client can retrieve runtime data about the service even
11273 after the service completed execution. Taking such a reference is
11274 available only for privileged clients and should be helpful to watch
11275 running services in a race-free manner, and in particular collect
11276 information about exit statuses and results.
11278 * The nss-resolve module has been changed to strictly return UNAVAIL
11279 when communication via D-Bus with resolved failed, and NOTFOUND when
11280 a lookup completed but was negative. This means it is now possible to
11281 neatly configure fallbacks using nsswitch.conf result checking
11282 expressions. Taking benefit of this, the new recommended
11283 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
11285 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
11287 * A new setting CtrlAltDelBurstAction= has been added to
11288 /etc/systemd/system.conf which may be used to configure the precise
11289 behaviour if the user on the console presses Ctrl-Alt-Del more often
11290 than 7 times in 2s. Previously this would unconditionally result in
11291 an expedited, immediate reboot. With this new setting the precise
11292 operation may be configured in more detail, and also turned off
11295 * In .netdev files two new settings RemoteChecksumTx= and
11296 RemoteChecksumRx= are now understood that permit configuring the
11297 remote checksumming logic for VXLAN networks.
11299 * The service manager learnt a new "invocation ID" concept for invoked
11300 services. Each runtime cycle of a service will get a new invocation
11301 ID (a 128-bit random UUID) assigned that identifies the current
11302 run of the service uniquely and globally. A new invocation ID
11303 is generated each time a service starts up. The journal will store
11304 the invocation ID of a service along with any logged messages, thus
11305 making the invocation ID useful for matching the online runtime of a
11306 service with the offline log data it generated in a safe way without
11307 relying on synchronized timestamps. In many ways this new service
11308 invocation ID concept is similar to the kernel's boot ID concept that
11309 uniquely and globally identifies the runtime of each boot. The
11310 invocation ID of a service is passed to the service itself via an
11311 environment variable ($INVOCATION_ID). A new bus call
11312 GetUnitByInvocationID() has been added that is similar to GetUnit()
11313 but instead of retrieving the bus path for a unit by its name
11314 retrieves it by its invocation ID. The returned path is valid only as
11315 long as the passed invocation ID is current.
11317 * systemd-resolved gained a new "DNSStubListener" setting in
11318 resolved.conf. It either takes a boolean value or the special values
11319 "udp" and "tcp", and configures whether to enable the stub DNS
11320 listener on 127.0.0.53:53.
11322 * IP addresses configured via networkd may now carry additional
11323 configuration settings supported by the kernel. New options include:
11324 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
11325 PrefixRoute=, AutoJoin=.
11327 * The PAM configuration fragment file for "user@.service" shipped with
11328 systemd (i.e. the --user instance of systemd) has been stripped to
11329 the minimum necessary to make the system boot. Previously, it
11330 contained Fedora-specific stanzas that did not apply to other
11331 distributions. It is expected that downstream distributions add
11332 additional configuration lines, matching their needs to this file,
11333 using it only as rough template of what systemd itself needs. Note
11334 that this reduced fragment does not even include an invocation of
11335 pam_limits which most distributions probably want to add, even though
11336 systemd itself does not need it. (There's also the new build time
11337 option --with-pamconfdir=no to disable installation of the PAM
11338 fragment entirely.)
11340 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
11341 capability is now also dropped from its set (in addition to
11342 CAP_SYS_MKNOD as before).
11344 * In service unit files it is now possible to connect a specific named
11345 file descriptor with stdin/stdout/stdout of an executed service. The
11346 name may be specified in matching .socket units using the
11347 FileDescriptorName= setting.
11349 * A number of journal settings may now be configured on the kernel
11350 command line. Specifically, the following options are now understood:
11351 systemd.journald.max_level_console=,
11352 systemd.journald.max_level_store=,
11353 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
11354 systemd.journald.max_level_wall=.
11356 * "systemctl is-enabled --full" will now show by which symlinks a unit
11357 file is enabled in the unit dependency tree.
11359 * Support for VeraCrypt encrypted partitions has been added to the
11360 "cryptsetup" logic and /etc/crypttab.
11362 * systemd-detect-virt gained support for a new --private-users switch
11363 that checks whether the invoking processes are running inside a user
11364 namespace. Similar, a new special value "private-users" for the
11365 existing ConditionVirtualization= setting has been added, permitting
11366 skipping of specific units in user namespace environments.
11368 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
11369 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
11370 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
11371 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
11372 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
11373 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
11374 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
11375 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
11376 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
11377 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
11378 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
11379 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
11380 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
11381 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
11382 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
11383 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
11384 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
11385 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
11386 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
11387 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
11388 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
11389 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
11390 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
11391 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
11392 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
11393 Jędrzejewski-Szmek, Zeal Jagannatha
11395 — Santa Fe, 2016-11-03
11399 * In service units the various ExecXYZ= settings have been extended
11400 with an additional special character as first argument of the
11401 assigned value: if the character '+' is used the specified command
11402 line it will be run with full privileges, regardless of User=,
11403 Group=, CapabilityBoundingSet= and similar options. The effect is
11404 similar to the existing PermissionsStartOnly= option, but allows
11405 configuration of this concept for each executed command line
11408 * Services may now alter the service watchdog timeout at runtime by
11409 sending a WATCHDOG_USEC= message via sd_notify().
11411 * MemoryLimit= and related unit settings now optionally take percentage
11412 specifications. The percentage is taken relative to the amount of
11413 physical memory in the system (or in case of containers, the assigned
11414 amount of memory). This allows scaling service resources neatly with
11415 the amount of RAM available on the system. Similarly, systemd-logind's
11416 RuntimeDirectorySize= option now also optionally takes percentage
11419 * In similar fashion TasksMax= takes percentage values now, too. The
11420 value is taken relative to the configured maximum number of processes
11421 on the system. The per-service task maximum has been changed to 15%
11422 using this functionality. (Effectively this is an increase of 512 →
11423 4915 for service units, given the kernel's default pid_max setting.)
11425 * Calendar time specifications in .timer units now understand a ".."
11426 syntax for time ranges. Example: "4..7:10" may now be used for
11427 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
11430 * The InaccessableDirectories=, ReadOnlyDirectories= and
11431 ReadWriteDirectories= unit file settings have been renamed to
11432 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
11433 applied to all kinds of file nodes, and not just directories, with
11434 the exception of symlinks. Specifically these settings may now be
11435 used on block and character device nodes, UNIX sockets and FIFOS as
11436 well as regular files. The old names of these settings remain
11437 available for compatibility.
11439 * systemd will now log about all service processes it kills forcibly
11440 (using SIGKILL) because they remained after the clean shutdown phase
11441 of the service completed. This should help identifying services that
11442 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
11443 systemd-logind's configuration a similar log message is generated for
11444 processes killed at the end of each session due to this setting.
11446 * systemd will now set the $JOURNAL_STREAM environment variable for all
11447 services whose stdout/stderr are connected to the Journal (which
11448 effectively means by default: all services). The variable contains
11449 the device and inode number of the file descriptor used for
11450 stdout/stderr. This may be used by invoked programs to detect whether
11451 their stdout/stderr is connected to the Journal, in which case they
11452 can switch over to direct Journal communication, thus being able to
11453 pass extended, structured metadata along with their log messages. As
11454 one example, this is now used by glib's logging primitives.
11456 * When using systemd's default tmp.mount unit for /tmp, the mount point
11457 will now be established with the "nosuid" and "nodev" options. This
11458 avoids privilege escalation attacks that put traps and exploits into
11459 /tmp. However, this might cause problems if you e.g. put container
11460 images or overlays into /tmp; if you need this, override tmp.mount's
11461 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
11464 * systemd now supports the "memory" cgroup controller also on
11467 * The systemd-cgtop tool now optionally takes a control group path as
11468 command line argument. If specified, the control group list shown is
11469 limited to subgroups of that group.
11471 * The SystemCallFilter= unit file setting gained support for
11472 pre-defined, named system call filter sets. For example
11473 SystemCallFilter=@clock is now an effective way to make all clock
11474 changing-related system calls unavailable to a service. A number of
11475 similar pre-defined groups are defined. Writing system call filters
11476 for system services is simplified substantially with this new
11477 concept. Accordingly, all of systemd's own, long-running services now
11478 enable system call filtering based on this, by default.
11480 * A new service setting MemoryDenyWriteExecute= has been added, taking
11481 a boolean value. If turned on, a service may no longer create memory
11482 mappings that are writable and executable at the same time. This
11483 enhances security for services where this is enabled as it becomes
11484 harder to dynamically write and then execute memory in exploited
11485 service processes. This option has been enabled for all of systemd's
11486 own long-running services.
11488 * A new RestrictRealtime= service setting has been added, taking a
11489 boolean argument. If set the service's processes may no longer
11490 acquire realtime scheduling. This improves security as realtime
11491 scheduling may otherwise be used to easily freeze the system.
11493 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
11494 value. This may be used for requesting that the system manager inside
11495 of the container reports start-up completion to nspawn which then
11496 propagates this notification further to the service manager
11497 supervising nspawn itself. A related option NotifyReady= in .nspawn
11498 files has been added too. This functionality allows ordering of the
11499 start-up of multiple containers using the usual systemd ordering
11502 * machinectl gained a new command "stop" that is an alias for
11505 * systemd-resolved gained support for contacting DNS servers on
11506 link-local IPv6 addresses.
11508 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
11509 its caches. A method call for requesting the same operation has been
11510 added to the bus API too, and is made available via "systemd-resolve
11513 * systemd-resolve gained a new --status switch. If passed a brief
11514 summary of the used DNS configuration with per-interface information
11517 * resolved.conf gained a new Cache= boolean option, defaulting to
11518 on. If turned off local DNS caching is disabled. This comes with a
11519 performance penalty in particular when DNSSEC is enabled. Note that
11520 resolved disables its internal caching implicitly anyway, when the
11521 configured DNS server is on a host-local IP address such as ::1 or
11522 127.0.0.1, thus automatically avoiding double local caching.
11524 * systemd-resolved now listens on the local IP address 127.0.0.53:53
11525 for DNS requests. This improves compatibility with local programs
11526 that do not use the libc NSS or systemd-resolved's bus APIs for name
11527 resolution. This minimal DNS service is only available to local
11528 programs and does not implement the full DNS protocol, but enough to
11529 cover local DNS clients. A new, static resolv.conf file, listing just
11530 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
11531 now recommended to make /etc/resolv.conf a symlink to this file in
11532 order to route all DNS lookups to systemd-resolved, regardless if
11533 done via NSS, the bus API or raw DNS packets. Note that this local
11534 DNS service is not as fully featured as the libc NSS or
11535 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
11536 used to deliver link-local address information (as this implies
11537 sending a local interface index along), LLMNR/mDNS support via this
11538 interface is severely restricted. It is thus strongly recommended for
11539 all applications to use the libc NSS API or native systemd-resolved
11542 * systemd-networkd's bridge support learned a new setting
11543 VLANFiltering= for controlling VLAN filtering. Moreover a new section
11544 in .network files has been added for configuring VLAN bridging in
11545 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
11547 * systemd-networkd's IPv6 Router Advertisement code now makes use of
11548 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
11549 now be acquired without relying on DHCPv6. Two new options
11550 UseDomains= and UseDNS= have been added to configure this behaviour.
11552 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
11553 renamed IPv6AcceptRA=, without altering its behaviour. The old
11554 setting name remains available for compatibility reasons.
11556 * The systemd-networkd VTI/VTI6 tunneling support gained new options
11557 Key=, InputKey= and OutputKey=.
11559 * systemd-networkd gained support for VRF ("Virtual Routing Function")
11560 interface configuration.
11562 * "systemctl edit" may now be used to create new unit files by
11563 specifying the --force switch.
11565 * sd-event gained a new function sd_event_get_iteration() for
11566 requesting the current iteration counter of the event loop. It starts
11567 at zero and is increased by one with each event loop iteration.
11569 * A new rpm macro %systemd_ordering is provided by the macros.systemd
11570 file. It can be used in lieu of %systemd_requires in packages which
11571 don't use any systemd functionality and are intended to be installed
11572 in minimal containers without systemd present. This macro provides
11573 ordering dependencies to ensure that if the package is installed in
11574 the same rpm transaction as systemd, systemd will be installed before
11575 the scriptlets for the package are executed, allowing unit presets
11578 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
11579 been added to simplify packaging of generators.
11581 * The os-release file gained VERSION_CODENAME field for the
11582 distribution nickname (e.g. VERSION_CODENAME=woody).
11584 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
11585 can be set to disable parsing of metadata and the creation
11586 of persistent symlinks for that device.
11588 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
11589 to make them available to logged-in users has been reverted.
11591 * Much of the common code of the various systemd components is now
11592 built into an internal shared library libsystemd-shared-231.so
11593 (incorporating the systemd version number in the name, to be updated
11594 with future releases) that the components link to. This should
11595 decrease systemd footprint both in memory during runtime and on
11596 disk. Note that the shared library is not for public use, and is
11597 neither API nor ABI stable, but is likely to change with every new
11598 released update. Packagers need to make sure that binaries
11599 linking to libsystemd-shared.so are updated in step with the
11602 * Configuration for "mkosi" is now part of the systemd
11603 repository. mkosi is a tool to easily build legacy-free OS images,
11604 and is available on github: https://github.com/systemd/mkosi. If
11605 "mkosi" is invoked in the build tree a new raw OS image is generated
11606 incorporating the systemd sources currently being worked on and a
11607 clean, fresh distribution installation. The generated OS image may be
11608 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
11609 UEFI PC. This functionality is particularly useful to easily test
11610 local changes made to systemd in a pristine, defined environment. See
11611 doc/HACKING for details.
11613 * configure learned the --with-support-url= option to specify the
11614 distribution's bugtracker.
11616 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
11617 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
11618 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
11619 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
11620 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
11621 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
11622 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
11623 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
11624 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
11625 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
11626 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
11627 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
11628 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
11629 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
11630 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
11631 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
11632 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
11633 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
11634 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
11636 — Berlin, 2016-07-25
11640 * DNSSEC is now turned on by default in systemd-resolved (in
11641 "allow-downgrade" mode), but may be turned off during compile time by
11642 passing "--with-default-dnssec=no" to "configure" (and of course,
11643 during runtime with DNSSEC= in resolved.conf). We recommend
11644 downstreams to leave this on at least during development cycles and
11645 report any issues with the DNSSEC logic upstream. We are very
11646 interested in collecting feedback about the DNSSEC validator and its
11647 limitations in the wild. Note however, that DNSSEC support is
11648 probably nothing downstreams should turn on in stable distros just
11649 yet, as it might create incompatibilities with a few DNS servers and
11650 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
11651 automatically whenever we detect such incompatible setups, but there
11652 might be systems we do not cover yet. Hence: please help us testing
11653 the DNSSEC code, leave this on where you can, report back, but then
11654 again don't consider turning this on in your stable, LTS or
11655 production release just yet. (Note that you have to enable
11656 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
11657 and its DNSSEC mode for hostname resolution from local
11660 * systemd-resolve conveniently resolves DANE records with the --tlsa
11661 option and OPENPGPKEY records with the --openpgp option. It also
11662 supports dumping raw DNS record data via the new --raw= switch.
11664 * systemd-logind will now by default terminate user processes that are
11665 part of the user session scope unit (session-XX.scope) when the user
11666 logs out. This behavior is controlled by the KillUserProcesses=
11667 setting in logind.conf, and the previous default of "no" is now
11668 changed to "yes". This means that user sessions will be properly
11669 cleaned up after, but additional steps are necessary to allow
11670 intentionally long-running processes to survive logout.
11672 While the user is logged in at least once, user@.service is running,
11673 and any service that should survive the end of any individual login
11674 session can be started at a user service or scope using systemd-run.
11675 systemd-run(1) man page has been extended with an example which shows
11676 how to run screen in a scope unit underneath user@.service. The same
11677 command works for tmux.
11679 After the user logs out of all sessions, user@.service will be
11680 terminated too, by default, unless the user has "lingering" enabled.
11681 To effectively allow users to run long-term tasks even if they are
11682 logged out, lingering must be enabled for them. See loginctl(1) for
11683 details. The default polkit policy was modified to allow users to
11684 set lingering for themselves without authentication.
11686 Previous defaults can be restored at compile time by the
11687 --without-kill-user-processes option to "configure".
11689 * systemd-logind gained new configuration settings SessionsMax= and
11690 InhibitorsMax=, both with a default of 8192. It will not register new
11691 user sessions or inhibitors above this limit.
11693 * systemd-logind will now reload configuration on SIGHUP.
11695 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
11696 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
11697 enable. Also, support for the "io" cgroup controller in the unified
11698 hierarchy has been added, so that the "memory", "pids" and "io" are
11699 now the controllers that are supported on the unified hierarchy.
11701 WARNING: it is not possible to use previous systemd versions with
11702 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
11703 is necessary to also update systemd in the initramfs if using the
11704 unified hierarchy. An updated SELinux policy is also required.
11706 * LLDP support has been extended, and both passive (receive-only) and
11707 active (sender) modes are supported. Passive mode ("routers-only") is
11708 enabled by default in systemd-networkd. Active LLDP mode is enabled
11709 by default for containers on the internal network. The "networkctl
11710 lldp" command may be used to list information gathered. "networkctl
11711 status" will also show basic LLDP information on connected peers now.
11713 * The IAID and DUID unique identifier sent in DHCP requests may now be
11714 configured for the system and each .network file managed by
11715 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
11717 * systemd-networkd gained support for configuring proxy ARP support for
11718 each interface, via the ProxyArp= setting in .network files. It also
11719 gained support for configuring the multicast querier feature of
11720 bridge devices, via the new MulticastQuerier= setting in .netdev
11721 files. Similarly, snooping on the IGMP traffic can be controlled
11722 via the new setting MulticastSnooping=.
11724 A new setting PreferredLifetime= has been added for addresses
11725 configured in .network file to configure the lifetime intended for an
11728 The systemd-networkd DHCP server gained the option EmitRouter=, which
11729 defaults to yes, to configure whether the DHCP Option 3 (Router)
11732 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
11733 systemd-socket-activate and installed into /usr/bin. It is now fully
11736 * systemd-journald now uses separate threads to flush changes to disk
11737 when closing journal files, thus reducing impact of slow disk I/O on
11738 logging performance.
11740 * The sd-journal API gained two new calls
11741 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
11742 can be used to open journal files using file descriptors instead of
11743 file or directory paths. sd_journal_open_container() has been
11744 deprecated, sd_journal_open_directory_fd() should be used instead
11745 with the flag SD_JOURNAL_OS_ROOT.
11747 * journalctl learned a new output mode "-o short-unix" that outputs log
11748 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
11749 UTC). It also gained support for a new --no-hostname setting to
11750 suppress the hostname column in the family of "short" output modes.
11752 * systemd-ask-password now optionally skips printing of the password to
11753 stdout with --no-output which can be useful in scripts.
11755 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
11756 (devices tagged with ID_MAKER_TOOL) are now tagged with
11757 "uaccess" and are available to logged in users.
11759 * The DeviceAllow= unit setting now supports specifiers (with "%").
11761 * "systemctl show" gained a new --value switch, which allows print a
11762 only the contents of a specific unit property, without also printing
11763 the property's name. Similar support was added to "show*" verbs
11764 of loginctl and machinectl that output "key=value" lists.
11766 * A new unit type "generated" was added for files dynamically generated
11767 by generator tools. Similarly, a new unit type "transient" is used
11768 for unit files created using the runtime API. "systemctl enable" will
11769 refuse to operate on such files.
11771 * A new command "systemctl revert" has been added that may be used to
11772 revert to the vendor version of a unit file, in case local changes
11773 have been made by adding drop-ins or overriding the unit file.
11775 * "machinectl clean" gained a new verb to automatically remove all or
11776 just hidden container images.
11778 * systemd-tmpfiles gained support for a new line type "e" for emptying
11779 directories, if they exist, without creating them if they don't.
11781 * systemd-nspawn gained support for automatically patching the UID/GIDs
11782 of the owners and the ACLs of all files and directories in a
11783 container tree to match the UID/GID user namespacing range selected
11784 for the container invocation. This mode is enabled via the new
11785 --private-users-chown switch. It also gained support for
11786 automatically choosing a free, previously unused UID/GID range when
11787 starting a container, via the new --private-users=pick setting (which
11788 implies --private-users-chown). Together, these options for the first
11789 time make user namespacing for nspawn containers fully automatic and
11790 thus deployable. The systemd-nspawn@.service template unit file has
11791 been changed to use this functionality by default.
11793 * systemd-nspawn gained a new --network-zone= switch, that allows
11794 creating ad-hoc virtual Ethernet links between multiple containers,
11795 that only exist as long as at least one container referencing them is
11796 running. This allows easy connecting of multiple containers with a
11797 common link that implements an Ethernet broadcast domain. Each of
11798 these network "zones" may be named relatively freely by the user, and
11799 may be referenced by any number of containers, but each container may
11800 only reference one of these "zones". On the lower level, this is
11801 implemented by an automatically managed bridge network interface for
11802 each zone, that is created when the first container referencing its
11803 zone is created and removed when the last one referencing its zone
11806 * The default start timeout may now be configured on the kernel command
11807 line via systemd.default_timeout_start_sec=. It was already
11808 configurable via the DefaultTimeoutStartSec= option in
11809 /etc/systemd/system.conf.
11811 * Socket units gained a new TriggerLimitIntervalSec= and
11812 TriggerLimitBurst= setting to configure a limit on the activation
11813 rate of the socket unit.
11815 * The LimitNICE= setting now optionally takes normal UNIX nice values
11816 in addition to the raw integer limit value. If the specified
11817 parameter is prefixed with "+" or "-" and is in the range -20…19 the
11818 value is understood as UNIX nice value. If not prefixed like this it
11819 is understood as raw RLIMIT_NICE limit.
11821 * Note that the effect of the PrivateDevices= unit file setting changed
11822 slightly with this release: the per-device /dev file system will be
11823 mounted read-only from this version on, and will have "noexec"
11824 set. This (minor) change of behavior might cause some (exceptional)
11825 legacy software to break, when PrivateDevices=yes is set for its
11826 service. Please leave PrivateDevices= off if you run into problems
11829 * systemd-bootchart has been split out to a separate repository:
11830 https://github.com/systemd/systemd-bootchart
11832 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
11833 merged into the kernel in its current form.
11835 * The compatibility libraries libsystemd-daemon.so,
11836 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
11837 which have been deprecated since systemd-209 have been removed along
11838 with the corresponding pkg-config files. All symbols provided by
11839 those libraries are provided by libsystemd.so.
11841 * The Capabilities= unit file setting has been removed (it is ignored
11842 for backwards compatibility). AmbientCapabilities= and
11843 CapabilityBoundingSet= should be used instead.
11845 * A new special target has been added, initrd-root-device.target,
11846 which creates a synchronization point for dependencies of the root
11847 device in early userspace. Initramfs builders must ensure that this
11848 target is now included in early userspace.
11850 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
11851 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
11852 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
11853 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
11854 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
11855 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
11856 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
11857 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
11858 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
11859 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
11860 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
11861 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
11862 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
11863 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
11864 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
11865 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
11866 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
11867 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
11868 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
11869 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
11870 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
11871 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
11872 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
11873 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
11874 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
11877 — Fairfax, 2016-05-21
11881 * The systemd-resolved DNS resolver service has gained a substantial
11882 set of new features, most prominently it may now act as a DNSSEC
11883 validating stub resolver. DNSSEC mode is currently turned off by
11884 default, but is expected to be turned on by default in one of the
11885 next releases. For now, we invite everybody to test the DNSSEC logic
11886 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
11887 service also gained a full set of D-Bus interfaces, including calls
11888 to configure DNS and DNSSEC settings per link (for use by external
11889 network management software). systemd-resolved and systemd-networkd
11890 now distinguish between "search" and "routing" domains. The former
11891 are used to qualify single-label names, the latter are used purely
11892 for routing lookups within certain domains to specific links.
11893 resolved now also synthesizes RRs for all entries from /etc/hosts.
11895 * The systemd-resolve tool (which is a client utility for
11896 systemd-resolved) has been improved considerably and is now fully
11897 supported and documented. Hence it has moved from /usr/lib/systemd to
11900 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
11903 * The coredump collection logic has been reworked: when a coredump is
11904 collected it is now written to disk, compressed and processed
11905 (including stacktrace extraction) from a new instantiated service
11906 systemd-coredump@.service, instead of directly from the
11907 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
11908 processing large coredumps can take up a substantial amount of
11909 resources and time, and this previously happened entirely outside of
11910 systemd's service supervision. With the new logic the core_pattern
11911 hook only does minimal metadata collection before passing off control
11912 to the new instantiated service, which is configured with a time
11913 limit, a nice level and other settings to minimize negative impact on
11914 the rest of the system. Also note that the new logic will honour the
11915 RLIMIT_CORE setting of the crashed process, which now allows users
11916 and processes to turn off coredumping for their processes by setting
11919 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
11920 and all forked processes by default. Previously, PID 1 would leave
11921 the setting at "0" for all processes, as set by the kernel. Note that
11922 the resource limit traditionally has no effect on the generated
11923 coredumps on the system if the /proc/sys/kernel/core_pattern hook
11924 logic is used. Since the limit is now honoured (see above) its
11925 default has been changed so that the coredumping logic is enabled by
11926 default for all processes, while allowing specific opt-out.
11928 * When the stacktrace is extracted from processes of system users, this
11929 is now done as "systemd-coredump" user, in order to sandbox this
11930 potentially security sensitive parsing operation. (Note that when
11931 processing coredumps of normal users this is done under the user ID
11932 of process that crashed, as before.) Packagers should take notice
11933 that it is now necessary to create the "systemd-coredump" system user
11934 and group at package installation time.
11936 * The systemd-activate socket activation testing tool gained support
11937 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
11938 and --seqpacket switches. It also has been extended to support both
11939 new-style and inetd-style file descriptor passing. Use the new
11940 --inetd switch to request inetd-style file descriptor passing.
11942 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
11943 variable, which takes a boolean value. If set to false, ANSI color
11944 output is disabled in the tools even when run on a terminal that
11947 * The VXLAN support in networkd now supports two new settings
11948 DestinationPort= and PortRange=.
11950 * A new systemd.machine_id= kernel command line switch has been added,
11951 that may be used to set the machine ID in /etc/machine-id if it is
11952 not initialized yet. This command line option has no effect if the
11953 file is already initialized.
11955 * systemd-nspawn gained a new --as-pid2 switch that invokes any
11956 specified command line as PID 2 rather than PID 1 in the
11957 container. In this mode PID 1 is a minimal stub init process that
11958 implements the special POSIX and Linux semantics of PID 1 regarding
11959 signal and child process management. Note that this stub init process
11960 is implemented in nspawn itself and requires no support from the
11961 container image. This new logic is useful to support running
11962 arbitrary commands in the container, as normal processes are
11963 generally not prepared to run as PID 1.
11965 * systemd-nspawn gained a new --chdir= switch for setting the current
11966 working directory for the process started in the container.
11968 * "journalctl /dev/sda" will now output all kernel log messages for
11969 specified device from the current boot, in addition to all devices
11970 that are parents of it. This should make log output about devices
11971 pretty useful, as long as kernel drivers attach enough metadata to
11972 the log messages. (The usual SATA drivers do.)
11974 * The sd-journal API gained two new calls
11975 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
11976 that report whether log data from /run or /var has been found.
11978 * journalctl gained a new switch "--fields" that prints all journal
11979 record field names currently in use in the journal. This is backed
11980 by two new sd-journal API calls sd_journal_enumerate_fields() and
11981 sd_journal_restart_fields().
11983 * Most configurable timeouts in systemd now expect an argument of
11984 "infinity" to turn them off, instead of "0" as before. The semantics
11985 from now on is that a timeout of "0" means "now", and "infinity"
11986 means "never". To maintain backwards compatibility, "0" continues to
11987 turn off previously existing timeout settings.
11989 * "systemctl reload-or-try-restart" has been renamed to "systemctl
11990 try-reload-or-restart" to clarify what it actually does: the "try"
11991 logic applies to both reloading and restarting, not just restarting.
11992 The old name continues to be accepted for compatibility.
11994 * On boot-up, when PID 1 detects that the system clock is behind the
11995 release date of the systemd version in use, the clock is now set
11996 to the latter. Previously, this was already done in timesyncd, in order
11997 to avoid running with clocks set to the various clock epochs such as
11998 1902, 1938 or 1970. With this change the logic is now done in PID 1
11999 in addition to timesyncd during early boot-up, so that it is enforced
12000 before the first process is spawned by systemd. Note that the logic
12001 in timesyncd remains, as it is more comprehensive and ensures
12002 clock monotonicity by maintaining a persistent timestamp file in
12003 /var. Since /var is generally not available in earliest boot or the
12004 initrd, this part of the logic remains in timesyncd, and is not done
12007 * Support for tweaking details in net_cls.class_id through the
12008 NetClass= configuration directive has been removed, as the kernel
12009 people have decided to deprecate that controller in cgroup v2.
12010 Userspace tools such as nftables are moving over to setting rules
12011 that are specific to the full cgroup path of a task, which obsoletes
12012 these controllers anyway. The NetClass= directive is kept around for
12013 legacy compatibility reasons. For a more in-depth description of the
12014 kernel change, please refer to the respective upstream commit:
12016 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
12018 * A new service setting RuntimeMaxSec= has been added that may be used
12019 to specify a maximum runtime for a service. If the timeout is hit, the
12020 service is terminated and put into a failure state.
12022 * A new service setting AmbientCapabilities= has been added. It allows
12023 configuration of additional Linux process capabilities that are
12024 passed to the activated processes. This is only available on very
12027 * The process resource limit settings in service units may now be used
12028 to configure hard and soft limits individually.
12030 * The various libsystemd APIs such as sd-bus or sd-event now publicly
12031 expose support for gcc's __attribute__((cleanup())) C extension.
12032 Specifically, for many object destructor functions alternative
12033 versions have been added that have names suffixed with "p" and take a
12034 pointer to a pointer to the object to destroy, instead of just a
12035 pointer to the object itself. This is useful because these destructor
12036 functions may be used directly as parameters to the cleanup
12037 construct. Internally, systemd has been a heavy user of this GCC
12038 extension for a long time, and with this change similar support is
12039 now available to consumers of the library outside of systemd. Note
12040 that by using this extension in your sources compatibility with old
12041 and strictly ANSI compatible C compilers is lost. However, all gcc or
12042 LLVM versions of recent years support this extension.
12044 * Timer units gained support for a new setting RandomizedDelaySec= that
12045 allows configuring some additional randomized delay to the configured
12046 time. This is useful to spread out timer events to avoid load peaks in
12047 clusters or larger setups.
12049 * Calendar time specifications now support sub-second accuracy.
12051 * Socket units now support listening on SCTP and UDP-lite protocol
12054 * The sd-event API now comes with a full set of man pages.
12056 * Older versions of systemd contained experimental support for
12057 compressing journal files and coredumps with the LZ4 compressor that
12058 was not compatible with the lz4 binary (due to API limitations of the
12059 lz4 library). This support has been removed; only support for files
12060 compatible with the lz4 binary remains. This LZ4 logic is now
12061 officially supported and no longer considered experimental.
12063 * The dkr image import logic has been removed again from importd. dkr's
12064 micro-services focus doesn't fit into the machine image focus of
12065 importd, and quickly got out of date with the upstream dkr API.
12067 * Creation of the /run/lock/lockdev/ directory was dropped from
12068 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
12069 been available for many years. If you still need this, you need to
12070 create your own tmpfiles.d config file with:
12072 d /run/lock/lockdev 0775 root lock -
12074 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
12075 and RebootArgument= have been moved from the [Service] section of
12076 unit files to [Unit], and they are now supported on all unit types,
12077 not just service units. Of course, systemd will continue to
12078 understand these settings also at the old location, in order to
12079 maintain compatibility.
12081 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
12082 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
12083 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
12084 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
12085 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
12086 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
12087 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
12088 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
12089 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
12090 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
12091 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
12092 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
12093 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
12094 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
12095 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
12096 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
12097 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
12098 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
12099 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
12101 — Berlin, 2016-02-11
12105 * A number of properties previously only settable in unit
12106 files are now also available as properties to set when
12107 creating transient units programmatically via the bus, as it
12108 is exposed with systemd-run's --property=
12109 setting. Specifically, these are: SyslogIdentifier=,
12110 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
12111 EnvironmentFile=, ReadWriteDirectories=,
12112 ReadOnlyDirectories=, InaccessibleDirectories=,
12113 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
12115 * When creating transient services via the bus API it is now
12116 possible to pass in a set of file descriptors to use as
12117 STDIN/STDOUT/STDERR for the invoked process.
12119 * Slice units may now be created transiently via the bus APIs,
12120 similar to the way service and scope units may already be
12121 created transiently.
12123 * Wherever systemd expects a calendar timestamp specification
12124 (like in journalctl's --since= and --until= switches) UTC
12125 timestamps are now supported. Timestamps suffixed with "UTC"
12126 are now considered to be in Universal Time Coordinated
12127 instead of the local timezone. Also, timestamps may now
12128 optionally be specified with sub-second accuracy. Both of
12129 these additions also apply to recurring calendar event
12130 specification, such as OnCalendar= in timer units.
12132 * journalctl gained a new "--sync" switch that asks the
12133 journal daemon to write all so far unwritten log messages to
12134 disk and sync the files, before returning.
12136 * systemd-tmpfiles learned two new line types "q" and "Q" that
12137 operate like "v", but also set up a basic btrfs quota
12138 hierarchy when used on a btrfs file system with quota
12141 * tmpfiles' "v", "q" and "Q" will now create a plain directory
12142 instead of a subvolume (even on a btrfs file system) if the
12143 root directory is a plain directory, and not a
12144 subvolume. This should simplify things with certain chroot()
12145 environments which are not aware of the concept of btrfs
12148 * systemd-detect-virt gained a new --chroot switch to detect
12149 whether execution takes place in a chroot() environment.
12151 * CPUAffinity= now takes CPU index ranges in addition to
12152 individual indexes.
12154 * The various memory-related resource limit settings (such as
12155 LimitAS=) now understand the usual K, M, G, … suffixes to
12156 the base of 1024 (IEC). Similar, the time-related resource
12157 limit settings understand the usual min, h, day, … suffixes
12160 * There's a new system.conf setting DefaultTasksMax= to
12161 control the default TasksMax= setting for services and
12162 scopes running on the system. (TasksMax= is the primary
12163 setting that exposes the "pids" cgroup controller on systemd
12164 and was introduced in the previous systemd release.) The
12165 setting now defaults to 512, which means services that are
12166 not explicitly configured otherwise will only be able to
12167 create 512 processes or threads at maximum, from this
12168 version on. Note that this means that thread- or
12169 process-heavy services might need to be reconfigured to set
12170 TasksMax= to a higher value. It is sufficient to set
12171 TasksMax= in these specific unit files to a higher value, or
12172 even "infinity". Similar, there's now a logind.conf setting
12173 UserTasksMax= that defaults to 4096 and limits the total
12174 number of processes or tasks each user may own
12175 concurrently. nspawn containers also have the TasksMax=
12176 value set by default now, to 8192. Note that all of this
12177 only has an effect if the "pids" cgroup controller is
12178 enabled in the kernel. The general benefit of these changes
12179 should be a more robust and safer system, that provides a
12180 certain amount of per-service fork() bomb protection.
12182 * systemd-nspawn gained the new --network-veth-extra= switch
12183 to define additional and arbitrarily-named virtual Ethernet
12184 links between the host and the container.
12186 * A new service execution setting PassEnvironment= has been
12187 added that allows importing select environment variables
12188 from PID1's environment block into the environment block of
12191 * Timer units gained support for a new RemainAfterElapse=
12192 setting which takes a boolean argument. It defaults to on,
12193 exposing behaviour unchanged to previous releases. If set to
12194 off, timer units are unloaded after they elapsed if they
12195 cannot elapse again. This is particularly useful for
12196 transient timer units, which shall not stay around longer
12197 than until they first elapse.
12199 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
12200 default now (the kernel default is 16). This is beneficial
12201 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
12202 allows substantially larger numbers of queued
12203 datagrams. This should increase the capability of systemd to
12204 parallelize boot-up, as logging and sd_notify() are unlikely
12205 to stall execution anymore. If you need to change the value
12206 from the new defaults, use the usual sysctl.d/ snippets.
12208 * The compression framing format used by the journal or
12209 coredump processing has changed to be in line with what the
12210 official LZ4 tools generate. LZ4 compression support in
12211 systemd was considered unsupported previously, as the format
12212 was not compatible with the normal tools. With this release
12213 this has changed now, and it is hence safe for downstream
12214 distributions to turn it on. While not compressing as well
12215 as the XZ, LZ4 is substantially faster, which makes
12216 it a good default choice for the compression logic in the
12217 journal and in coredump handling.
12219 * Any reference to /etc/mtab has been dropped from
12220 systemd. The file has been obsolete since a while, but
12221 systemd refused to work on systems where it was incorrectly
12222 set up (it should be a symlink or non-existent). Please make
12223 sure to update to util-linux 2.27.1 or newer in conjunction
12224 with this systemd release, which also drops any reference to
12225 /etc/mtab. If you maintain a distribution make sure that no
12226 software you package still references it, as this is a
12227 likely source of bugs. There's also a glibc bug pending,
12228 asking for removal of any reference to this obsolete file:
12230 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
12232 Note that only util-linux versions built with
12233 --enable-libmount-force-mountinfo are supported.
12235 * Support for the ".snapshot" unit type has been removed. This
12236 feature turned out to be little useful and little used, and
12237 has now been removed from the core and from systemctl.
12239 * The dependency types RequiresOverridable= and
12240 RequisiteOverridable= have been removed from systemd. They
12241 have been used only very sparingly to our knowledge and
12242 other options that provide a similar effect (such as
12243 systemctl --mode=ignore-dependencies) are much more useful
12244 and commonly used. Moreover, they were only half-way
12245 implemented as the option to control behaviour regarding
12246 these dependencies was never added to systemctl. By removing
12247 these dependency types the execution engine becomes a bit
12248 simpler. Unit files that use these dependencies should be
12249 changed to use the non-Overridable dependency types
12250 instead. In fact, when parsing unit files with these
12251 options, that's what systemd will automatically convert them
12252 too, but it will also warn, asking users to fix the unit
12253 files accordingly. Removal of these dependency types should
12254 only affect a negligible number of unit files in the wild.
12256 * Behaviour of networkd's IPForward= option changed
12257 (again). It will no longer maintain a per-interface setting,
12258 but propagate one way from interfaces where this is enabled
12259 to the global kernel setting. The global setting will be
12260 enabled when requested by a network that is set up, but
12261 never be disabled again. This change was made to make sure
12262 IPv4 and IPv6 behaviour regarding packet forwarding is
12263 similar (as the Linux IPv6 stack does not support
12264 per-interface control of this setting) and to minimize
12267 * In unit files the behaviour of %u, %U, %h, %s has
12268 changed. These specifiers will now unconditionally resolve
12269 to the various user database fields of the user that the
12270 systemd instance is running as, instead of the user
12271 configured in the specific unit via User=. Note that this
12272 effectively doesn't change much, as resolving of these
12273 specifiers was already turned off in the --system instance
12274 of systemd, as we cannot do NSS lookups from PID 1. In the
12275 --user instance of systemd these specifiers where correctly
12276 resolved, but hardly made any sense, since the user instance
12277 lacks privileges to do user switches anyway, and User= is
12278 hence useless. Moreover, even in the --user instance of
12279 systemd behaviour was awkward as it would only take settings
12280 from User= assignment placed before the specifier into
12281 account. In order to unify and simplify the logic around
12282 this the specifiers will now always resolve to the
12283 credentials of the user invoking the manager (which in case
12284 of PID 1 is the root user).
12286 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
12287 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
12288 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
12289 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
12290 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
12291 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
12292 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
12293 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
12294 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
12295 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
12296 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
12297 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
12298 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
12299 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
12302 — Berlin, 2015-11-18
12306 * systemd now depends on util-linux v2.27. More specifically,
12307 the newly added mount monitor feature in libmount now
12308 replaces systemd's former own implementation.
12310 * libmount mandates /etc/mtab not to be regular file, and
12311 systemd now enforces this condition at early boot.
12312 /etc/mtab has been deprecated and warned about for a very
12313 long time, so systems running systemd should already have
12314 stopped having this file around as anything else than a
12315 symlink to /proc/self/mounts.
12317 * Support for the "pids" cgroup controller has been added. It
12318 allows accounting the number of tasks in a cgroup and
12319 enforcing limits on it. This adds two new setting
12320 TasksAccounting= and TasksMax= to each unit, as well as a
12321 global option DefaultTasksAccounting=.
12323 * Support for the "net_cls" cgroup controller has been added.
12324 It allows assigning a net class ID to each task in the
12325 cgroup, which can then be used in firewall rules and traffic
12326 shaping configurations. Note that the kernel netfilter net
12327 class code does not currently work reliably for ingress
12328 packets on unestablished sockets.
12330 This adds a new config directive called NetClass= to CGroup
12331 enabled units. Allowed values are positive numbers for fixed
12332 assignments and "auto" for picking a free value
12335 * 'systemctl is-system-running' now returns 'offline' if the
12336 system is not booted with systemd. This command can now be
12337 used as a substitute for 'systemd-notify --booted'.
12339 * Watchdog timeouts have been increased to 3 minutes for all
12340 in-tree service files. Apparently, disk IO issues are more
12341 frequent than we hoped, and user reported >1 minute waiting
12344 * 'machine-id-commit' functionality has been merged into
12345 'machine-id-setup --commit'. The separate binary has been
12348 * The WorkingDirectory= directive in unit files may now be set
12349 to the special value '~'. In this case, the working
12350 directory is set to the home directory of the user
12351 configured in User=.
12353 * "machinectl shell" will now open the shell in the home
12354 directory of the selected user by default.
12356 * The CrashChVT= configuration file setting is renamed to
12357 CrashChangeVT=, following our usual logic of not
12358 abbreviating unnecessarily. The old directive is still
12359 supported for compat reasons. Also, this directive now takes
12360 an integer value between 1 and 63, or a boolean value. The
12361 formerly supported '-1' value for disabling stays around for
12364 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
12365 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
12366 RootDirectory= properties can now be set for transient
12369 * The systemd-analyze tool gained a new "set-log-target" verb
12370 to change the logging target the system manager logs to
12371 dynamically during runtime. This is similar to how
12372 "systemd-analyze set-log-level" already changes the log
12375 * In nspawn /sys is now mounted as tmpfs, with only a selected
12376 set of subdirectories mounted in from the real sysfs. This
12377 enhances security slightly, and is useful for ensuring user
12378 namespaces work correctly.
12380 * Support for USB FunctionFS activation has been added. This
12381 allows implementation of USB gadget services that are
12382 activated as soon as they are requested, so that they don't
12383 have to run continuously, similar to classic socket
12386 * The "systemctl exit" command now optionally takes an
12387 additional parameter that sets the exit code to return from
12388 the systemd manager when exiting. This is only relevant when
12389 running the systemd user instance, or when running the
12390 system instance in a container.
12392 * sd-bus gained the new API calls sd_bus_path_encode_many()
12393 and sd_bus_path_decode_many() that allow easy encoding and
12394 decoding of multiple identifier strings inside a D-Bus
12395 object path. Another new call sd_bus_default_flush_close()
12396 has been added to flush and close per-thread default
12399 * systemd-cgtop gained support for a -M/--machine= switch to
12400 show the control groups within a certain container only.
12402 * "systemctl kill" gained support for an optional --fail
12403 switch. If specified the requested operation will fail of no
12404 processes have been killed, because the unit had no
12405 processes attached, or similar.
12407 * A new systemd.crash_reboot=1 kernel command line option has
12408 been added that triggers a reboot after crashing. This can
12409 also be set through CrashReboot= in systemd.conf.
12411 * The RuntimeDirectory= setting now understands unit
12412 specifiers like %i or %f.
12414 * A new (still internal) library API sd-ipv4acd has been added,
12415 that implements address conflict detection for IPv4. It's
12416 based on code from sd-ipv4ll, and will be useful for
12417 detecting DHCP address conflicts.
12419 * File descriptors passed during socket activation may now be
12420 named. A new API sd_listen_fds_with_names() is added to
12421 access the names. The default names may be overridden,
12422 either in the .socket file using the FileDescriptorName=
12423 parameter, or by passing FDNAME= when storing the file
12424 descriptors using sd_notify().
12426 * systemd-networkd gained support for:
12428 - Setting the IPv6 Router Advertisement settings via
12429 IPv6AcceptRouterAdvertisements= in .network files.
12431 - Configuring the HelloTimeSec=, MaxAgeSec= and
12432 ForwardDelaySec= bridge parameters in .netdev files.
12434 - Configuring PreferredSource= for static routes in
12437 * The "ask-password" framework used to query for LUKS harddisk
12438 passwords or SSL passwords during boot gained support for
12439 caching passwords in the kernel keyring, if it is
12440 available. This makes sure that the user only has to type in
12441 a passphrase once if there are multiple objects to unlock
12442 with the same one. Previously, such password caching was
12443 available only when Plymouth was used; this moves the
12444 caching logic into the systemd codebase itself. The
12445 "systemd-ask-password" utility gained a new --keyname=
12446 switch to control which kernel keyring key to use for
12447 caching a password in. This functionality is also useful for
12448 enabling display managers such as gdm to automatically
12449 unlock the user's GNOME keyring if its passphrase, the
12450 user's password and the harddisk password are the same, if
12451 gdm-autologin is used.
12453 * When downloading tar or raw images using "machinectl
12454 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
12455 file is now also downloaded, if it is available and stored
12456 next to the image file.
12458 * Units of type ".socket" gained a new boolean setting
12459 Writable= which is only useful in conjunction with
12460 ListenSpecial=. If true, enables opening the specified
12461 special file in O_RDWR mode rather than O_RDONLY mode.
12463 * systemd-rfkill has been reworked to become a singleton
12464 service that is activated through /dev/rfkill on each rfkill
12465 state change and saves the settings to disk. This way,
12466 systemd-rfkill is now compatible with devices that exist
12467 only intermittendly, and even restores state if the previous
12468 system shutdown was abrupt rather than clean.
12470 * The journal daemon gained support for vacuuming old journal
12471 files controlled by the number of files that shall remain,
12472 in addition to the already existing control by size and by
12473 date. This is useful as journal interleaving performance
12474 degrades with too many separate journal files, and allows
12475 putting an effective limit on them. The new setting defaults
12476 to 100, but this may be changed by setting SystemMaxFiles=
12477 and RuntimeMaxFiles= in journald.conf. Also, the
12478 "journalctl" tool gained the new --vacuum-files= switch to
12479 manually vacuum journal files to leave only the specified
12480 number of files in place.
12482 * udev will now create /dev/disk/by-path links for ATA devices
12483 on kernels where that is supported.
12485 * Galician, Serbian, Turkish and Korean translations were added.
12487 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
12488 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
12489 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
12490 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
12491 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
12492 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
12493 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
12494 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
12495 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
12496 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
12497 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
12498 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
12499 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
12500 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
12501 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
12502 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
12503 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
12504 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
12506 — Berlin, 2015-10-07
12510 * The DHCP implementation of systemd-networkd gained a set of
12513 - The DHCP server now supports emitting DNS and NTP
12514 information. It may be enabled and configured via
12515 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
12516 and NTP information is enabled, but no servers are
12517 configured, the corresponding uplink information (if there
12518 is any) is propagated.
12520 - Server and client now support transmission and reception
12521 of timezone information. It can be configured via the
12522 newly introduced network options UseTimezone=,
12523 EmitTimezone=, and Timezone=. Transmission of timezone
12524 information is enabled between host and containers by
12525 default now: the container will change its local timezone
12526 to what the host has set.
12528 - Lease timeouts can now be configured via
12529 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
12531 - The DHCP server improved on the stability of
12532 leases. Clients are more likely to get the same lease
12533 information back, even if the server loses state.
12535 - The DHCP server supports two new configuration options to
12536 control the lease address pool metrics, PoolOffset= and
12539 * The encapsulation limit of tunnels in systemd-networkd may
12540 now be configured via 'EncapsulationLimit='. It allows
12541 modifying the maximum additional levels of encapsulation
12542 that are permitted to be prepended to a packet.
12544 * systemd now supports the concept of user buses replacing
12545 session buses, if used with dbus-1.10 (and enabled via dbus
12546 --enable-user-session). It previously only supported this on
12547 kdbus-enabled systems, and this release expands this to
12548 'dbus-daemon' systems.
12550 * systemd-networkd now supports predictable interface names
12551 for virtio devices.
12553 * systemd now optionally supports the new Linux kernel
12554 "unified" control group hierarchy. If enabled via the kernel
12555 command-line option 'systemd.unified_cgroup_hierarchy=1',
12556 systemd will try to mount the unified cgroup hierarchy
12557 directly on /sys/fs/cgroup. If not enabled, or not
12558 available, systemd will fall back to the legacy cgroup
12559 hierarchy setup, as before. Host system and containers can
12560 mix and match legacy and unified hierarchies as they
12561 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
12562 environment variable to individually select the hierarchy to
12563 use for executed containers. By default, nspawn will use the
12564 unified hierarchy for the containers if the host uses the
12565 unified hierarchy, and the legacy hierarchy otherwise.
12566 Please note that at this point the unified hierarchy is an
12567 experimental kernel feature and is likely to change in one
12568 of the next kernel releases. Therefore, it should not be
12569 enabled by default in downstream distributions yet. The
12570 minimum required kernel version for the unified hierarchy to
12571 work is 4.2. Note that when the unified hierarchy is used
12572 for the first time delegated access to controllers is
12573 safe. Because of this systemd-nspawn containers will get
12574 access to controllers now, as will systemd user
12575 sessions. This means containers and user sessions may now
12576 manage their own resources, partitioning up what the system
12579 * A new special scope unit "init.scope" has been introduced
12580 that encapsulates PID 1 of the system. It may be used to
12581 determine resource usage and enforce resource limits on PID
12582 1 itself. PID 1 hence moved out of the root of the control
12585 * The cgtop tool gained support for filtering out kernel
12586 threads when counting tasks in a control group. Also, the
12587 count of processes is now recursively summed up by
12588 default. Two options -k and --recursive= have been added to
12589 revert to old behaviour. The tool has also been updated to
12590 work correctly in containers now.
12592 * systemd-nspawn's --bind= and --bind-ro= options have been
12593 extended to allow creation of non-recursive bind mounts.
12595 * libsystemd gained two new calls sd_pid_get_cgroup() and
12596 sd_peer_get_cgroup() which return the control group path of
12597 a process or peer of a connected AF_UNIX socket. This
12598 function call is particularly useful when implementing
12599 delegated subtrees support in the control group hierarchy.
12601 * The "sd-event" event loop API of libsystemd now supports
12602 correct dequeuing of real-time signals, without losing
12605 * When systemd requests a polkit decision when managing units it
12606 will now add additional fields to the request, including unit
12607 name and desired operation. This enables more powerful polkit
12608 policies, that make decisions depending on these parameters.
12610 * nspawn learnt support for .nspawn settings files, that may
12611 accompany the image files or directories of containers, and
12612 may contain additional settings for the container. This is
12613 an alternative to configuring container parameters via the
12614 nspawn command line.
12616 Contributions from: Cristian Rodríguez, Daniel Mack, David
12617 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
12618 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
12619 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
12620 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
12621 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
12622 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
12623 Andersen, Tom Gundersen, Torstein Husebø
12625 — Berlin, 2015-09-08
12629 * machinectl gained a new verb 'shell' which opens a fresh
12630 shell on the target container or the host. It is similar to
12631 the existing 'login' command of machinectl, but spawns the
12632 shell directly without prompting for username or
12633 password. The pseudo machine '.host' now refers to the local
12634 host and is used by default. Hence, 'machinectl shell' can
12635 be used as replacement for 'su -' which spawns a session as
12636 a fresh systemd unit in a way that is fully isolated from
12637 the originating session.
12639 * systemd-networkd learned to cope with private-zone DHCP
12640 options and allows other programs to query the values.
12642 * SELinux access control when enabling/disabling units is no
12643 longer enforced with this release. The previous implementation
12644 was incorrect, and a new corrected implementation is not yet
12645 available. As unit file operations are still protected via
12646 polkit and D-Bus policy this is not a security problem. Yet,
12647 distributions which care about optimal SELinux support should
12648 probably not stabilize on this release.
12650 * sd-bus gained support for matches of type "arg0has=", that
12651 test for membership of strings in string arrays sent in bus
12654 * systemd-resolved now dumps the contents of its DNS and LLMNR
12655 caches to the logs on reception of the SIGUSR1 signal. This
12656 is useful to debug DNS behaviour.
12658 * The coredumpctl tool gained a new --directory= option to
12659 operate on journal files in a specific directory.
12661 * "systemctl reboot" and related commands gained a new
12662 "--message=" option which may be used to set a free-text
12663 wall message when shutting down or rebooting the
12664 system. This message is also logged, which is useful for
12665 figuring out the reason for a reboot or shutdown a
12668 * The "systemd-resolve-host" tool's -i switch now takes
12669 network interface numbers as alternative to interface names.
12671 * A new unit file setting for services has been introduced:
12672 UtmpMode= allows configuration of how precisely systemd
12673 handles utmp and wtmp entries for the service if this is
12674 enabled. This allows writing services that appear similar to
12675 user sessions in the output of the "w", "who", "last" and
12678 * systemd-resolved will now locally synthesize DNS resource
12679 records for the "localhost" and "gateway" domains as well as
12680 the local hostname. This should ensure that clients querying
12681 RRs via resolved will get similar results as those going via
12682 NSS, if nss-myhostname is enabled.
12684 Contributions from: Alastair Hughes, Alex Crawford, Daniel
12685 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
12686 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
12687 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
12688 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
12689 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
12690 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
12691 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
12692 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
12693 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
12694 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
12695 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
12697 — Berlin, 2015-08-27
12701 * The systemd-efi-boot-generator functionality was merged into
12702 systemd-gpt-auto-generator.
12704 * systemd-networkd now supports Group Policy for vxlan
12705 devices. It can be enabled via the new boolean configuration
12706 option called 'GroupPolicyExtension='.
12708 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
12709 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
12710 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
12712 — Berlin, 2015-07-31
12716 * The python-systemd code has been removed from the systemd repository.
12717 A new repository has been created which accommodates the code from
12718 now on, and we kindly ask distributions to create a separate package
12719 for this: https://github.com/systemd/python-systemd
12721 * The systemd daemon will now reload its main configuration
12722 (/etc/systemd/system.conf) on daemon-reload.
12724 * sd-dhcp now exposes vendor specific extensions via
12725 sd_dhcp_lease_get_vendor_specific().
12727 * systemd-networkd gained a number of new configuration options.
12729 - A new boolean configuration option for TAP devices called
12730 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
12731 device, thus allowing to send and receive GSO packets.
12733 - A new tunnel configuration option called 'CopyDSCP='.
12734 If enabled, the DSCP field of ip6 tunnels is copied into the
12735 decapsulated packet.
12737 - A set of boolean bridge configuration options were added.
12738 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
12739 and 'UnicastFlood=' are now parsed by networkd and applied to the
12740 respective bridge link device via the respective IFLA_BRPORT_*
12743 - A new string configuration option to override the hostname sent
12744 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
12745 is true, networkd will use the configured hostname instead of the
12746 system hostname when sending DHCP requests.
12748 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
12749 networkd will configure the IPv6 flow-label of the tunnel device
12750 according to RFC2460.
12752 - The 'macvtap' virtual network devices are now supported, similar to
12753 the already supported 'macvlan' devices.
12755 * systemd-resolved now implements RFC5452 to improve resilience against
12756 cache poisoning. Additionally, source port randomization is enabled
12757 by default to further protect against DNS spoofing attacks.
12759 * nss-mymachines now supports translating UIDs and GIDs of running
12760 containers with user-namespaces enabled. If a container 'foo'
12761 translates a host uid 'UID' to the container uid 'TUID', then
12762 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
12763 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
12764 mapped as 'vg-foo-TGID'.
12766 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
12767 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
12768 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
12769 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
12770 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
12771 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
12772 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
12773 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
12774 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
12775 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
12777 — Berlin, 2015-07-29
12781 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
12782 There are no known issues with current sysfs, and udev does not need
12783 or should be used to work around such bugs.
12785 * udev does no longer enable USB HID power management. Several reports
12786 indicate, that some devices cannot handle that setting.
12788 * The udev accelerometer helper was removed. The functionality
12789 is now fully included in iio-sensor-proxy. But this means,
12790 older iio-sensor-proxy versions will no longer provide
12791 accelerometer/orientation data with this systemd version.
12792 Please upgrade iio-sensor-proxy to version 1.0.
12794 * networkd gained a new configuration option IPv6PrivacyExtensions=
12795 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
12796 for Stateless Address") on selected networks.
12798 * For the sake of fewer build-time dependencies and less code in the
12799 main repository, the python bindings are about to be removed in the
12800 next release. A new repository has been created which accommodates
12801 the code from now on, and we kindly ask distributions to create a
12802 separate package for this. The removal will take place in v223.
12804 https://github.com/systemd/python-systemd
12806 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
12807 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
12808 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
12809 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
12810 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
12811 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
12812 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
12813 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
12814 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12815 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
12817 — Berlin, 2015-07-07
12821 * The sd-bus.h and sd-event.h APIs have now been declared
12822 stable and have been added to the official interface of
12823 libsystemd.so. sd-bus implements an alternative D-Bus client
12824 library, that is relatively easy to use, very efficient and
12825 supports both classic D-Bus as well as kdbus as transport
12826 backend. sd-event is a generic event loop abstraction that
12827 is built around Linux epoll, but adds features such as event
12828 prioritization or efficient timer handling. Both APIs are good
12829 choices for C programs looking for a bus and/or event loop
12830 implementation that is minimal and does not have to be
12831 portable to other kernels.
12833 * kdbus support is no longer compile-time optional. It is now
12834 always built-in. However, it can still be disabled at
12835 runtime using the kdbus=0 kernel command line setting, and
12836 that setting may be changed to default to off, by specifying
12837 --disable-kdbus at build-time. Note though that the kernel
12838 command line setting has no effect if the kdbus.ko kernel
12839 module is not installed, in which case kdbus is (obviously)
12840 also disabled. We encourage all downstream distributions to
12841 begin testing kdbus by adding it to the kernel images in the
12842 development distributions, and leaving kdbus support in
12845 * The minimal required util-linux version has been bumped to
12848 * Support for chkconfig (--enable-chkconfig) was removed in
12849 favor of calling an abstraction tool
12850 /lib/systemd/systemd-sysv-install. This needs to be
12851 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
12852 in README for details.
12854 * If there's a systemd unit and a SysV init script for the
12855 same service name, and the user executes "systemctl enable"
12856 for it (or a related call), then this will now enable both
12857 (or execute the related operation on both), not just the
12860 * The libudev API documentation has been converted from gtkdoc
12863 * gudev has been removed from the systemd tree, it is now an
12866 * The systemd-cgtop tool learnt a new --raw switch to generate
12867 "raw" (machine parsable) output.
12869 * networkd's IPForwarding= .network file setting learnt the
12870 new setting "kernel", which ensures that networkd does not
12871 change the IP forwarding sysctl from the default kernel
12874 * The systemd-logind bus API now exposes a new boolean
12875 property "Docked" that reports whether logind considers the
12876 system "docked", i.e. connected to a docking station or not.
12878 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
12879 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
12880 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
12881 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
12882 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
12883 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
12884 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
12885 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
12886 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
12887 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
12888 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
12889 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
12890 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
12891 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12892 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
12893 Fink, Zbigniew Jędrzejewski-Szmek
12895 — Berlin, 2015-06-19
12899 * The gudev library has been extracted into a separate repository
12900 available at: https://git.gnome.org/browse/libgudev/
12901 It is now managed as part of the Gnome project. Distributions
12902 are recommended to pass --disable-gudev to systemd and use
12903 gudev from the Gnome project instead. gudev is still included
12904 in systemd, for now. It will be removed soon, though. Please
12905 also see the announcement-thread on systemd-devel:
12906 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
12908 * systemd now exposes a CPUUsageNSec= property for each
12909 service unit on the bus, that contains the overall consumed
12910 CPU time of a service (the sum of what each process of the
12911 service consumed). This value is only available if
12912 CPUAccounting= is turned on for a service, and is then shown
12913 in the "systemctl status" output.
12915 * Support for configuring alternative mappings of the old SysV
12916 runlevels to systemd targets has been removed. They are now
12917 hardcoded in a way that runlevels 2, 3, 4 all map to
12918 multi-user.target and 5 to graphical.target (which
12919 previously was already the default behaviour).
12921 * The auto-mounter logic gained support for mount point
12922 expiry, using a new TimeoutIdleSec= setting in .automount
12923 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
12925 * The EFI System Partition (ESP) as mounted to /boot by
12926 systemd-efi-boot-generator will now be unmounted
12927 automatically after 2 minutes of not being used. This should
12928 minimize the risk of ESP corruptions.
12930 * New /etc/fstab options x-systemd.requires= and
12931 x-systemd.requires-mounts-for= are now supported to express
12932 additional dependencies for mounts. This is useful for
12933 journaling file systems that support external journal
12934 devices or overlay file systems that require underlying file
12935 systems to be mounted.
12937 * systemd does not support direct live-upgrades (via systemctl
12938 daemon-reexec) from versions older than v44 anymore. As no
12939 distribution we are aware of shipped such old versions in a
12940 stable release this should not be problematic.
12942 * When systemd forks off a new per-connection service instance
12943 it will now set the $REMOTE_ADDR environment variable to the
12944 remote IP address, and $REMOTE_PORT environment variable to
12945 the remote IP port. This behaviour is similar to the
12946 corresponding environment variables defined by CGI.
12948 * systemd-networkd gained support for uplink failure
12949 detection. The BindCarrier= option allows binding interface
12950 configuration dynamically to the link sense of other
12951 interfaces. This is useful to achieve behaviour like in
12954 * systemd-networkd gained support for configuring the DHCP
12955 client identifier to use when requesting leases.
12957 * systemd-networkd now has a per-network UseNTP= option to
12958 configure whether NTP server information acquired via DHCP
12959 is passed on to services like systemd-timesyncd.
12961 * systemd-networkd gained support for vti6 tunnels.
12963 * Note that systemd-networkd manages the sysctl variable
12964 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
12965 it is configured for since v219. The variable controls IP
12966 forwarding, and is a per-interface alternative to the global
12967 /proc/sys/net/ipv[46]/ip_forward. This setting is
12968 configurable in the IPForward= option, which defaults to
12969 "no". This means if networkd is used for an interface it is
12970 no longer sufficient to set the global sysctl option to turn
12971 on IP forwarding! Instead, the .network file option
12972 IPForward= needs to be turned on! Note that the
12973 implementation of this behaviour was broken in v219 and has
12974 been fixed in v220.
12976 * Many bonding and vxlan options are now configurable in
12979 * systemd-nspawn gained a new --property= setting to set unit
12980 properties for the container scope. This is useful for
12981 setting resource parameters (e.g. "CPUShares=500") on
12982 containers started from the command line.
12984 * systemd-nspawn gained a new --private-users= switch to make
12985 use of user namespacing available on recent Linux kernels.
12987 * systemd-nspawn may now be called as part of a shell pipeline
12988 in which case the pipes used for stdin and stdout are passed
12989 directly to the process invoked in the container, without
12990 indirection via a pseudo tty.
12992 * systemd-nspawn gained a new switch to control the UNIX
12993 signal to use when killing the init process of the container
12994 when shutting down.
12996 * systemd-nspawn gained a new --overlay= switch for mounting
12997 overlay file systems into the container using the new kernel
13000 * When a container image is imported via systemd-importd and
13001 the host file system is not btrfs, a loopback block device
13002 file is created in /var/lib/machines.raw with a btrfs file
13003 system inside. It is then mounted to /var/lib/machines to
13004 enable btrfs features for container management. The loopback
13005 file and btrfs file system is grown as needed when container
13006 images are imported via systemd-importd.
13008 * systemd-machined/systemd-importd gained support for btrfs
13009 quota, to enforce container disk space limits on disk. This
13010 is exposed in "machinectl set-limit".
13012 * systemd-importd now can import containers from local .tar,
13013 .raw and .qcow2 images, and export them to .tar and .raw. It
13014 can also import dkr v2 images now from the network (on top
13017 * systemd-importd gained support for verifying downloaded
13018 images with gpg2 (previously only gpg1 was supported).
13020 * systemd-machined, systemd-logind, systemd: most bus calls are
13021 now accessible to unprivileged processes via polkit. Also,
13022 systemd-logind will now allow users to kill their own sessions
13023 without further privileges or authorization.
13025 * systemd-shutdownd has been removed. This service was
13026 previously responsible for implementing scheduled shutdowns
13027 as exposed in /usr/bin/shutdown's time parameter. This
13028 functionality has now been moved into systemd-logind and is
13029 accessible via a bus interface.
13031 * "systemctl reboot" gained a new switch --firmware-setup that
13032 can be used to reboot into the EFI firmware setup, if that
13033 is available. systemd-logind now exposes an API on the bus
13034 to trigger such reboots, in case graphical desktop UIs want
13035 to cover this functionality.
13037 * "systemctl enable", "systemctl disable" and "systemctl mask"
13038 now support a new "--now" switch. If specified the units
13039 that are enabled will also be started, and the ones
13040 disabled/masked also stopped.
13042 * The Gummiboot EFI boot loader tool has been merged into
13043 systemd, and renamed to "systemd-boot". The bootctl tool has been
13044 updated to support systemd-boot.
13046 * An EFI kernel stub has been added that may be used to create
13047 kernel EFI binaries that contain not only the actual kernel,
13048 but also an initrd, boot splash, command line and OS release
13049 information. This combined binary can then be signed as a
13050 single image, so that the firmware can verify it all in one
13051 step. systemd-boot has special support for EFI binaries created
13052 like this and can extract OS release information from them
13053 and show them in the boot menu. This functionality is useful
13054 to implement cryptographically verified boot schemes.
13056 * Optional support has been added to systemd-fsck to pass
13057 fsck's progress report to an AF_UNIX socket in the file
13060 * udev will no longer create device symlinks for all block devices by
13061 default. A deny list for excluding special block devices from this
13062 logic has been turned into an allow list that requires picking block
13063 devices explicitly that require device symlinks.
13065 * A new (currently still internal) API sd-device.h has been
13066 added to libsystemd. This modernized API is supposed to
13067 replace libudev eventually. In fact, already much of libudev
13068 is now just a wrapper around sd-device.h.
13070 * A new hwdb database for storing metadata about pointing
13071 stick devices has been added.
13073 * systemd-tmpfiles gained support for setting file attributes
13074 similar to the "chattr" tool with new 'h' and 'H' lines.
13076 * systemd-journald will no longer unconditionally set the
13077 btrfs NOCOW flag on new journal files. This is instead done
13078 with tmpfiles snippet using the new 'h' line type. This
13079 allows easy disabling of this logic, by masking the
13080 journal-nocow.conf tmpfiles file.
13082 * systemd-journald will now translate audit message types to
13083 human readable identifiers when writing them to the
13084 journal. This should improve readability of audit messages.
13086 * The LUKS logic gained support for the offset= and skip=
13087 options in /etc/crypttab, as previously implemented by
13090 * /usr/lib/os-release gained a new optional field VARIANT= for
13091 distributions that support multiple variants (such as a
13092 desktop edition, a server edition, …)
13094 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
13095 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
13096 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
13097 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
13098 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
13099 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
13100 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
13101 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
13102 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
13103 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
13104 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
13105 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
13106 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
13107 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
13108 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
13109 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
13110 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
13111 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
13112 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
13113 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
13114 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
13115 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
13116 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
13117 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
13118 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
13119 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
13120 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
13122 — Berlin, 2015-05-22
13126 * Introduce a new API "sd-hwdb.h" for querying the hardware
13127 metadata database. With this minimal interface one can query
13128 and enumerate the udev hwdb, decoupled from the old libudev
13129 library. libudev's interface for this is now only a wrapper
13130 around sd-hwdb. A new tool systemd-hwdb has been added to
13131 interface with and update the database.
13133 * When any of systemd's tools copies files (for example due to
13134 tmpfiles' C lines) a btrfs reflink will attempted first,
13135 before bytewise copying is done.
13137 * systemd-nspawn gained a new --ephemeral switch. When
13138 specified a btrfs snapshot is taken of the container's root
13139 directory, and immediately removed when the container
13140 terminates again. Thus, a container can be started whose
13141 changes never alter the container's root directory, and are
13142 lost on container termination. This switch can also be used
13143 for starting a container off the root file system of the
13144 host without affecting the host OS. This switch is only
13145 available on btrfs file systems.
13147 * systemd-nspawn gained a new --template= switch. It takes the
13148 path to a container tree to use as template for the tree
13149 specified via --directory=, should that directory be
13150 missing. This allows instantiating containers dynamically,
13151 on first run. This switch is only available on btrfs file
13154 * When a .mount unit refers to a mount point on which multiple
13155 mounts are stacked, and the .mount unit is stopped all of
13156 the stacked mount points will now be unmounted until no
13157 mount point remains.
13159 * systemd now has an explicit notion of supported and
13160 unsupported unit types. Jobs enqueued for unsupported unit
13161 types will now fail with an "unsupported" error code. More
13162 specifically .swap, .automount and .device units are not
13163 supported in containers, .busname units are not supported on
13164 non-kdbus systems. .swap and .automount are also not
13165 supported if their respective kernel compile time options
13168 * machinectl gained support for two new "copy-from" and
13169 "copy-to" commands for copying files from a running
13170 container to the host or vice versa.
13172 * machinectl gained support for a new "bind" command to bind
13173 mount host directories into local containers. This is
13174 currently only supported for nspawn containers.
13176 * networkd gained support for configuring bridge forwarding
13177 database entries (fdb) from .network files.
13179 * A new tiny daemon "systemd-importd" has been added that can
13180 download container images in tar, raw, qcow2 or dkr formats,
13181 and make them available locally in /var/lib/machines, so
13182 that they can run as nspawn containers. The daemon can GPG
13183 verify the downloads (not supported for dkr, since it has no
13184 provisions for verifying downloads). It will transparently
13185 decompress bz2, xz, gzip compressed downloads if necessary,
13186 and restore sparse files on disk. The daemon uses privilege
13187 separation to ensure the actual download logic runs with
13188 fewer privileges than the daemon itself. machinectl has
13189 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
13190 make the functionality of importd available to the
13191 user. With this in place the Fedora and Ubuntu "Cloud"
13192 images can be downloaded and booted as containers unmodified
13193 (the Fedora images lack the appropriate GPG signature files
13194 currently, so they cannot be verified, but this will change
13195 soon, hopefully). Note that downloading images is currently
13196 only fully supported on btrfs.
13198 * machinectl is now able to list container images found in
13199 /var/lib/machines, along with some metadata about sizes of
13200 disk and similar. If the directory is located on btrfs and
13201 quota is enabled, this includes quota display. A new command
13202 "image-status" has been added that shows additional
13203 information about images.
13205 * machinectl is now able to clone container images
13206 efficiently, if the underlying file system (btrfs) supports
13207 it, with the new "machinectl clone" command. It also
13208 gained commands for renaming and removing images, as well as
13209 marking them read-only or read-write (supported also on
13210 legacy file systems).
13212 * networkd gained support for collecting LLDP network
13213 announcements, from hardware that supports this. This is
13214 shown in networkctl output.
13216 * systemd-run gained support for a new -t (--pty) switch for
13217 invoking a binary on a pty whose input and output is
13218 connected to the invoking terminal. This allows executing
13219 processes as system services while interactively
13220 communicating with them via the terminal. Most interestingly
13221 this is supported across container boundaries. Invoking
13222 "systemd-run -t /bin/bash" is an alternative to running a
13223 full login session, the difference being that the former
13224 will not register a session, nor go through the PAM session
13227 * tmpfiles gained support for a new "v" line type for creating
13228 btrfs subvolumes. If the underlying file system is a legacy
13229 file system, this automatically degrades to creating a
13230 normal directory. Among others /var/lib/machines is now
13231 created like this at boot, should it be missing.
13233 * The directory /var/lib/containers/ has been deprecated and
13234 been replaced by /var/lib/machines. The term "machines" has
13235 been used in the systemd context as generic term for both
13236 VMs and containers, and hence appears more appropriate for
13237 this, as the directory can also contain raw images bootable
13240 * systemd-nspawn when invoked with -M but without --directory=
13241 or --image= is now capable of searching for the container
13242 root directory, subvolume or disk image automatically, in
13243 /var/lib/machines. systemd-nspawn@.service has been updated
13244 to make use of this, thus allowing it to be used for raw
13247 * A new machines.target unit has been introduced that is
13248 supposed to group all containers/VMs invoked as services on
13249 the system. systemd-nspawn@.service has been updated to
13250 integrate with that.
13252 * machinectl gained a new "start" command, for invoking a
13253 container as a service. "machinectl start foo" is mostly
13254 equivalent to "systemctl start systemd-nspawn@foo.service",
13255 but handles escaping in a nicer way.
13257 * systemd-nspawn will now mount most of the cgroupfs tree
13258 read-only into each container, with the exception of the
13259 container's own subtree in the name=systemd hierarchy.
13261 * journald now sets the special FS_NOCOW file flag for its
13262 journal files. This should improve performance on btrfs, by
13263 avoiding heavy fragmentation when journald's write-pattern
13264 is used on COW file systems. It degrades btrfs' data
13265 integrity guarantees for the files to the same levels as for
13266 ext3/ext4 however. This should be OK though as journald does
13267 its own data integrity checks and all its objects are
13268 checksummed on disk. Also, journald should handle btrfs disk
13269 full events a lot more gracefully now, by processing SIGBUS
13270 errors, and not relying on fallocate() anymore.
13272 * When journald detects that journal files it is writing to
13273 have been deleted it will immediately start new journal
13276 * systemd now provides a way to store file descriptors
13277 per-service in PID 1. This is useful for daemons to ensure
13278 that fds they require are not lost during a daemon
13279 restart. The fds are passed to the daemon on the next
13280 invocation in the same way socket activation fds are
13281 passed. This is now used by journald to ensure that the
13282 various sockets connected to all the system's stdout/stderr
13283 are not lost when journald is restarted. File descriptors
13284 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
13285 an extension to sd_notify(). Note that a limit is enforced
13286 on the number of fds a service can store in PID 1, and it
13287 defaults to 0, so that no fds may be stored, unless this is
13288 explicitly turned on.
13290 * The default TERM variable to use for units connected to a
13291 terminal, when no other value is explicitly is set is now
13292 vt220 rather than vt102. This should be fairly safe still,
13293 but allows PgUp/PgDn work.
13295 * The /etc/crypttab option header= as known from Debian is now
13298 * "loginctl user-status" and "loginctl session-status" will
13299 now show the last 10 lines of log messages of the
13300 user/session following the status output. Similar,
13301 "machinectl status" will show the last 10 log lines
13302 associated with a virtual machine or container
13303 service. (Note that this is usually not the log messages
13304 done in the VM/container itself, but simply what the
13305 container manager logs. For nspawn this includes all console
13308 * "loginctl session-status" without further argument will now
13309 show the status of the session of the caller. Similar,
13310 "lock-session", "unlock-session", "activate",
13311 "enable-linger", "disable-linger" may now be called without
13312 session/user parameter in which case they apply to the
13313 caller's session/user.
13315 * An X11 session scriptlet is now shipped that uploads
13316 $DISPLAY and $XAUTHORITY into the environment of the systemd
13317 --user daemon if a session begins. This should improve
13318 compatibility with X11 enabled applications run as systemd
13321 * Generators are now subject to masking via /etc and /run, the
13322 same way as unit files.
13324 * networkd .network files gained support for configuring
13325 per-link IPv4/IPv6 packet forwarding as well as IPv4
13326 masquerading. This is by default turned on for veth links to
13327 containers, as registered by systemd-nspawn. This means that
13328 nspawn containers run with --network-veth will now get
13329 automatic routed access to the host's networks without any
13330 further configuration or setup, as long as networkd runs on
13333 * systemd-nspawn gained the --port= (-p) switch to expose TCP
13334 or UDP posts of a container on the host. With this in place
13335 it is possible to run containers with private veth links
13336 (--network-veth), and have their functionality exposed on
13337 the host as if their services were running directly on the
13340 * systemd-nspawn's --network-veth switch now gained a short
13341 version "-n", since with the changes above it is now truly
13342 useful out-of-the-box. The systemd-nspawn@.service has been
13343 updated to make use of it too by default.
13345 * systemd-nspawn will now maintain a per-image R/W lock, to
13346 ensure that the same image is not started more than once
13347 writable. (It's OK to run an image multiple times
13348 simultaneously in read-only mode.)
13350 * systemd-nspawn's --image= option is now capable of
13351 dissecting and booting MBR and GPT disk images that contain
13352 only a single active Linux partition. Previously it
13353 supported only GPT disk images with proper GPT type
13354 IDs. This allows running cloud images from major
13355 distributions directly with systemd-nspawn, without
13358 * In addition to collecting mouse dpi data in the udev
13359 hardware database, there's now support for collecting angle
13360 information for mouse scroll wheels. The database is
13361 supposed to guarantee similar scrolling behavior on mice
13362 that it knows about. There's also support for collecting
13363 information about Touchpad types.
13365 * udev's input_id built-in will now also collect touch screen
13366 dimension data and attach it to probed devices.
13368 * /etc/os-release gained support for a Distribution Privacy
13371 * networkd gained support for creating "ipvlan", "gretap",
13372 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
13374 * systemd-tmpfiles gained support for "a" lines for setting
13377 * systemd-nspawn will now mount /tmp in the container to
13378 tmpfs, automatically.
13380 * systemd now exposes the memory.usage_in_bytes cgroup
13381 attribute and shows it for each service in the "systemctl
13382 status" output, if available.
13384 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
13385 immediate reboot is triggered. This useful if shutdown is
13386 hung and is unable to complete, to expedite the
13387 operation. Note that this kind of reboot will still unmount
13388 all file systems, and hence should not result in fsck being
13389 run on next reboot.
13391 * A .device unit for an optical block device will now be
13392 considered active only when a medium is in the drive. Also,
13393 mount units are now bound to their backing devices thus
13394 triggering automatic unmounting when devices become
13395 unavailable. With this in place systemd will now
13396 automatically unmount left-over mounts when a CD-ROM is
13397 ejected or a USB stick is yanked from the system.
13399 * networkd-wait-online now has support for waiting for
13400 specific interfaces only (with globbing), and for giving up
13401 after a configurable timeout.
13403 * networkd now exits when idle. It will be automatically
13404 restarted as soon as interfaces show up, are removed or
13405 change state. networkd will stay around as long as there is
13406 at least one DHCP state machine or similar around, that keep
13409 * networkd may now configure IPv6 link-local addressing in
13410 addition to IPv4 link-local addressing.
13412 * The IPv6 "token" for use in SLAAC may now be configured for
13413 each .network interface in networkd.
13415 * Routes configured with networkd may now be assigned a scope
13418 * networkd's [Match] sections now support globbing and lists
13419 of multiple space-separated matches per item.
13421 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
13422 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
13423 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
13424 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
13425 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
13426 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
13427 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
13428 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
13429 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
13430 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
13431 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
13432 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
13433 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
13434 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
13435 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
13436 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
13437 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
13438 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
13439 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
13440 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
13441 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
13442 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
13443 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
13444 Hoffmann, Zbigniew Jędrzejewski-Szmek
13446 — Berlin, 2015-02-16
13450 * When querying unit file enablement status (for example via
13451 "systemctl is-enabled"), a new state "indirect" is now known
13452 which indicates that a unit might not be enabled itself, but
13453 another unit listed in its Also= setting might be.
13455 * Similar to the various existing ConditionXYZ= settings for
13456 units, there are now matching AssertXYZ= settings. While
13457 failing conditions cause a unit to be skipped, but its job
13458 to succeed, failing assertions declared like this will cause
13459 a unit start operation and its job to fail.
13461 * hostnamed now knows a new chassis type "embedded".
13463 * systemctl gained a new "edit" command. When used on a unit
13464 file, this allows extending unit files with .d/ drop-in
13465 configuration snippets or editing the full file (after
13466 copying it from /usr/lib to /etc). This will invoke the
13467 user's editor (as configured with $EDITOR), and reload the
13468 modified configuration after editing.
13470 * "systemctl status" now shows the suggested enablement state
13471 for a unit, as declared in the (usually vendor-supplied)
13472 system preset files.
13474 * nss-myhostname will now resolve the single-label hostname
13475 "gateway" to the locally configured default IP routing
13476 gateways, ordered by their metrics. This assigns a stable
13477 name to the used gateways, regardless which ones are
13478 currently configured. Note that the name will only be
13479 resolved after all other name sources (if nss-myhostname is
13480 configured properly) and should hence not negatively impact
13481 systems that use the single-label hostname "gateway" in
13484 * systemd-inhibit now allows filtering by mode when listing
13487 * Scope and service units gained a new "Delegate" boolean
13488 property, which, when set, allows processes running inside the
13489 unit to further partition resources. This is primarily
13490 useful for systemd user instances as well as container
13493 * journald will now pick up audit messages directly from
13494 the kernel, and log them like any other log message. The
13495 audit fields are split up and fully indexed. This means that
13496 journalctl in many ways is now a (nicer!) alternative to
13497 ausearch, the traditional audit client. Note that this
13498 implements only a minimal audit client. If you want the
13499 special audit modes like reboot-on-log-overflow, please use
13500 the traditional auditd instead, which can be used in
13501 parallel to journald.
13503 * The ConditionSecurity= unit file option now understands the
13504 special string "audit" to check whether auditing is
13507 * journalctl gained two new commands --vacuum-size= and
13508 --vacuum-time= to delete old journal files until the
13509 remaining ones take up no more than the specified size on disk,
13510 or are not older than the specified time.
13512 * A new, native PPPoE library has been added to sd-network,
13513 systemd's library of light-weight networking protocols. This
13514 library will be used in a future version of networkd to
13515 enable PPPoE communication without an external pppd daemon.
13517 * The busctl tool now understands a new "capture" verb that
13518 works similar to "monitor", but writes a packet capture
13519 trace to STDOUT that can be redirected to a file which is
13520 compatible with libcap's capture file format. This can then
13521 be loaded in Wireshark and similar tools to inspect bus
13524 * The busctl tool now understands a new "tree" verb that shows
13525 the object trees of a specific service on the bus, or of all
13528 * The busctl tool now understands a new "introspect" verb that
13529 shows all interfaces and members of objects on the bus,
13530 including their signature and values. This is particularly
13531 useful to get more information about bus objects shown by
13532 the new "busctl tree" command.
13534 * The busctl tool now understands new verbs "call",
13535 "set-property" and "get-property" for invoking bus method
13536 calls, setting and getting bus object properties in a
13539 * busctl gained a new --augment-creds= argument that controls
13540 whether the tool shall augment credential information it
13541 gets from the bus with data from /proc, in a possibly
13544 * nspawn's --link-journal= switch gained two new values
13545 "try-guest" and "try-host" that work like "guest" and
13546 "host", but do not fail if the host has no persistent
13547 journaling enabled. -j is now equivalent to
13548 --link-journal=try-guest.
13550 * macvlan network devices created by nspawn will now have
13551 stable MAC addresses.
13553 * A new SmackProcessLabel= unit setting has been added, which
13554 controls the SMACK security label processes forked off by
13555 the respective unit shall use.
13557 * If compiled with --enable-xkbcommon, systemd-localed will
13558 verify x11 keymap settings by compiling the given keymap. It
13559 will spew out warnings if the compilation fails. This
13560 requires libxkbcommon to be installed.
13562 * When a coredump is collected, a larger number of metadata
13563 fields is now collected and included in the journal records
13564 created for it. More specifically, control group membership,
13565 environment variables, memory maps, working directory,
13566 chroot directory, /proc/$PID/status, and a list of open file
13567 descriptors is now stored in the log entry.
13569 * The udev hwdb now contains DPI information for mice. For
13572 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
13574 * All systemd programs that read standalone configuration
13575 files in /etc now also support a corresponding series of
13576 .conf.d configuration directories in /etc/, /run/,
13577 /usr/local/lib/, /usr/lib/, and (if configured with
13578 --enable-split-usr) /lib/. In particular, the following
13579 configuration files now have corresponding configuration
13580 directories: system.conf user.conf, logind.conf,
13581 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
13582 resolved.conf, timesyncd.conf, journal-remote.conf, and
13583 journal-upload.conf. Note that distributions should use the
13584 configuration directories in /usr/lib/; the directories in
13585 /etc/ are reserved for the system administrator.
13587 * systemd-rfkill will no longer take the rfkill device name
13588 into account when storing rfkill state on disk, as the name
13589 might be dynamically assigned and not stable. Instead, the
13590 ID_PATH udev variable combined with the rfkill type (wlan,
13591 bluetooth, …) is used.
13593 * A new service systemd-machine-id-commit.service has been
13594 added. When used on systems where /etc is read-only during
13595 boot, and /etc/machine-id is not initialized (but an empty
13596 file), this service will copy the temporary machine ID
13597 created as replacement into /etc after the system is fully
13598 booted up. This is useful for systems that are freshly
13599 installed with a non-initialized machine ID, but should get
13600 a fixed machine ID for subsequent boots.
13602 * networkd's .netdev files now provide a large set of
13603 configuration parameters for VXLAN devices. Similarly, the
13604 bridge port cost parameter is now configurable in .network
13605 files. There's also new support for configuring IP source
13606 routing. networkd .link files gained support for a new
13607 OriginalName= match that is useful to match against the
13608 original interface name the kernel assigned. .network files
13609 may include MTU= and MACAddress= fields for altering the MTU
13610 and MAC address while being connected to a specific network
13613 * The LUKS logic gained supported for configuring
13614 UUID-specific key files. There's also new support for naming
13615 LUKS device from the kernel command line, using the new
13616 luks.name= argument.
13618 * Timer units may now be transiently created via the bus API
13619 (this was previously already available for scope and service
13620 units). In addition it is now possible to create multiple
13621 transient units at the same time with a single bus call. The
13622 "systemd-run" tool has been updated to make use of this for
13623 running commands on a specified time, in at(1)-style.
13625 * tmpfiles gained support for "t" lines, for assigning
13626 extended attributes to files. Among other uses this may be
13627 used to assign SMACK labels to files.
13629 Contributions from: Alin Rauta, Alison Chaiken, Andrej
13630 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
13631 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
13632 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
13633 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
13634 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
13635 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
13636 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
13637 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
13638 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
13639 Michael Biebl, Michael Chapman, Michael Marineau, Michal
13640 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
13641 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
13642 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
13643 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
13644 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
13645 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
13646 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
13648 — Berlin, 2014-12-10
13652 * journalctl gained the new options -t/--identifier= to match
13653 on the syslog identifier (aka "tag"), as well as --utc to
13654 show log timestamps in the UTC timezone. journalctl now also
13655 accepts -n/--lines=all to disable line capping in a pager.
13657 * journalctl gained a new switch, --flush, that synchronously
13658 flushes logs from /run/log/journal to /var/log/journal if
13659 persistent storage is enabled. systemd-journal-flush.service
13660 now waits until the operation is complete.
13662 * Services can notify the manager before they start a reload
13663 (by sending RELOADING=1) or shutdown (by sending
13664 STOPPING=1). This allows the manager to track and show the
13665 internal state of daemons and closes a race condition when
13666 the process is still running but has closed its D-Bus
13669 * Services with Type=oneshot do not have to have any ExecStart
13672 * User units are now loaded also from
13673 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
13674 /run/systemd/user directory that was already previously
13675 supported, but is under the control of the user.
13677 * Job timeouts (i.e. timeouts on the time a job that is
13678 queued stays in the run queue) can now optionally result in
13679 immediate reboot or power-off actions (JobTimeoutAction= and
13680 JobTimeoutRebootArgument=). This is useful on ".target"
13681 units, to limit the maximum time a target remains
13682 undispatched in the run queue, and to trigger an emergency
13683 operation in such a case. This is now used by default to
13684 turn off the system if boot-up (as defined by everything in
13685 basic.target) hangs and does not complete for at least
13686 15min. Also, if power-off or reboot hang for at least 30min
13687 an immediate power-off/reboot operation is triggered. This
13688 functionality is particularly useful to increase reliability
13689 on embedded devices, but also on laptops which might
13690 accidentally get powered on when carried in a backpack and
13691 whose boot stays stuck in a hard disk encryption passphrase
13694 * systemd-logind can be configured to also handle lid switch
13695 events even when the machine is docked or multiple displays
13696 are attached (HandleLidSwitchDocked= option).
13698 * A helper binary and a service have been added which can be
13699 used to resume from hibernation in the initramfs. A
13700 generator will parse the resume= option on the kernel
13701 command line to trigger resume.
13703 * A user console daemon systemd-consoled has been
13704 added. Currently, it is a preview, and will so far open a
13705 single terminal on each session of the user marked as
13706 Desktop=systemd-console.
13708 * Route metrics can be specified for DHCP routes added by
13711 * The SELinux context of socket-activated services can be set
13712 from the information provided by the networking stack
13713 (SELinuxContextFromNet= option).
13715 * Userspace firmware loading support has been removed and
13716 the minimum supported kernel version is thus bumped to 3.7.
13718 * Timeout for udev workers has been increased from 1 to 3
13719 minutes, but a warning will be printed after 1 minute to
13720 help diagnose kernel modules that take a long time to load.
13722 * Udev rules can now remove tags on devices with TAG-="foobar".
13724 * systemd's readahead implementation has been removed. In many
13725 circumstances it didn't give expected benefits even for
13726 rotational disk drives and was becoming less relevant in the
13727 age of SSDs. As none of the developers has been using
13728 rotating media anymore, and nobody stepped up to actively
13729 maintain this component of systemd it has now been removed.
13731 * Swap units can use Options= to specify discard options.
13732 Discard options specified for swaps in /etc/fstab are now
13735 * Docker containers are now detected as a separate type of
13738 * The Password Agent protocol gained support for queries where
13739 the user input is shown, useful e.g. for user names.
13740 systemd-ask-password gained a new --echo option to turn that
13743 * The default sysctl.d/ snippets will now set:
13745 net.core.default_qdisc = fq_codel
13747 This selects Fair Queuing Controlled Delay as the default
13748 queuing discipline for network interfaces. fq_codel helps
13749 fight the network bufferbloat problem. It is believed to be
13750 a good default with no tuning required for most workloads.
13751 Downstream distributions may override this choice. On 10Gbit
13752 servers that do not do forwarding, "fq" may perform better.
13753 Systems without a good clocksource should use "pfifo_fast".
13755 * If kdbus is enabled during build a new option BusPolicy= is
13756 available for service units, that allows locking all service
13757 processes into a stricter bus policy, in order to limit
13758 access to various bus services, or even hide most of them
13759 from the service's view entirely.
13761 * networkctl will now show the .network and .link file
13762 networkd has applied to a specific interface.
13764 * sd-login gained a new API call sd_session_get_desktop() to
13765 query which desktop environment has been selected for a
13768 * UNIX utmp support is now compile-time optional to support
13769 legacy-free systems.
13771 * systemctl gained two new commands "add-wants" and
13772 "add-requires" for pulling in units from specific targets
13775 * If the word "rescue" is specified on the kernel command line
13776 the system will now boot into rescue mode (aka
13777 rescue.target), which was previously available only by
13778 specifying "1" or "systemd.unit=rescue.target" on the kernel
13779 command line. This new kernel command line option nicely
13780 mirrors the already existing "emergency" kernel command line
13783 * New kernel command line options mount.usr=, mount.usrflags=,
13784 mount.usrfstype= have been added that match root=, rootflags=,
13785 rootfstype= but allow mounting a specific file system to
13788 * The $NOTIFY_SOCKET is now also passed to control processes of
13789 services, not only the main process.
13791 * This version reenables support for fsck's -l switch. This
13792 means at least version v2.25 of util-linux is required for
13793 operation, otherwise dead-locks on device nodes may
13794 occur. Again: you need to update util-linux to at least
13795 v2.25 when updating systemd to v217.
13797 * The "multi-seat-x" tool has been removed from systemd, as
13798 its functionality has been integrated into X servers 1.16,
13799 and the tool is hence redundant. It is recommended to update
13800 display managers invoking this tool to simply invoke X
13801 directly from now on, again.
13803 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
13804 message flag has been added for all of systemd's polkit
13805 authenticated method calls has been added. In particular this
13806 now allows optional interactive authorization via polkit for
13807 many of PID1's privileged operations such as unit file
13808 enabling and disabling.
13810 * "udevadm hwdb --update" learnt a new switch "--usr" for
13811 placing the rebuilt hardware database in /usr instead of
13812 /etc. When used only hardware database entries stored in
13813 /usr will be used, and any user database entries in /etc are
13814 ignored. This functionality is useful for vendors to ship a
13815 pre-built database on systems where local configuration is
13816 unnecessary or unlikely.
13818 * Calendar time specifications in .timer units now also
13819 understand the strings "semi-annually", "quarterly" and
13820 "minutely" as shortcuts (in addition to the preexisting
13821 "annually", "hourly", …).
13823 * systemd-tmpfiles will now correctly create files in /dev
13824 at boot which are marked for creation only at boot. It is
13825 recommended to always create static device nodes with 'c!'
13826 and 'b!', so that they are created only at boot and not
13827 overwritten at runtime.
13829 * When the watchdog logic is used for a service (WatchdogSec=)
13830 and the watchdog timeout is hit the service will now be
13831 terminated with SIGABRT (instead of just SIGTERM), in order
13832 to make sure a proper coredump and backtrace is
13833 generated. This ensures that hanging services will result in
13834 similar coredump/backtrace behaviour as services that hit a
13835 segmentation fault.
13837 Contributions from: Andreas Henriksson, Andrei Borzenkov,
13838 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
13839 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
13840 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
13841 Herrmann, David Sommerseth, David Strauss, Emil Renner
13842 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
13843 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
13844 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
13845 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
13846 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
13847 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
13848 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
13849 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
13850 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
13851 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
13852 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
13853 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
13854 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
13855 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
13856 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
13857 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
13860 — Berlin, 2014-10-28
13864 * timedated no longer reads NTP implementation unit names from
13865 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
13866 implementations should add a
13868 Conflicts=systemd-timesyncd.service
13870 to their unit files to take over and replace systemd's NTP
13871 default functionality.
13873 * systemd-sysusers gained a new line type "r" for configuring
13874 which UID/GID ranges to allocate system users/groups
13875 from. Lines of type "u" may now add an additional column
13876 that specifies the home directory for the system user to be
13877 created. Also, systemd-sysusers may now optionally read user
13878 information from STDIN instead of a file. This is useful for
13879 invoking it from RPM preinst scriptlets that need to create
13880 users before the first RPM file is installed since these
13881 files might need to be owned by them. A new
13882 %sysusers_create_inline RPM macro has been introduced to do
13883 just that. systemd-sysusers now updates the shadow files as
13884 well as the user/group databases, which should enhance
13885 compatibility with certain tools like grpck.
13887 * A number of bus APIs of PID 1 now optionally consult polkit to
13888 permit access for otherwise unprivileged clients under certain
13889 conditions. Note that this currently doesn't support
13890 interactive authentication yet, but this is expected to be
13891 added eventually, too.
13893 * /etc/machine-info now has new fields for configuring the
13894 deployment environment of the machine, as well as the
13895 location of the machine. hostnamectl has been updated with
13896 new command to update these fields.
13898 * systemd-timesyncd has been updated to automatically acquire
13899 NTP server information from systemd-networkd, which might
13900 have been discovered via DHCP.
13902 * systemd-resolved now includes a caching DNS stub resolver
13903 and a complete LLMNR name resolution implementation. A new
13904 NSS module "nss-resolve" has been added which can be used
13905 instead of glibc's own "nss-dns" to resolve hostnames via
13906 systemd-resolved. Hostnames, addresses and arbitrary RRs may
13907 be resolved via systemd-resolved D-Bus APIs. In contrast to
13908 the glibc internal resolver systemd-resolved is aware of
13909 multi-homed system, and keeps DNS server and caches separate
13910 and per-interface. Queries are sent simultaneously on all
13911 interfaces that have DNS servers configured, in order to
13912 properly handle VPNs and local LANs which might resolve
13913 separate sets of domain names. systemd-resolved may acquire
13914 DNS server information from systemd-networkd automatically,
13915 which in turn might have discovered them via DHCP. A tool
13916 "systemd-resolve-host" has been added that may be used to
13917 query the DNS logic in resolved. systemd-resolved implements
13918 IDNA and automatically uses IDNA or UTF-8 encoding depending
13919 on whether classic DNS or LLMNR is used as transport. In the
13920 next releases we intend to add a DNSSEC and mDNS/DNS-SD
13921 implementation to systemd-resolved.
13923 * A new NSS module nss-mymachines has been added, that
13924 automatically resolves the names of all local registered
13925 containers to their respective IP addresses.
13927 * A new client tool "networkctl" for systemd-networkd has been
13928 added. It currently is entirely passive and will query
13929 networking configuration from udev, rtnetlink and networkd,
13930 and present it to the user in a very friendly
13931 way. Eventually, we hope to extend it to become a full
13932 control utility for networkd.
13934 * .socket units gained a new DeferAcceptSec= setting that
13935 controls the kernels' TCP_DEFER_ACCEPT sockopt for
13936 TCP. Similarly, support for controlling TCP keep-alive
13937 settings has been added (KeepAliveTimeSec=,
13938 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
13939 turning off Nagle's algorithm on TCP has been added
13942 * logind learned a new session type "web", for use in projects
13943 like Cockpit which register web clients as PAM sessions.
13945 * timer units with at least one OnCalendar= setting will now
13946 be started only after time-sync.target has been
13947 reached. This way they will not elapse before the system
13948 clock has been corrected by a local NTP client or
13949 similar. This is particular useful on RTC-less embedded
13950 machines, that come up with an invalid system clock.
13952 * systemd-nspawn's --network-veth= switch should now result in
13953 stable MAC addresses for both the outer and the inner side
13956 * systemd-nspawn gained a new --volatile= switch for running
13957 container instances with /etc or /var unpopulated.
13959 * The kdbus client code has been updated to use the new Linux
13960 3.17 memfd subsystem instead of the old kdbus-specific one.
13962 * systemd-networkd's DHCP client and server now support
13963 FORCERENEW. There are also new configuration options to
13964 configure the vendor client identifier and broadcast mode
13967 * systemd will no longer inform the kernel about the current
13968 timezone, as this is necessarily incorrect and racy as the
13969 kernel has no understanding of DST and similar
13970 concepts. This hence means FAT timestamps will be always
13971 considered UTC, similar to what Android is already
13972 doing. Also, when the RTC is configured to the local time
13973 (rather than UTC) systemd will never synchronize back to it,
13974 as this might confuse Windows at a later boot.
13976 * systemd-analyze gained a new command "verify" for offline
13977 validation of unit files.
13979 * systemd-networkd gained support for a couple of additional
13980 settings for bonding networking setups. Also, the metric for
13981 statically configured routes may now be configured. For
13982 network interfaces where this is appropriate the peer IP
13983 address may now be configured.
13985 * systemd-networkd's DHCP client will no longer request
13986 broadcasting by default, as this tripped up some networks.
13987 For hardware where broadcast is required the feature should
13988 be switched back on using RequestBroadcast=yes.
13990 * systemd-networkd will now set up IPv4LL addresses (when
13991 enabled) even if DHCP is configured successfully.
13993 * udev will now default to respect network device names given
13994 by the kernel when the kernel indicates that these are
13995 predictable. This behavior can be tweaked by changing
13996 NamePolicy= in the relevant .link file.
13998 * A new library systemd-terminal has been added that
13999 implements full TTY stream parsing and rendering. This
14000 library is supposed to be used later on for implementing a
14001 full userspace VT subsystem, replacing the current kernel
14004 * A new tool systemd-journal-upload has been added to push
14005 journal data to a remote system running
14006 systemd-journal-remote.
14008 * journald will no longer forward all local data to another
14009 running syslog daemon. This change has been made because
14010 rsyslog (which appears to be the most commonly used syslog
14011 implementation these days) no longer makes use of this, and
14012 instead pulls the data out of the journal on its own. Since
14013 forwarding the messages to a non-existent syslog server is
14014 more expensive than we assumed we have now turned this
14015 off. If you run a syslog server that is not a recent rsyslog
14016 version, you have to turn this option on again
14017 (ForwardToSyslog= in journald.conf).
14019 * journald now optionally supports the LZ4 compressor for
14020 larger journal fields. This compressor should perform much
14021 better than XZ which was the previous default.
14023 * machinectl now shows the IP addresses of local containers,
14024 if it knows them, plus the interface name of the container.
14026 * A new tool "systemd-escape" has been added that makes it
14027 easy to escape strings to build unit names and similar.
14029 * sd_notify() messages may now include a new ERRNO= field
14030 which is parsed and collected by systemd and shown among the
14031 "systemctl status" output for a service.
14033 * A new component "systemd-firstboot" has been added that
14034 queries the most basic systemd information (timezone,
14035 hostname, root password) interactively on first
14036 boot. Alternatively it may also be used to provision these
14037 things offline on OS images installed into directories.
14039 * The default sysctl.d/ snippets will now set
14041 net.ipv4.conf.default.promote_secondaries=1
14043 This has the benefit of no flushing secondary IP addresses
14044 when primary addresses are removed.
14046 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
14047 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
14048 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
14049 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
14050 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
14051 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
14052 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
14053 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
14054 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
14055 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
14056 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
14057 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
14058 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
14059 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
14060 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
14062 — Berlin, 2014-08-19
14066 * A new tool systemd-sysusers has been added. This tool
14067 creates system users and groups in /etc/passwd and
14068 /etc/group, based on static declarative system user/group
14069 definitions in /usr/lib/sysusers.d/. This is useful to
14070 enable factory resets and volatile systems that boot up with
14071 an empty /etc directory, and thus need system users and
14072 groups created during early boot. systemd now also ships
14073 with two default sysusers.d/ files for the most basic
14074 users and groups systemd and the core operating system
14077 * A new tmpfiles snippet has been added that rebuilds the
14078 essential files in /etc on boot, should they be missing.
14080 * A directive for ensuring automatic clean-up of
14081 /var/cache/man/ has been removed from the default
14082 configuration. This line should now be shipped by the man
14083 implementation. The necessary change has been made to the
14084 man-db implementation. Note that you need to update your man
14085 implementation to one that ships this line, otherwise no
14086 automatic clean-up of /var/cache/man will take place.
14088 * A new condition ConditionNeedsUpdate= has been added that
14089 may conditionalize services to only run when /etc or /var
14090 are "older" than the vendor operating system resources in
14091 /usr. This is useful for reconstructing or updating /etc
14092 after an offline update of /usr or a factory reset, on the
14093 next reboot. Services that want to run once after such an
14094 update or reset should use this condition and order
14095 themselves before the new systemd-update-done.service, which
14096 will mark the two directories as fully updated. A number of
14097 service files have been added making use of this, to rebuild
14098 the udev hardware database, the journald message catalog and
14099 dynamic loader cache (ldconfig). The systemd-sysusers tool
14100 described above also makes use of this now. With this in
14101 place it is now possible to start up a minimal operating
14102 system with /etc empty cleanly. For more information on the
14103 concepts involved see this recent blog story:
14105 https://0pointer.de/blog/projects/stateless.html
14107 * A new system group "input" has been introduced, and all
14108 input device nodes get this group assigned. This is useful
14109 for system-level software to get access to input devices. It
14110 complements what is already done for "audio" and "video".
14112 * systemd-networkd learnt minimal DHCPv4 server support in
14113 addition to the existing DHCPv4 client support. It also
14114 learnt DHCPv6 client and IPv6 Router Solicitation client
14115 support. The DHCPv4 client gained support for static routes
14116 passed in from the server. Note that the [DHCPv4] section
14117 known in older systemd-networkd versions has been renamed to
14118 [DHCP] and is now also used by the DHCPv6 client. Existing
14119 .network files using settings of this section should be
14120 updated, though compatibility is maintained. Optionally, the
14121 client hostname may now be sent to the DHCP server.
14123 * networkd gained support for vxlan virtual networks as well
14124 as tun/tap and dummy devices.
14126 * networkd gained support for automatic allocation of address
14127 ranges for interfaces from a system-wide pool of
14128 addresses. This is useful for dynamically managing a large
14129 number of interfaces with a single network configuration
14130 file. In particular this is useful to easily assign
14131 appropriate IP addresses to the veth links of a large number
14132 of nspawn instances.
14134 * RPM macros for processing sysusers, sysctl and binfmt
14135 drop-in snippets at package installation time have been
14138 * The /etc/os-release file should now be placed in
14139 /usr/lib/os-release. The old location is automatically
14140 created as symlink. /usr/lib is the more appropriate
14141 location of this file, since it shall actually describe the
14142 vendor operating system shipped in /usr, and not the
14143 configuration stored in /etc.
14145 * .mount units gained a new boolean SloppyOptions= setting
14146 that maps to mount(8)'s -s option which enables permissive
14147 parsing of unknown mount options.
14149 * tmpfiles learnt a new "L+" directive which creates a symlink
14150 but (unlike "L") deletes a pre-existing file first, should
14151 it already exist and not already be the correct
14152 symlink. Similarly, "b+", "c+" and "p+" directives have been
14153 added as well, which create block and character devices, as
14154 well as fifos in the filesystem, possibly removing any
14155 pre-existing files of different types.
14157 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
14158 'argument' field (which so far specified the source to
14159 symlink/copy the files from) is now optional. If omitted the
14160 same file os copied from /usr/share/factory/ suffixed by the
14161 full destination path. This is useful for populating /etc
14162 with essential files, by copying them from vendor defaults
14163 shipped in /usr/share/factory/etc.
14165 * A new command "systemctl preset-all" has been added that
14166 applies the service preset settings to all installed unit
14167 files. A new switch --preset-mode= has been added that
14168 controls whether only enable or only disable operations
14171 * A new command "systemctl is-system-running" has been added
14172 that allows checking the overall state of the system, for
14173 example whether it is fully up and running.
14175 * When the system boots up with an empty /etc, the equivalent
14176 to "systemctl preset-all" is executed during early boot, to
14177 make sure all default services are enabled after a factory
14180 * systemd now contains a minimal preset file that enables the
14181 most basic services systemd ships by default.
14183 * Unit files' [Install] section gained a new DefaultInstance=
14184 field for defining the default instance to create if a
14185 template unit is enabled with no instance specified.
14187 * A new passive target cryptsetup-pre.target has been added
14188 that may be used by services that need to make they run and
14189 finish before the first LUKS cryptographic device is set up.
14191 * The /dev/loop-control and /dev/btrfs-control device nodes
14192 are now owned by the "disk" group by default, opening up
14193 access to this group.
14195 * systemd-coredump will now automatically generate a
14196 stack trace of all core dumps taking place on the system,
14197 based on elfutils' libdw library. This stack trace is logged
14200 * systemd-coredump may now optionally store coredumps directly
14201 on disk (in /var/lib/systemd/coredump, possibly compressed),
14202 instead of storing them unconditionally in the journal. This
14203 mode is the new default. A new configuration file
14204 /etc/systemd/coredump.conf has been added to configure this
14205 and other parameters of systemd-coredump.
14207 * coredumpctl gained a new "info" verb to show details about a
14208 specific coredump. A new switch "-1" has also been added
14209 that makes sure to only show information about the most
14210 recent entry instead of all entries. Also, as the tool is
14211 generally useful now the "systemd-" prefix of the binary
14212 name has been removed. Distributions that want to maintain
14213 compatibility with the old name should add a symlink from
14214 the old name to the new name.
14216 * journald's SplitMode= now defaults to "uid". This makes sure
14217 that unprivileged users can access their own coredumps with
14218 coredumpctl without restrictions.
14220 * New kernel command line options "systemd.wants=" (for
14221 pulling an additional unit during boot), "systemd.mask="
14222 (for masking a specific unit for the boot), and
14223 "systemd.debug-shell" (for enabling the debug shell on tty9)
14224 have been added. This is implemented in the new generator
14225 "systemd-debug-generator".
14227 * systemd-nspawn will now by default filter a couple of
14228 syscalls for containers, among them those required for
14229 kernel module loading, direct x86 IO port access, swap
14230 management, and kexec. Most importantly though
14231 open_by_handle_at() is now prohibited for containers,
14232 closing a hole similar to a recently discussed vulnerability
14233 in docker regarding access to files on file hierarchies the
14234 container should normally not have access to. Note that, for
14235 nspawn, we generally make no security claims anyway (and
14236 this is explicitly documented in the man page), so this is
14237 just a fix for one of the most obvious problems.
14239 * A new man page file-hierarchy(7) has been added that
14240 contains a minimized, modernized version of the file system
14241 layout systemd expects, similar in style to the FHS
14242 specification or hier(5). A new tool systemd-path(1) has
14243 been added to query many of these paths for the local
14246 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
14247 longer done. Since the directory now has a per-user size
14248 limit, and is cleaned on logout this appears unnecessary,
14249 in particular since this now brings the lifecycle of this
14250 directory closer in line with how IPC objects are handled.
14252 * systemd.pc now exports a number of additional directories,
14253 including $libdir (which is useful to identify the library
14254 path for the primary architecture of the system), and a
14255 couple of drop-in directories.
14257 * udev's predictable network interface names now use the dev_port
14258 sysfs attribute, introduced in linux 3.15 instead of dev_id to
14259 distinguish between ports of the same PCI function. dev_id should
14260 only be used for ports using the same HW address, hence the need
14263 * machined has been updated to export the OS version of a
14264 container (read from /etc/os-release and
14265 /usr/lib/os-release) on the bus. This is now shown in
14266 "machinectl status" for a machine.
14268 * A new service setting RestartForceExitStatus= has been
14269 added. If configured to a set of exit signals or process
14270 return values, the service will be restarted when the main
14271 daemon process exits with any of them, regardless of the
14274 * systemctl's -H switch for connecting to remote systemd
14275 machines has been extended so that it may be used to
14276 directly connect to a specific container on the
14277 host. "systemctl -H root@foobar:waldi" will now connect as
14278 user "root" to host "foobar", and then proceed directly to
14279 the container named "waldi". Note that currently you have to
14280 authenticate as user "root" for this to work, as entering
14281 containers is a privileged operation.
14283 Contributions from: Andreas Henriksson, Benjamin Steinwender,
14284 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
14285 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
14286 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
14287 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
14288 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
14289 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
14290 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
14291 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
14292 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
14293 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
14294 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
14296 — Berlin, 2014-07-03
14300 * As an experimental feature, udev now tries to lock the
14301 disk device node (flock(LOCK_SH|LOCK_NB)) while it
14302 executes events for the disk or any of its partitions.
14303 Applications like partitioning programs can lock the
14304 disk device node (flock(LOCK_EX)) and claim temporary
14305 device ownership that way; udev will entirely skip all event
14306 handling for this disk and its partitions. If the disk
14307 was opened for writing, the close will trigger a partition
14308 table rescan in udev's "watch" facility, and if needed
14309 synthesize "change" events for the disk and all its partitions.
14310 This is now unconditionally enabled, and if it turns out to
14311 cause major problems, we might turn it on only for specific
14312 devices, or might need to disable it entirely. Device Mapper
14313 devices are excluded from this logic.
14315 * We temporarily dropped the "-l" switch for fsck invocations,
14316 since they collide with the flock() logic above. util-linux
14317 upstream has been changed already to avoid this conflict,
14318 and we will re-add "-l" as soon as util-linux with this
14319 change has been released.
14321 * The dependency on libattr has been removed. Since a long
14322 time, the extended attribute calls have moved to glibc, and
14323 libattr is thus unnecessary.
14325 * Virtualization detection works without privileges now. This
14326 means the systemd-detect-virt binary no longer requires
14327 CAP_SYS_PTRACE file capabilities, and our daemons can run
14328 with fewer privileges.
14330 * systemd-networkd now runs under its own "systemd-network"
14331 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
14332 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
14333 loses the ability to write to files owned by root this way.
14335 * Similarly, systemd-resolved now runs under its own
14336 "systemd-resolve" user with no capabilities remaining.
14338 * Similarly, systemd-bus-proxyd now runs under its own
14339 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
14341 * systemd-networkd gained support for setting up "veth"
14342 virtual Ethernet devices for container connectivity, as well
14343 as GRE and VTI tunnels.
14345 * systemd-networkd will no longer automatically attempt to
14346 manually load kernel modules necessary for certain tunnel
14347 transports. Instead, it is assumed the kernel loads them
14348 automatically when required. This only works correctly on
14349 very new kernels. On older kernels, please consider adding
14350 the kernel modules to /etc/modules-load.d/ as a work-around.
14352 * The resolv.conf file systemd-resolved generates has been
14353 moved to /run/systemd/resolve/. If you have a symlink from
14354 /etc/resolv.conf, it might be necessary to correct it.
14356 * Two new service settings, ProtectHome= and ProtectSystem=,
14357 have been added. When enabled, they will make the user data
14358 (such as /home) inaccessible or read-only and the system
14359 (such as /usr) read-only, for specific services. This allows
14360 very light-weight per-service sandboxing to avoid
14361 modifications of user data or system files from
14362 services. These two new switches have been enabled for all
14363 of systemd's long-running services, where appropriate.
14365 * Socket units gained new SocketUser= and SocketGroup=
14366 settings to set the owner user and group of AF_UNIX sockets
14367 and FIFOs in the file system.
14369 * Socket units gained a new RemoveOnStop= setting. If enabled,
14370 all FIFOS and sockets in the file system will be removed
14371 when the specific socket unit is stopped.
14373 * Socket units gained a new Symlinks= setting. It takes a list
14374 of symlinks to create to file system sockets or FIFOs
14375 created by the specific Unix sockets. This is useful to
14376 manage symlinks to socket nodes with the same lifecycle as
14379 * The /dev/log socket and /dev/initctl FIFO have been moved to
14380 /run, and have been replaced by symlinks. This allows
14381 connecting to these facilities even if PrivateDevices=yes is
14382 used for a service (which makes /dev/log itself unavailable,
14383 but /run is left). This also has the benefit of ensuring
14384 that /dev only contains device nodes, directories and
14385 symlinks, and nothing else.
14387 * sd-daemon gained two new calls sd_pid_notify() and
14388 sd_pid_notifyf(). They are similar to sd_notify() and
14389 sd_notifyf(), but allow overriding of the source PID of
14390 notification messages if permissions permit this. This is
14391 useful to send notify messages on behalf of a different
14392 process (for example, the parent process). The
14393 systemd-notify tool has been updated to make use of this
14394 when sending messages (so that notification messages now
14395 originate from the shell script invoking systemd-notify and
14396 not the systemd-notify process itself. This should minimize
14397 a race where systemd fails to associate notification
14398 messages to services when the originating process already
14401 * A new "on-abnormal" setting for Restart= has been added. If
14402 set, it will result in automatic restarts on all "abnormal"
14403 reasons for a process to exit, which includes unclean
14404 signals, core dumps, timeouts and watchdog timeouts, but
14405 does not include clean and unclean exit codes or clean
14406 signals. Restart=on-abnormal is an alternative for
14407 Restart=on-failure for services that shall be able to
14408 terminate and avoid restarts on certain errors, by
14409 indicating so with an unclean exit code. Restart=on-failure
14410 or Restart=on-abnormal is now the recommended setting for
14411 all long-running services.
14413 * If the InaccessibleDirectories= service setting points to a
14414 mount point (or if there are any submounts contained within
14415 it), it is now attempted to completely unmount it, to make
14416 the file systems truly unavailable for the respective
14419 * The ReadOnlyDirectories= service setting and
14420 systemd-nspawn's --read-only parameter are now recursively
14421 applied to all submounts, too.
14423 * Mount units may now be created transiently via the bus APIs.
14425 * The support for SysV and LSB init scripts has been removed
14426 from the systemd daemon itself. Instead, it is now
14427 implemented as a generator that creates native systemd units
14428 from these scripts when needed. This enables us to remove a
14429 substantial amount of legacy code from PID 1, following the
14430 fact that many distributions only ship a very small number
14431 of LSB/SysV init scripts nowadays.
14433 * Privileged Xen (dom0) domains are not considered
14434 virtualization anymore by the virtualization detection
14435 logic. After all, they generally have unrestricted access to
14436 the hardware and usually are used to manage the unprivileged
14439 * systemd-tmpfiles gained a new "C" line type, for copying
14440 files or entire directories.
14442 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
14443 lines. So far, they have been non-globbing versions of the
14444 latter, and have thus been redundant. In future, it is
14445 recommended to only use "z". "m" has hence been removed
14446 from the documentation, even though it stays supported.
14448 * A tmpfiles snippet to recreate the most basic structure in
14449 /var has been added. This is enough to create the /var/run →
14450 /run symlink and create a couple of structural
14451 directories. This allows systems to boot up with an empty or
14452 volatile /var. Of course, while with this change, the core OS
14453 now is capable with dealing with a volatile /var, not all
14454 user services are ready for it. However, we hope that sooner
14455 or later, many service daemons will be changed upstream so
14456 that they are able to automatically create their necessary
14457 directories in /var at boot, should they be missing. This is
14458 the first step to allow state-less systems that only require
14459 the vendor image for /usr to boot.
14461 * systemd-nspawn has gained a new --tmpfs= switch to mount an
14462 empty tmpfs instance to a specific directory. This is
14463 particularly useful for making use of the automatic
14464 reconstruction of /var (see above), by passing --tmpfs=/var.
14466 * Access modes specified in tmpfiles snippets may now be
14467 prefixed with "~", which indicates that they shall be masked
14468 by whether the existing file or directory is currently
14469 writable, readable or executable at all. Also, if specified,
14470 the sgid/suid/sticky bits will be masked for all
14473 * A new passive target unit "network-pre.target" has been
14474 added which is useful for services that shall run before any
14475 network is configured, for example firewall scripts.
14477 * The "floppy" group that previously owned the /dev/fd*
14478 devices is no longer used. The "disk" group is now used
14479 instead. Distributions should probably deprecate usage of
14482 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
14483 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
14484 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
14485 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
14486 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
14487 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
14490 — Berlin, 2014-06-11
14494 * A new "systemd-timesyncd" daemon has been added for
14495 synchronizing the system clock across the network. It
14496 implements an SNTP client. In contrast to NTP
14497 implementations such as chrony or the NTP reference server,
14498 this only implements a client side, and does not bother with
14499 the full NTP complexity, focusing only on querying time from
14500 one remote server and synchronizing the local clock to
14501 it. Unless you intend to serve NTP to networked clients or
14502 want to connect to local hardware clocks, this simple NTP
14503 client should be more than appropriate for most
14504 installations. The daemon runs with minimal privileges, and
14505 has been hooked up with networkd to only operate when
14506 network connectivity is available. The daemon saves the
14507 current clock to disk every time a new NTP sync has been
14508 acquired, and uses this to possibly correct the system clock
14509 early at bootup, in order to accommodate for systems that
14510 lack an RTC such as the Raspberry Pi and embedded devices,
14511 and to make sure that time monotonically progresses on these
14512 systems, even if it is not always correct. To make use of
14513 this daemon, a new system user and group "systemd-timesync"
14514 needs to be created on installation of systemd.
14516 * The queue "seqnum" interface of libudev has been disabled, as
14517 it was generally incompatible with device namespacing as
14518 sequence numbers of devices go "missing" if the devices are
14519 part of a different namespace.
14521 * "systemctl list-timers" and "systemctl list-sockets" gained
14522 a --recursive switch for showing units of these types also
14523 for all local containers, similar in style to the already
14524 supported --recursive switch for "systemctl list-units".
14526 * A new RebootArgument= setting has been added for service
14527 units, which may be used to specify a kernel reboot argument
14528 to use when triggering reboots with StartLimitAction=.
14530 * A new FailureAction= setting has been added for service
14531 units which may be used to specify an operation to trigger
14532 when a service fails. This works similarly to
14533 StartLimitAction=, but unlike it, controls what is done
14534 immediately rather than only after several attempts to
14535 restart the service in question.
14537 * hostnamed got updated to also expose the kernel name,
14538 release, and version on the bus. This is useful for
14539 executing commands like hostnamectl with the -H switch.
14540 systemd-analyze makes use of this to properly display
14541 details when running non-locally.
14543 * The bootchart tool can now show cgroup information in the
14544 graphs it generates.
14546 * The CFS CPU quota cgroup attribute is now exposed for
14547 services. The new CPUQuota= switch has been added for this
14548 which takes a percentage value. Setting this will have the
14549 result that a service may never get more CPU time than the
14550 specified percentage, even if the machine is otherwise idle.
14552 * systemd-networkd learned IPIP and SIT tunnel support.
14554 * LSB init scripts exposing a dependency on $network will now
14555 get a dependency on network-online.target rather than simply
14556 network.target. This should bring LSB handling closer to
14557 what it was on SysV systems.
14559 * A new fsck.repair= kernel option has been added to control
14560 how fsck shall deal with unclean file systems at boot.
14562 * The (.ini) configuration file parser will now silently ignore
14563 sections whose names begin with "X-". This may be used to maintain
14564 application-specific extension sections in unit files.
14566 * machined gained a new API to query the IP addresses of
14567 registered containers. "machinectl status" has been updated
14568 to show these addresses in its output.
14570 * A new call sd_uid_get_display() has been added to the
14571 sd-login APIs for querying the "primary" session of a
14572 user. The "primary" session of the user is elected from the
14573 user's sessions and generally a graphical session is
14574 preferred over a text one.
14576 * A minimal systemd-resolved daemon has been added. It
14577 currently simply acts as a companion to systemd-networkd and
14578 manages resolv.conf based on per-interface DNS
14579 configuration, possibly supplied via DHCP. In the long run
14580 we hope to extend this into a local DNSSEC enabled DNS and
14583 * The systemd-networkd-wait-online tool is now enabled by
14584 default. It will delay network-online.target until a network
14585 connection has been configured. The tool primarily integrates
14586 with networkd, but will also make a best effort to make sense
14587 of network configuration performed in some other way.
14589 * Two new service options StartupCPUShares= and
14590 StartupBlockIOWeight= have been added that work similarly to
14591 CPUShares= and BlockIOWeight= however only apply during
14592 system startup. This is useful to prioritize certain services
14593 differently during bootup than during normal runtime.
14595 * hostnamed has been changed to prefer the statically
14596 configured hostname in /etc/hostname (unless set to
14597 'localhost' or empty) over any dynamic one supplied by
14598 dhcp. With this change, the rules for picking the hostname
14599 match more closely the rules of other configuration settings
14600 where the local administrator's configuration in /etc always
14601 overrides any other settings.
14603 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
14604 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
14605 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
14606 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
14607 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
14608 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
14609 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
14610 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
14611 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
14612 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
14613 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
14614 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
14615 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
14616 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
14617 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
14618 Lindskog, WaLyong Cho, Will Woods, Zbigniew
14621 — Beijing, 2014-05-28
14625 * When restoring the screen brightness at boot, stay away from
14626 the darkest setting or from the lowest 5% of the available
14627 range, depending on which is the larger value of both. This
14628 should effectively protect the user from rebooting into a
14629 black screen, should the brightness have been set to minimum
14632 * sd-login gained a new sd_machine_get_class() call to
14633 determine the class ("vm" or "container") of a machine
14634 registered with machined.
14636 * sd-login gained new calls
14637 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
14638 to query the identity of the peer of a local AF_UNIX
14639 connection. They operate similarly to their sd_pid_get_xyz()
14642 * PID 1 will now maintain a system-wide system state engine
14643 with the states "starting", "running", "degraded",
14644 "maintenance", "stopping". These states are bound to system
14645 startup, normal runtime, runtime with at least one failed
14646 service, rescue/emergency mode and system shutdown. This
14647 state is shown in the "systemctl status" output when no unit
14648 name is passed. It is useful to determine system state, in
14649 particularly when doing so for many systems or containers at
14652 * A new command "list-machines" has been added to "systemctl"
14653 that lists all local OS containers and shows their system
14654 state (see above), if systemd runs inside of them.
14656 * systemctl gained a new "-r" switch to recursively enumerate
14657 units on all local containers, when used with the
14658 "list-unit" command (which is the default one that is
14659 executed when no parameters are specified).
14661 * The GPT automatic partition discovery logic will now honour
14662 two GPT partition flags: one may be set on a partition to
14663 cause it to be mounted read-only, and the other may be set
14664 on a partition to ignore it during automatic discovery.
14666 * Two new GPT type UUIDs have been added for automatic root
14667 partition discovery, for 32-bit and 64-bit ARM. This is not
14668 particularly useful for discovering the root directory on
14669 these architectures during bare-metal boots (since UEFI is
14670 not common there), but still very useful to allow booting of
14671 ARM disk images in nspawn with the -i option.
14673 * MAC addresses of interfaces created with nspawn's
14674 --network-interface= switch will now be generated from the
14675 machine name, and thus be stable between multiple invocations
14678 * logind will now automatically remove all IPC objects owned
14679 by a user if she or he fully logs out. This makes sure that
14680 users who are logged out cannot continue to consume IPC
14681 resources. This covers SysV memory, semaphores and message
14682 queues as well as POSIX shared memory and message
14683 queues. Traditionally, SysV and POSIX IPC had no lifecycle
14684 limits. With this functionality, that is corrected. This may
14685 be turned off by using the RemoveIPC= switch of logind.conf.
14687 * The systemd-machine-id-setup and tmpfiles tools gained a
14688 --root= switch to operate on a specific root directory,
14691 * journald can now forward logged messages to the TTYs of all
14692 logged in users ("wall"). This is the default for all
14693 emergency messages now.
14695 * A new tool systemd-journal-remote has been added to stream
14696 journal log messages across the network.
14698 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
14699 controller trees are mounted into it. Note that the
14700 directories mounted beneath it are not read-only. This is a
14701 security measure and is particularly useful because glibc
14702 actually includes a search logic to pick any tmpfs it can
14703 find to implement shm_open() if /dev/shm is not available
14704 (which it might very well be in namespaced setups).
14706 * machinectl gained a new "poweroff" command to cleanly power
14707 down a local OS container.
14709 * The PrivateDevices= unit file setting will now also drop the
14710 CAP_MKNOD capability from the capability bound set, and
14711 imply DevicePolicy=closed.
14713 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
14714 comprehensively on all long-running systemd services where
14715 this is appropriate.
14717 * systemd-udevd will now run in a disassociated mount
14718 namespace. To mount directories from udev rules, make sure to
14719 pull in mount units via SYSTEMD_WANTS properties.
14721 * The kdbus support gained support for uploading policy into
14722 the kernel. sd-bus gained support for creating "monitoring"
14723 connections that can eavesdrop into all bus communication
14724 for debugging purposes.
14726 * Timestamps may now be specified in seconds since the UNIX
14727 epoch Jan 1st, 1970 by specifying "@" followed by the value
14730 * Native tcpwrap support in systemd has been removed. tcpwrap
14731 is old code, not really maintained anymore and has serious
14732 shortcomings, and better options such as firewalls
14733 exist. For setups that require tcpwrap usage, please
14734 consider invoking your socket-activated service via tcpd,
14735 like on traditional inetd.
14737 * A new system.conf configuration option
14738 DefaultTimerAccuracySec= has been added that controls the
14739 default AccuracySec= setting of .timer units.
14741 * Timer units gained a new WakeSystem= switch. If enabled,
14742 timers configured this way will cause the system to resume
14743 from system suspend (if the system supports that, which most
14746 * Timer units gained a new Persistent= switch. If enabled,
14747 timers configured this way will save to disk when they have
14748 been last triggered. This information is then used on next
14749 reboot to possible execute overdue timer events, that
14750 could not take place because the system was powered off.
14751 This enables simple anacron-like behaviour for timer units.
14753 * systemctl's "list-timers" will now also list the time a
14754 timer unit was last triggered in addition to the next time
14755 it will be triggered.
14757 * systemd-networkd will now assign predictable IPv4LL
14758 addresses to its local interfaces.
14760 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
14761 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
14762 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
14763 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
14764 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
14765 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
14766 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
14767 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
14770 — Berlin, 2014-03-25
14774 * A new unit file setting RestrictAddressFamilies= has been
14775 added to restrict which socket address families unit
14776 processes gain access to. This takes address family names
14777 like "AF_INET" or "AF_UNIX", and is useful to minimize the
14778 attack surface of services via exotic protocol stacks. This
14779 is built on seccomp system call filters.
14781 * Two new unit file settings RuntimeDirectory= and
14782 RuntimeDirectoryMode= have been added that may be used to
14783 manage a per-daemon runtime directories below /run. This is
14784 an alternative for setting up directory permissions with
14785 tmpfiles snippets, and has the advantage that the runtime
14786 directory's lifetime is bound to the daemon runtime and that
14787 the daemon starts up with an empty directory each time. This
14788 is particularly useful when writing services that drop
14789 privileges using the User= or Group= setting.
14791 * The DeviceAllow= unit setting now supports globbing for
14792 matching against device group names.
14794 * The systemd configuration file system.conf gained new
14795 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
14796 DefaultMemoryAccounting= to globally turn on/off accounting
14797 for specific resources (cgroups) for all units. These
14798 settings may still be overridden individually in each unit
14801 * systemd-gpt-auto-generator is now able to discover /srv and
14802 root partitions in addition to /home and swap partitions. It
14803 also supports LUKS-encrypted partitions now. With this in
14804 place, automatic discovery of partitions to mount following
14805 the Discoverable Partitions Specification
14806 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
14807 is now a lot more complete. This allows booting without
14808 /etc/fstab and without root= on the kernel command line on
14809 systems prepared appropriately.
14811 * systemd-nspawn gained a new --image= switch which allows
14812 booting up disk images and Linux installations on any block
14813 device that follow the Discoverable Partitions Specification
14814 (see above). This means that installations made with
14815 appropriately updated installers may now be started and
14816 deployed using container managers, completely
14817 unmodified. (We hope that libvirt-lxc will add support for
14818 this feature soon, too.)
14820 * systemd-nspawn gained a new --network-macvlan= setting to
14821 set up a private macvlan interface for the
14822 container. Similarly, systemd-networkd gained a new
14823 Kind=macvlan setting in .netdev files.
14825 * systemd-networkd now supports configuring local addresses
14828 * A new tool systemd-network-wait-online has been added to
14829 synchronously wait for network connectivity using
14832 * The sd-bus.h bus API gained a new sd_bus_track object for
14833 tracking the lifecycle of bus peers. Note that sd-bus.h is
14834 still not a public API though (unless you specify
14835 --enable-kdbus on the configure command line, which however
14836 voids your warranty and you get no API stability guarantee).
14838 * The $XDG_RUNTIME_DIR runtime directories for each user are
14839 now individual tmpfs instances, which has the benefit of
14840 introducing separate pools for each user, with individual
14841 size limits, and thus making sure that unprivileged clients
14842 can no longer negatively impact the system or other users by
14843 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
14844 RuntimeDirectorySize= has been introduced that allows
14845 controlling the default size limit for all users. It
14846 defaults to 10% of the available physical memory. This is no
14847 replacement for quotas on tmpfs though (which the kernel
14848 still does not support), as /dev/shm and /tmp are still
14849 shared resources used by both the system and unprivileged
14852 * logind will now automatically turn off automatic suspending
14853 on laptop lid close when more than one display is
14854 connected. This was previously expected to be implemented
14855 individually in desktop environments (such as GNOME),
14856 however has been added to logind now, in order to fix a
14857 boot-time race where a desktop environment might not have
14858 been started yet and thus not been able to take an inhibitor
14859 lock at the time where logind already suspends the system
14860 due to a closed lid.
14862 * logind will now wait at least 30s after each system
14863 suspend/resume cycle, and 3min after system boot before
14864 suspending the system due to a closed laptop lid. This
14865 should give USB docking stations and similar enough time to
14866 be probed and configured after system resume and boot in
14867 order to then act as suspend blocker.
14869 * systemd-run gained a new --property= setting which allows
14870 initialization of resource control properties (and others)
14871 for the created scope or service unit. Example: "systemd-run
14872 --property=BlockIOWeight=10 updatedb" may be used to run
14873 updatedb at a low block IO scheduling weight.
14875 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
14876 now also work in --scope mode.
14878 * When systemd is compiled with kdbus support, basic support
14879 for enforced policies is now in place. (Note that enabling
14880 kdbus still voids your warranty and no API compatibility
14881 promises are made.)
14883 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
14884 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14885 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
14886 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
14887 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
14888 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
14889 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
14890 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
14891 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
14892 Zbigniew Jędrzejewski-Szmek
14894 — Berlin, 2014-03-12
14898 * systemd will now relabel /dev after loading the SMACK policy
14899 according to SMACK rules.
14901 * A new unit file option AppArmorProfile= has been added to
14902 set the AppArmor profile for the processes of a unit.
14904 * A new condition check ConditionArchitecture= has been added
14905 to conditionalize units based on the system architecture, as
14906 reported by uname()'s "machine" field.
14908 * systemd-networkd now supports matching on the system
14909 virtualization, architecture, kernel command line, hostname
14912 * logind is now a lot more aggressive when suspending the
14913 machine due to a closed laptop lid. Instead of acting only
14914 on the lid close action, it will continuously watch the lid
14915 status and act on it. This is useful for laptops where the
14916 power button is on the outside of the chassis so that it can
14917 be reached without opening the lid (such as the Lenovo
14918 Yoga). On those machines, logind will now immediately
14919 re-suspend the machine if the power button has been
14920 accidentally pressed while the laptop was suspended and in a
14921 backpack or similar.
14923 * logind will now watch SW_DOCK switches and inhibit reaction
14924 to the lid switch if it is pressed. This means that logind
14925 will not suspend the machine anymore if the lid is closed
14926 and the system is docked, if the laptop supports SW_DOCK
14927 notifications via the input layer. Note that ACPI docking
14928 stations do not generate this currently. Also note that this
14929 logic is usually not fully sufficient and Desktop
14930 Environments should take a lid switch inhibitor lock when an
14931 external display is connected, as systemd will not watch
14934 * nspawn will now make use of the devices cgroup controller by
14935 default, and only permit creation of and access to the usual
14936 API device nodes like /dev/null or /dev/random, as well as
14937 access to (but not creation of) the pty devices.
14939 * We will now ship a default .network file for
14940 systemd-networkd that automatically configures DHCP for
14941 network interfaces created by nspawn's --network-veth or
14942 --network-bridge= switches.
14944 * systemd will now understand the usual M, K, G, T suffixes
14945 according to SI conventions (i.e. to the base 1000) when
14946 referring to throughput and hardware metrics. It will stay
14947 with IEC conventions (i.e. to the base 1024) for software
14948 metrics, according to what is customary according to
14949 Wikipedia. We explicitly document which base applies for
14950 each configuration option.
14952 * The DeviceAllow= setting in unit files now supports a syntax to
14953 allow-list an entire group of devices node majors at once, based on
14954 the /proc/devices listing. For example, with the string "char-pts",
14955 it is now possible to allow-list all current and future pseudo-TTYs
14958 * sd-event learned a new "post" event source. Event sources of
14959 this type are triggered by the dispatching of any event
14960 source of a type that is not "post". This is useful for
14961 implementing clean-up and check event sources that are
14962 triggered by other work being done in the program.
14964 * systemd-networkd is no longer statically enabled, but uses
14965 the usual [Install] sections so that it can be
14966 enabled/disabled using systemctl. It still is enabled by
14969 * When creating a veth interface pair with systemd-nspawn, the
14970 host side will now be prefixed with "vb-" if
14971 --network-bridge= is used, and with "ve-" if --network-veth
14972 is used. This way, it is easy to distinguish these cases on
14973 the host, for example to apply different configuration to
14974 them with systemd-networkd.
14976 * The compatibility libraries for libsystemd-journal.so,
14977 libsystem-id128.so, libsystemd-login.so and
14978 libsystemd-daemon.so do not make use of IFUNC
14979 anymore. Instead, we now build libsystemd.so multiple times
14980 under these alternative names. This means that the footprint
14981 is drastically increased, but given that these are
14982 transitional compatibility libraries, this should not matter
14983 much. This change has been made necessary to support the ARM
14984 platform for these compatibility libraries, as the ARM
14985 toolchain is not really at the same level as the toolchain
14986 for other architectures like x86 and does not support
14987 IFUNC. Please make sure to use --enable-compat-libs only
14988 during a transitional period!
14990 * The .include syntax has been deprecated and is not documented
14991 anymore. Drop-in files in .d directories should be used instead.
14993 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
14994 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14995 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
14996 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
14997 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
14998 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
14999 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
15000 Zbigniew Jędrzejewski-Szmek
15002 — Berlin, 2014-02-24
15006 * A new component "systemd-networkd" has been added that can
15007 be used to configure local network interfaces statically or
15008 via DHCP. It is capable of bringing up bridges, VLANs, and
15009 bonding. Currently, no hook-ups for interactive network
15010 configuration are provided. Use this for your initrd,
15011 container, embedded, or server setup if you need a simple,
15012 yet powerful, network configuration solution. This
15013 configuration subsystem is quite nifty, as it allows wildcard
15014 hotplug matching in interfaces. For example, with a single
15015 configuration snippet, you can configure that all Ethernet
15016 interfaces showing up are automatically added to a bridge,
15017 or similar. It supports link-sensing and more.
15019 * A new tool "systemd-socket-proxyd" has been added which can
15020 act as a bidirectional proxy for TCP sockets. This is
15021 useful for adding socket activation support to services that
15022 do not actually support socket activation, including virtual
15023 machines and the like.
15025 * Add a new tool to save/restore rfkill state on
15028 * Save/restore state of keyboard backlights in addition to
15029 display backlights on shutdown/boot.
15031 * udev learned a new SECLABEL{} construct to label device
15032 nodes with a specific security label when they appear. For
15033 now, only SECLABEL{selinux} is supported, but the syntax is
15034 prepared for additional security frameworks.
15036 * udev gained a new scheme to configure link-level attributes
15037 from files in /etc/systemd/network/*.link. These files can
15038 match against MAC address, device path, driver name and type,
15039 and will apply attributes like the naming policy, link speed,
15040 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
15041 address assignment policy (randomized, …).
15043 * The configuration of network interface naming rules for
15044 "permanent interface names" has changed: a new NamePolicy=
15045 setting in the [Link] section of .link files determines the
15046 priority of possible naming schemes (onboard, slot, MAC,
15047 path). The default value of this setting is determined by
15048 /usr/lib/net/links/99-default.link. Old
15049 80-net-name-slot.rules udev configuration file has been
15050 removed, so local configuration overriding this file should
15051 be adapted to override 99-default.link instead.
15053 * When the User= switch is used in a unit file, also
15054 initialize $SHELL= based on the user database entry.
15056 * systemd no longer depends on libdbus. All communication is
15057 now done with sd-bus, systemd's low-level bus library
15060 * kdbus support has been added to PID 1 itself. When kdbus is
15061 enabled, this causes PID 1 to set up the system bus and
15062 enable support for a new ".busname" unit type that
15063 encapsulates bus name activation on kdbus. It works a little
15064 bit like ".socket" units, except for bus names. A new
15065 generator has been added that converts classic dbus1 service
15066 activation files automatically into native systemd .busname
15067 and .service units.
15069 * sd-bus: add a light-weight vtable implementation that allows
15070 defining objects on the bus with a simple static const
15071 vtable array of its methods, signals and properties.
15073 * systemd will not generate or install static dbus
15074 introspection data anymore to /usr/share/dbus-1/interfaces,
15075 as the precise format of these files is unclear, and
15076 nothing makes use of it.
15078 * A proxy daemon is now provided to proxy clients connecting
15079 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
15080 compatibility with classic D-Bus.
15082 * A bus driver implementation has been added that supports the
15083 classic D-Bus bus driver calls on kdbus, also for
15084 compatibility purposes.
15086 * A new API "sd-event.h" has been added that implements a
15087 minimal event loop API built around epoll. It provides a
15088 couple of features that direct epoll usage is lacking:
15089 prioritization of events, scales to large numbers of timer
15090 events, per-event timer slack (accuracy), system-wide
15091 coalescing of timer events, exit handlers, watchdog
15092 supervision support using systemd's sd_notify() API, child
15095 * A new API "sd-rntl.h" has been added that provides an API
15096 around the route netlink interface of the kernel, similar in
15097 style to "sd-bus.h".
15099 * A new API "sd-dhcp-client.h" has been added that provides a
15100 small DHCPv4 client-side implementation. This is used by
15101 "systemd-networkd".
15103 * There is a new kernel command line option
15104 "systemd.restore_state=0|1". When set to "0", none of the
15105 systemd tools will restore saved runtime state to hardware
15106 devices. More specifically, the rfkill and backlight states
15109 * The FsckPassNo= compatibility option in mount/service units
15110 has been removed. The fstab generator will now add the
15111 necessary dependencies automatically, and does not require
15112 PID1's support for that anymore.
15114 * journalctl gained a new switch, --list-boots, that lists
15115 recent boots with their times and boot IDs.
15117 * The various tools like systemctl, loginctl, timedatectl,
15118 busctl, systemd-run, … have gained a new switch "-M" to
15119 connect to a specific, local OS container (as direct
15120 connection, without requiring SSH). This works on any
15121 container that is registered with machined, such as those
15122 created by libvirt-lxc or nspawn.
15124 * systemd-run and systemd-analyze also gained support for "-H"
15125 to connect to remote hosts via SSH. This is particularly
15126 useful for systemd-run because it enables queuing of jobs
15127 onto remote systems.
15129 * machinectl gained a new command "login" to open a getty
15130 login in any local container. This works with any container
15131 that is registered with machined (such as those created by
15132 libvirt-lxc or nspawn), and which runs systemd inside.
15134 * machinectl gained a new "reboot" command that may be used to
15135 trigger a reboot on a specific container that is registered
15136 with machined. This works on any container that runs an init
15137 system of some kind.
15139 * systemctl gained a new "list-timers" command to print a nice
15140 listing of installed timer units with the times they elapse
15143 * Alternative reboot() parameters may now be specified on the
15144 "systemctl reboot" command line and are passed to the
15145 reboot() system call.
15147 * systemctl gained a new --job-mode= switch to configure the
15148 mode to queue a job with. This is a more generic version of
15149 --fail, --irreversible, and --ignore-dependencies, which are
15150 still available but not advertised anymore.
15152 * /etc/systemd/system.conf gained new settings to configure
15153 various default timeouts of units, as well as the default
15154 start limit interval and burst. These may still be overridden
15157 * PID1 will now export on the bus profile data of the security
15158 policy upload process (such as the SELinux policy upload to
15161 * journald: when forwarding logs to the console, include
15162 timestamps (following the setting in
15163 /sys/module/printk/parameters/time).
15165 * OnCalendar= in timer units now understands the special
15166 strings "yearly" and "annually". (Both are equivalent)
15168 * The accuracy of timer units is now configurable with the new
15169 AccuracySec= setting. It defaults to 1min.
15171 * A new dependency type JoinsNamespaceOf= has been added that
15172 allows running two services within the same /tmp and network
15173 namespace, if PrivateNetwork= or PrivateTmp= are used.
15175 * A new command "cat" has been added to systemctl. It outputs
15176 the original unit file of a unit, and concatenates the
15177 contents of additional "drop-in" unit file snippets, so that
15178 the full configuration is shown.
15180 * systemctl now supports globbing on the various "list-xyz"
15181 commands, like "list-units" or "list-sockets", as well as on
15182 those commands which take multiple unit names.
15184 * journalctl's --unit= switch gained support for globbing.
15186 * All systemd daemons now make use of the watchdog logic so
15187 that systemd automatically notices when they hang.
15189 * If the $container_ttys environment variable is set,
15190 getty-generator will automatically spawn a getty for each
15191 listed tty. This is useful for container managers to request
15192 login gettys to be spawned on as many ttys as needed.
15194 * %h, %s, %U specifier support is not available anymore when
15195 used in unit files for PID 1. This is because NSS calls are
15196 not safe from PID 1. They stay available for --user
15197 instances of systemd, and as special case for the root user.
15199 * loginctl gained a new "--no-legend" switch to turn off output
15200 of the legend text.
15202 * The "sd-login.h" API gained three new calls:
15203 sd_session_is_remote(), sd_session_get_remote_user(),
15204 sd_session_get_remote_host() to query information about
15207 * The udev hardware database now also carries vendor/product
15208 information of SDIO devices.
15210 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
15211 determine whether watchdog notifications are requested by
15212 the system manager.
15214 * Socket-activated per-connection services now include a
15215 short description of the connection parameters in the
15218 * tmpfiles gained a new "--boot" option. When this is not used,
15219 only lines where the command character is not suffixed with
15220 "!" are executed. When this option is specified, those
15221 options are executed too. This partitions tmpfiles
15222 directives into those that can be safely executed at any
15223 time, and those which should be run only at boot (for
15224 example, a line that creates /run/nologin).
15226 * A new API "sd-resolve.h" has been added which provides a simple
15227 asynchronous wrapper around glibc NSS hostname resolution
15228 calls, such as getaddrinfo(). In contrast to glibc's
15229 getaddrinfo_a(), it does not use signals. In contrast to most
15230 other asynchronous name resolution libraries, this one does
15231 not reimplement DNS, but reuses NSS, so that alternate
15232 hostname resolution systems continue to work, such as mDNS,
15233 LDAP, etc. This API is based on libasyncns, but it has been
15234 cleaned up for inclusion in systemd.
15236 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
15237 "sd-daemon.h" are no longer found in individual libraries
15238 libsystemd-journal.so, libsystemd-login.so,
15239 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
15240 merged them into a single library, libsystemd.so, which
15241 provides all symbols. The reason for this is cyclic
15242 dependencies, as these libraries tend to use each other's
15243 symbols. So far, we have managed to workaround that by linking
15244 a copy of a good part of our code into each of these
15245 libraries again and again, which, however, makes certain
15246 things hard to do, like sharing static variables. Also, it
15247 substantially increases footprint. With this change, there
15248 is only one library for the basic APIs systemd
15249 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
15250 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
15251 library as well, however are subject to the --enable-kdbus
15252 switch (see below). Note that "sd-dhcp-client.h" is not part
15253 of this library (this is because it only consumes, never
15254 provides, services of/to other APIs). To make the transition
15255 easy from the separate libraries to the unified one, we
15256 provide the --enable-compat-libs compile-time switch which
15257 will generate stub libraries that are compatible with the
15258 old ones but redirect all calls to the new one.
15260 * All of the kdbus logic and the new APIs "sd-bus.h",
15261 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
15262 and "sd-utf8.h" are compile-time optional via the
15263 "--enable-kdbus" switch, and they are not compiled in by
15264 default. To make use of kdbus, you have to explicitly enable
15265 the switch. Note however, that neither the kernel nor the
15266 userspace API for all of this is considered stable yet. We
15267 want to maintain the freedom to still change the APIs for
15268 now. By specifying this build-time switch, you acknowledge
15269 that you are aware of the instability of the current
15272 * Also, note that while kdbus is pretty much complete,
15273 it lacks one thing: proper policy support. This means you
15274 can build a fully working system with all features; however,
15275 it will be highly insecure. Policy support will be added in
15276 one of the next releases, at the same time that we will
15277 declare the APIs stable.
15279 * When the kernel command line argument "kdbus" is specified,
15280 systemd will automatically load the kdbus.ko kernel module. At
15281 this stage of development, it is only useful for testing kdbus
15282 and should not be used in production. Note: if "--enable-kdbus"
15283 is specified, and the kdbus.ko kernel module is available, and
15284 "kdbus" is added to the kernel command line, the entire system
15285 runs with kdbus instead of dbus-daemon, with the above mentioned
15286 problem of missing the system policy enforcement. Also a future
15287 version of kdbus.ko or a newer systemd will not be compatible with
15288 each other, and will unlikely be able to boot the machine if only
15289 one of them is updated.
15291 * systemctl gained a new "import-environment" command which
15292 uploads the caller's environment (or parts thereof) into the
15293 service manager so that it is inherited by services started
15294 by the manager. This is useful to upload variables like
15295 $DISPLAY into the user service manager.
15297 * A new PrivateDevices= switch has been added to service units
15298 which allows running a service with a namespaced /dev
15299 directory that does not contain any device nodes for
15300 physical devices. More specifically, it only includes devices
15301 such as /dev/null, /dev/urandom, and /dev/zero which are API
15304 * logind has been extended to support behaviour like VT
15305 switching on seats that do not support a VT. This makes
15306 multi-session available on seats that are not the first seat
15307 (seat0), and on systems where kernel support for VTs has
15308 been disabled at compile-time.
15310 * If a process holds a delay lock for system sleep or shutdown
15311 and fails to release it in time, we will now log its
15312 identity. This makes it easier to identify processes that
15313 cause slow suspends or power-offs.
15315 * When parsing /etc/crypttab, support for a new key-slot=
15316 option as supported by Debian is added. It allows indicating
15317 which LUKS slot to use on disk, speeding up key loading.
15319 * The sd_journal_sendv() API call has been checked and
15320 officially declared to be async-signal-safe so that it may
15321 be invoked from signal handlers for logging purposes.
15323 * Boot-time status output is now enabled automatically after a
15324 short timeout if boot does not progress, in order to give
15325 the user an indication what she or he is waiting for.
15327 * The boot-time output has been improved to show how much time
15328 remains until jobs expire.
15330 * The KillMode= switch in service units gained a new possible
15331 value "mixed". If set, and the unit is shut down, then the
15332 initial SIGTERM signal is sent only to the main daemon
15333 process, while the following SIGKILL signal is sent to
15334 all remaining processes of the service.
15336 * When a scope unit is registered, a new property "Controller"
15337 may be set. If set to a valid bus name, systemd will send a
15338 RequestStop() signal to this name when it would like to shut
15339 down the scope. This may be used to hook manager logic into
15340 the shutdown logic of scope units. Also, scope units may now
15341 be put in a special "abandoned" state, in which case the
15342 manager process which created them takes no further
15343 responsibilities for it.
15345 * When reading unit files, systemd will now verify
15346 the access mode of these files, and warn about certain
15347 suspicious combinations. This has been added to make it
15348 easier to track down packaging bugs where unit files are
15349 marked executable or world-writable.
15351 * systemd-nspawn gained a new "--setenv=" switch to set
15352 container-wide environment variables. The similar option in
15353 systemd-activate was renamed from "--environment=" to
15354 "--setenv=" for consistency.
15356 * systemd-nspawn has been updated to create a new kdbus domain
15357 for each container that is invoked, thus allowing each
15358 container to have its own set of system and user buses,
15359 independent of the host.
15361 * systemd-nspawn gained a new --drop-capability= switch to run
15362 the container with less capabilities than the default. Both
15363 --drop-capability= and --capability= now take the special
15364 string "all" for dropping or keeping all capabilities.
15366 * systemd-nspawn gained new switches for executing containers
15367 with specific SELinux labels set.
15369 * systemd-nspawn gained a new --quiet switch to not generate
15370 any additional output but the container's own console
15373 * systemd-nspawn gained a new --share-system switch to run a
15374 container without PID namespacing enabled.
15376 * systemd-nspawn gained a new --register= switch to control
15377 whether the container is registered with systemd-machined or
15378 not. This is useful for containers that do not run full
15379 OS images, but only specific apps.
15381 * systemd-nspawn gained a new --keep-unit which may be used
15382 when invoked as the only program from a service unit, and
15383 results in registration of the unit service itself in
15384 systemd-machined, instead of a newly opened scope unit.
15386 * systemd-nspawn gained a new --network-interface= switch for
15387 moving arbitrary interfaces to the container. The new
15388 --network-veth switch creates a virtual Ethernet connection
15389 between host and container. The new --network-bridge=
15390 switch then allows assigning the host side of this virtual
15391 Ethernet connection to a bridge device.
15393 * systemd-nspawn gained a new --personality= switch for
15394 setting the kernel personality for the container. This is
15395 useful when running a 32-bit container on a 64-bit host. A
15396 similar option Personality= is now also available for service
15399 * logind will now also track a "Desktop" identifier for each
15400 session which encodes the desktop environment of it. This is
15401 useful for desktop environments that want to identify
15402 multiple running sessions of itself easily.
15404 * A new SELinuxContext= setting for service units has been
15405 added that allows setting a specific SELinux execution
15406 context for a service.
15408 * Most systemd client tools will now honour $SYSTEMD_LESS for
15409 settings of the "less" pager. By default, these tools will
15410 override $LESS to allow certain operations to work, such as
15411 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
15412 influence this logic.
15414 * systemd's "seccomp" hook-up has been changed to make use of
15415 the libseccomp library instead of using its own
15416 implementation. This has benefits for portability among
15419 * For usage together with SystemCallFilter=, a new
15420 SystemCallErrorNumber= setting has been introduced that
15421 allows configuration of a system error number to be returned
15422 on filtered system calls, instead of immediately killing the
15423 process. Also, SystemCallArchitectures= has been added to
15424 limit access to system calls of a particular architecture
15425 (in order to turn off support for unused secondary
15426 architectures). There is also a global
15427 SystemCallArchitectures= setting in system.conf now to turn
15428 off support for non-native system calls system-wide.
15430 * systemd requires a kernel with a working name_to_handle_at(),
15431 please see the kernel config requirements in the README file.
15433 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
15434 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
15435 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
15436 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
15437 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
15438 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
15439 Elia Pinto, Florian Weimer, George McCollister, Goffredo
15440 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
15441 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
15442 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
15443 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
15444 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
15445 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
15446 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
15447 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
15448 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
15449 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
15450 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
15451 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
15452 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
15453 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
15454 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
15455 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
15456 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
15458 — Berlin, 2014-02-20
15462 * logind has gained support for facilitating privileged input
15463 and drm device access for unprivileged clients. This work is
15464 useful to allow Wayland display servers (and similar
15465 programs, such as kmscon) to run under the user's ID and
15466 access input and drm devices which are normally
15467 protected. When this is used (and the kernel is new enough)
15468 logind will "mute" IO on the file descriptors passed to
15469 Wayland as long as it is in the background and "unmute" it
15470 if it returns into the foreground. This allows secure
15471 session switching without allowing background sessions to
15472 eavesdrop on input and display data. This also introduces
15473 session switching support if VT support is turned off in the
15474 kernel, and on seats that are not seat0.
15476 * A new kernel command line option luks.options= is understood
15477 now which allows specifying LUKS options for usage for LUKS
15478 encrypted partitions specified with luks.uuid=.
15480 * tmpfiles.d(5) snippets may now use specifier expansion in
15481 path names. More specifically %m, %b, %H, %v, are now
15482 replaced by the local machine id, boot id, hostname, and
15483 kernel version number.
15485 * A new tmpfiles.d(5) command "m" has been introduced which
15486 may be used to change the owner/group/access mode of a file
15487 or directory if it exists, but do nothing if it does not.
15489 * This release removes high-level support for the
15490 MemorySoftLimit= cgroup setting. The underlying kernel
15491 cgroup attribute memory.soft_limit= is currently badly
15492 designed and likely to be removed from the kernel API in its
15493 current form, hence we should not expose it for now.
15495 * The memory.use_hierarchy cgroup attribute is now enabled for
15496 all cgroups systemd creates in the memory cgroup
15497 hierarchy. This option is likely to be come the built-in
15498 default in the kernel anyway, and the non-hierarchical mode
15499 never made much sense in the intrinsically hierarchical
15502 * A new field _SYSTEMD_SLICE= is logged along with all journal
15503 messages containing the slice a message was generated
15504 from. This is useful to allow easy per-customer filtering of
15505 logs among other things.
15507 * systemd-journald will no longer adjust the group of journal
15508 files it creates to the "systemd-journal" group. Instead we
15509 rely on the journal directory to be owned by the
15510 "systemd-journal" group, and its setgid bit set, so that the
15511 kernel file system layer will automatically enforce that
15512 journal files inherit this group assignment. The reason for
15513 this change is that we cannot allow NSS look-ups from
15514 journald which would be necessary to resolve
15515 "systemd-journal" to a numeric GID, because this might
15516 create deadlocks if NSS involves synchronous queries to
15517 other daemons (such as nscd, or sssd) which in turn are
15518 logging clients of journald and might block on it, which
15519 would then dead lock. A tmpfiles.d(5) snippet included in
15520 systemd will make sure the setgid bit and group are
15521 properly set on the journal directory if it exists on every
15522 boot. However, we recommend adjusting it manually after
15523 upgrades too (or from RPM scriptlets), so that the change is
15524 not delayed until next reboot.
15526 * Backlight and random seed files in /var/lib/ have moved into
15527 the /var/lib/systemd/ directory, in order to centralize all
15528 systemd generated files in one directory.
15530 * Boot time performance measurements (as displayed by
15531 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
15532 performance information if that's available to determine how
15533 much time BIOS and boot loader initialization required. With
15534 a sufficiently new BIOS you hence no longer need to boot
15535 with Gummiboot to get access to such information.
15537 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
15538 Cristian Rodríguez, Dave Reisner, David Herrmann, David
15539 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
15540 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
15541 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
15542 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
15543 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15545 — Berlin, 2013-10-02
15549 * The Restart= option for services now understands a new
15550 on-watchdog setting, which will restart the service
15551 automatically if the service stops sending out watchdog keep
15552 alive messages (as configured with WatchdogSec=).
15554 * The getty generator (which is responsible for bringing up a
15555 getty on configured serial consoles) will no longer only
15556 start a getty on the primary kernel console but on all
15557 others, too. This makes the order in which console= is
15558 specified on the kernel command line less important.
15560 * libsystemd-logind gained a new sd_session_get_vt() call to
15561 retrieve the VT number of a session.
15563 * If the option "tries=0" is set for an entry of /etc/crypttab
15564 its passphrase is queried indefinitely instead of any
15565 maximum number of tries.
15567 * If a service with a configure PID file terminates its PID
15568 file will now be removed automatically if it still exists
15569 afterwards. This should put an end to stale PID files.
15571 * systemd-run will now also take relative binary path names
15572 for execution and no longer insists on absolute paths.
15574 * InaccessibleDirectories= and ReadOnlyDirectories= now take
15575 paths that are optionally prefixed with "-" to indicate that
15576 it should not be considered a failure if they do not exist.
15578 * journalctl -o (and similar commands) now understands a new
15579 output mode "short-precise", it is similar to "short" but
15580 shows timestamps with usec accuracy.
15582 * The option "discard" (as known from Debian) is now
15583 synonymous to "allow-discards" in /etc/crypttab. In fact,
15584 "discard" is preferred now (since it is easier to remember
15587 * Some licensing clean-ups were made, so that more code is now
15588 LGPL-2.1 licensed than before.
15590 * A minimal tool to save/restore the display backlight
15591 brightness across reboots has been added. It will store the
15592 backlight setting as late as possible at shutdown, and
15593 restore it as early as possible during reboot.
15595 * A logic to automatically discover and enable home and swap
15596 partitions on GPT disks has been added. With this in place
15597 /etc/fstab becomes optional for many setups as systemd can
15598 discover certain partitions located on the root disk
15599 automatically. Home partitions are recognized under their
15600 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
15601 partitions are recognized under their GPT type ID
15602 0657fd6da4ab43c484e50933c84b4f4f.
15604 * systemd will no longer pass any environment from the kernel
15605 or initrd to system services. If you want to set an
15606 environment for all services, do so via the kernel command
15607 line systemd.setenv= assignment.
15609 * The systemd-sysctl tool no longer natively reads the file
15610 /etc/sysctl.conf. If desired, the file should be symlinked
15611 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
15612 legacy support by a symlink rather than built-in code, it
15613 also makes the otherwise hidden order of application of the
15614 different files visible. (Note that this partly reverts to a
15615 pre-198 application order of sysctl knobs!)
15617 * The "systemctl set-log-level" and "systemctl dump" commands
15618 have been moved to systemd-analyze.
15620 * systemd-run learned the new --remain-after-exit switch,
15621 which causes the scope unit not to be cleaned up
15622 automatically after the process terminated.
15624 * tmpfiles learned a new --exclude-prefix= switch to exclude
15625 certain paths from operation.
15627 * journald will now automatically flush all messages to disk
15628 as soon as a message at the log level CRIT, ALERT or EMERG
15631 Contributions from: Andrew Cook, Brandon Philips, Christian
15632 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
15633 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
15634 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
15635 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
15636 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
15637 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
15638 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
15639 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
15640 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
15641 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
15642 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
15643 William Giokas, Zbigniew Jędrzejewski-Szmek
15645 — Berlin, 2013-09-13
15649 * The documentation has been updated to cover the various new
15650 concepts introduced with 205.
15652 * Unit files now understand the new %v specifier which
15653 resolves to the kernel version string as returned by "uname
15656 * systemctl now supports filtering the unit list output by
15657 load state, active state and sub state, using the new
15658 --state= parameter.
15660 * "systemctl status" will now show the results of the
15661 condition checks (like ConditionPathExists= and similar) of
15662 the last start attempts of the unit. They are also logged to
15665 * "journalctl -b" may now be used to look for boot output of a
15666 specific boot. Try "journalctl -b -1" for the previous boot,
15667 but the syntax is substantially more powerful.
15669 * "journalctl --show-cursor" has been added which prints the
15670 cursor string the last shown log line. This may then be used
15671 with the new "journalctl --after-cursor=" switch to continue
15672 browsing logs from that point on.
15674 * "journalctl --force" may now be used to force regeneration
15677 * Creation of "dead" device nodes has been moved from udev
15678 into kmod and tmpfiles. Previously, udev would read the kmod
15679 databases to pre-generate dead device nodes based on meta
15680 information contained in kernel modules, so that these would
15681 be auto-loaded on access rather then at boot. As this
15682 does not really have much to do with the exposing actual
15683 kernel devices to userspace this has always been slightly
15684 alien in the udev codebase. Following the new scheme kmod
15685 will now generate a runtime snippet for tmpfiles from the
15686 module meta information and it now is tmpfiles' job to the
15687 create the nodes. This also allows overriding access and
15688 other parameters for the nodes using the usual tmpfiles
15689 facilities. As side effect this allows us to remove the
15690 CAP_SYS_MKNOD capability bit from udevd entirely.
15692 * logind's device ACLs may now be applied to these "dead"
15693 devices nodes too, thus finally allowing managed access to
15694 devices such as /dev/snd/sequencer without loading the
15695 backing module right-away.
15697 * A new RPM macro has been added that may be used to apply
15698 tmpfiles configuration during package installation.
15700 * systemd-detect-virt and ConditionVirtualization= now can
15701 detect User-Mode-Linux machines (UML).
15703 * journald will now implicitly log the effective capabilities
15704 set of processes in the message metadata.
15706 * systemd-cryptsetup has gained support for TrueCrypt volumes.
15708 * The initrd interface has been simplified (more specifically,
15709 support for passing performance data via environment
15710 variables and fsck results via files in /run has been
15711 removed). These features were non-essential, and are
15712 nowadays available in a much nicer way by having systemd in
15713 the initrd serialize its state and have the hosts systemd
15714 deserialize it again.
15716 * The udev "keymap" data files and tools to apply keyboard
15717 specific mappings of scan to key codes, and force-release
15718 scan code lists have been entirely replaced by a udev
15719 "keyboard" builtin and a hwdb data file.
15721 * systemd will now honour the kernel's "quiet" command line
15722 argument also during late shutdown, resulting in a
15723 completely silent shutdown when used.
15725 * There's now an option to control the SO_REUSEPORT socket
15726 option in .socket units.
15728 * Instance units will now automatically get a per-template
15729 subslice of system.slice unless something else is explicitly
15730 configured. For example, instances of sshd@.service will now
15731 implicitly be placed in system-sshd.slice rather than
15732 system.slice as before.
15734 * Test coverage support may now be enabled at build time.
15736 Contributions from: Dave Reisner, Frederic Crozat, Harald
15737 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
15738 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
15739 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
15740 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
15741 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
15742 Giokas, Zbigniew Jędrzejewski-Szmek
15744 — Berlin, 2013-07-23
15748 * Two new unit types have been introduced:
15750 Scope units are very similar to service units, however, are
15751 created out of pre-existing processes — instead of PID 1
15752 forking off the processes. By using scope units it is
15753 possible for system services and applications to group their
15754 own child processes (worker processes) in a powerful way
15755 which then maybe used to organize them, or kill them
15756 together, or apply resource limits on them.
15758 Slice units may be used to partition system resources in an
15759 hierarchical fashion and then assign other units to them. By
15760 default there are now three slices: system.slice (for all
15761 system services), user.slice (for all user sessions),
15762 machine.slice (for VMs and containers).
15764 Slices and scopes have been introduced primarily in
15765 context of the work to move cgroup handling to a
15766 single-writer scheme, where only PID 1
15767 creates/removes/manages cgroups.
15769 * There's a new concept of "transient" units. In contrast to
15770 normal units these units are created via an API at runtime,
15771 not from configuration from disk. More specifically this
15772 means it is now possible to run arbitrary programs as
15773 independent services, with all execution parameters passed
15774 in via bus APIs rather than read from disk. Transient units
15775 make systemd substantially more dynamic then it ever was,
15776 and useful as a general batch manager.
15778 * logind has been updated to make use of scope and slice units
15779 for managing user sessions. As a user logs in he will get
15780 his own private slice unit, to which all sessions are added
15781 as scope units. We also added support for automatically
15782 adding an instance of user@.service for the user into the
15783 slice. Effectively logind will no longer create cgroup
15784 hierarchies on its own now, it will defer entirely to PID 1
15785 for this by means of scope, service and slice units. Since
15786 user sessions this way become entities managed by PID 1
15787 the output of "systemctl" is now a lot more comprehensive.
15789 * A new mini-daemon "systemd-machined" has been added which
15790 may be used by virtualization managers to register local
15791 VMs/containers. nspawn has been updated accordingly, and
15792 libvirt will be updated shortly. machined will collect a bit
15793 of meta information about the VMs/containers, and assign
15794 them their own scope unit (see above). The collected
15795 meta-data is then made available via the "machinectl" tool,
15796 and exposed in "ps" and similar tools. machined/machinectl
15797 is compile-time optional.
15799 * As discussed earlier, the low-level cgroup configuration
15800 options ControlGroup=, ControlGroupModify=,
15801 ControlGroupPersistent=, ControlGroupAttribute= have been
15802 removed. Please use high-level attribute settings instead as
15803 well as slice units.
15805 * A new bus call SetUnitProperties() has been added to alter
15806 various runtime parameters of a unit. This is primarily
15807 useful to alter cgroup parameters dynamically in a nice way,
15808 but will be extended later on to make more properties
15809 modifiable at runtime. systemctl gained a new set-properties
15810 command that wraps this call.
15812 * A new tool "systemd-run" has been added which can be used to
15813 run arbitrary command lines as transient services or scopes,
15814 while configuring a number of settings via the command
15815 line. This tool is currently very basic, however already
15816 very useful. We plan to extend this tool to even allow
15817 queuing of execution jobs with time triggers from the
15818 command line, similar in fashion to "at".
15820 * nspawn will now inform the user explicitly that kernels with
15821 audit enabled break containers, and suggest the user to turn
15824 * Support for detecting the IMA and AppArmor security
15825 frameworks with ConditionSecurity= has been added.
15827 * journalctl gained a new "-k" switch for showing only kernel
15828 messages, mimicking dmesg output; in addition to "--user"
15829 and "--system" switches for showing only user's own logs
15832 * systemd-delta can now show information about drop-in
15833 snippets extending unit files.
15835 * libsystemd-bus has been substantially updated but is still
15836 not available as public API.
15838 * systemd will now look for the "debug" argument on the kernel
15839 command line and enable debug logging, similar to what
15840 "systemd.log_level=debug" already did before.
15842 * "systemctl set-default", "systemctl get-default" has been
15843 added to configure the default.target symlink, which
15844 controls what to boot into by default.
15846 * "systemctl set-log-level" has been added as a convenient
15847 way to raise and lower systemd logging threshold.
15849 * "systemd-analyze plot" will now show the time the various
15850 generators needed for execution, as well as information
15851 about the unit file loading.
15853 * libsystemd-journal gained a new sd_journal_open_files() call
15854 for opening specific journal files. journactl also gained a
15855 new switch to expose this new functionality. Previously we
15856 only supported opening all files from a directory, or all
15857 files from the system, as opening individual files only is
15858 racy due to journal file rotation.
15860 * systemd gained the new DefaultEnvironment= setting in
15861 /etc/systemd/system.conf to set environment variables for
15864 * If a privileged process logs a journal message with the
15865 OBJECT_PID= field set, then journald will automatically
15866 augment this with additional OBJECT_UID=, OBJECT_GID=,
15867 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
15868 system services want to log events about specific client
15869 processes. journactl/systemctl has been updated to make use
15870 of this information if all log messages regarding a specific
15873 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
15874 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
15875 Reisner, David Coppa, David King, David Strauss, Eelco
15876 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
15877 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
15878 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
15879 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
15880 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
15881 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
15882 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
15883 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
15884 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
15885 Łukasz Stelmach, 장동준
15889 * The Python bindings gained some minimal support for the APIs
15890 exposed by libsystemd-logind.
15892 * ConditionSecurity= gained support for detecting SMACK. Since
15893 this condition already supports SELinux and AppArmor we only
15894 miss IMA for this. Patches welcome!
15896 Contributions from: Karol Lewandowski, Lennart Poettering,
15897 Zbigniew Jędrzejewski-Szmek
15901 * systemd-nspawn will now create /etc/resolv.conf if
15902 necessary, before bind-mounting the host's file onto it.
15904 * systemd-nspawn will now store meta information about a
15905 container on the container's cgroup as extended attribute
15906 fields, including the root directory.
15908 * The cgroup hierarchy has been reworked in many ways. All
15909 objects any of the components systemd creates in the cgroup
15910 tree are now suffixed. More specifically, user sessions are
15911 now placed in cgroups suffixed with ".session", users in
15912 cgroups suffixed with ".user", and nspawn containers in
15913 cgroups suffixed with ".nspawn". Furthermore, all cgroup
15914 names are now escaped in a simple scheme to avoid collision
15915 of userspace object names with kernel filenames. This work
15916 is preparation for making these objects relocatable in the
15917 cgroup tree, in order to allow easy resource partitioning of
15918 these objects without causing naming conflicts.
15920 * systemctl list-dependencies gained the new switches
15921 --plain, --reverse, --after and --before.
15923 * systemd-inhibit now shows the process name of processes that
15924 have taken an inhibitor lock.
15926 * nss-myhostname will now also resolve "localhost"
15927 implicitly. This makes /etc/hosts an optional file and
15928 nicely handles that on IPv6 ::1 maps to both "localhost" and
15929 the local hostname.
15931 * libsystemd-logind.so gained a new call
15932 sd_get_machine_names() to enumerate running containers and
15933 VMs (currently only supported by very new libvirt and
15934 nspawn). sd_login_monitor can now be used to watch
15935 VMs/containers coming and going.
15937 * .include is not allowed recursively anymore, and only in
15938 unit files. Usually it is better to use drop-in snippets in
15939 .d/*.conf anyway, as introduced with systemd 198.
15941 * systemd-analyze gained a new "critical-chain" command that
15942 determines the slowest chain of units run during system
15943 boot-up. It is very useful for tracking down where
15944 optimizing boot time is the most beneficial.
15946 * systemd will no longer allow manipulating service paths in
15947 the name=systemd:/system cgroup tree using ControlGroup= in
15948 units. (But is still fine with it in all other dirs.)
15950 * There's a new systemd-nspawn@.service service file that may
15951 be used to easily run nspawn containers as system
15952 services. With the container's root directory in
15953 /var/lib/container/foobar it is now sufficient to run
15954 "systemctl start systemd-nspawn@foobar.service" to boot it.
15956 * systemd-cgls gained a new parameter "--machine" to list only
15957 the processes within a certain container.
15959 * ConditionSecurity= now can check for "apparmor". We still
15960 are lacking checks for SMACK and IMA for this condition
15961 check though. Patches welcome!
15963 * A new configuration file /etc/systemd/sleep.conf has been
15964 added that may be used to configure which kernel operation
15965 systemd is supposed to execute when "suspend", "hibernate"
15966 or "hybrid-sleep" is requested. This makes the new kernel
15967 "freeze" state accessible to the user.
15969 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
15970 the passed argument if applicable.
15972 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
15973 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
15974 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
15975 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
15976 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
15977 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
15978 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
15983 * The output of 'systemctl list-jobs' got some polishing. The
15984 '--type=' argument may now be passed more than once. A new
15985 command 'systemctl list-sockets' has been added which shows
15986 a list of kernel sockets systemd is listening on with the
15987 socket units they belong to, plus the units these socket
15990 * The experimental libsystemd-bus library got substantial
15991 updates to work in conjunction with the (also experimental)
15992 kdbus kernel project. It works well enough to exchange
15993 messages with some sophistication. Note that kdbus is not
15994 ready yet, and the library is mostly an elaborate test case
15995 for now, and not installable.
15997 * systemd gained a new unit 'systemd-static-nodes.service'
15998 that generates static device nodes earlier during boot, and
15999 can run in conjunction with udev.
16001 * libsystemd-login gained a new call sd_pid_get_user_unit()
16002 to retrieve the user systemd unit a process is running
16003 in. This is useful for systems where systemd is used as
16006 * systemd-nspawn now places all containers in the new /machine
16007 top-level cgroup directory in the name=systemd
16008 hierarchy. libvirt will soon do the same, so that we get a
16009 uniform separation of /system, /user and /machine for system
16010 services, user processes and containers/virtual
16011 machines. This new cgroup hierarchy is also useful to stick
16012 stable names to specific container instances, which can be
16013 recognized later this way (this name may be controlled
16014 via systemd-nspawn's new -M switch). libsystemd-login also
16015 gained a new call sd_pid_get_machine_name() to retrieve the
16016 name of the container/VM a specific process belongs to.
16018 * bootchart can now store its data in the journal.
16020 * libsystemd-journal gained a new call
16021 sd_journal_add_conjunction() for AND expressions to the
16022 matching logic. This can be used to express more complex
16023 logical expressions.
16025 * journactl can now take multiple --unit= and --user-unit=
16028 * The cryptsetup logic now understands the "luks.key=" kernel
16029 command line switch for specifying a file to read the
16030 decryption key from. Also, if a configured key file is not
16031 found the tool will now automatically fall back to prompting
16034 * Python systemd.journal module was updated to wrap recently
16035 added functions from libsystemd-journal. The interface was
16036 changed to bring the low level interface in s.j._Reader
16037 closer to the C API, and the high level interface in
16038 s.j.Reader was updated to wrap and convert all data about
16041 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
16042 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
16043 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
16044 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
16045 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
16046 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
16050 * journalctl --update-catalog now understands a new --root=
16051 option to operate on catalogs found in a different root
16054 * During shutdown after systemd has terminated all running
16055 services a final killing loop kills all remaining left-over
16056 processes. We will now print the name of these processes
16057 when we send SIGKILL to them, since this usually indicates a
16060 * If /etc/crypttab refers to password files stored on
16061 configured mount points automatic dependencies will now be
16062 generated to ensure the specific mount is established first
16063 before the key file is attempted to be read.
16065 * 'systemctl status' will now show information about the
16066 network sockets a socket unit is listening on.
16068 * 'systemctl status' will also shown information about any
16069 drop-in configuration file for units. (Drop-In configuration
16070 files in this context are files such as
16071 /etc/systemd/system/foobar.service.d/*.conf)
16073 * systemd-cgtop now optionally shows summed up CPU times of
16074 cgroups. Press '%' while running cgtop to switch between
16075 percentage and absolute mode. This is useful to determine
16076 which cgroups use up the most CPU time over the entire
16077 runtime of the system. systemd-cgtop has also been updated
16078 to be 'pipeable' for processing with further shell tools.
16080 * 'hostnamectl set-hostname' will now allow setting of FQDN
16083 * The formatting and parsing of time span values has been
16084 changed. The parser now understands fractional expressions
16085 such as "5.5h". The formatter will now output fractional
16086 expressions for all time spans under 1min, i.e. "5.123456s"
16087 rather than "5s 123ms 456us". For time spans under 1s
16088 millisecond values are shown, for those under 1ms
16089 microsecond values are shown. This should greatly improve
16090 all time-related output of systemd.
16092 * libsystemd-login and libsystemd-journal gained new
16093 functions for querying the poll() events mask and poll()
16094 timeout value for integration into arbitrary event
16097 * localectl gained the ability to list available X11 keymaps
16098 (models, layouts, variants, options).
16100 * 'systemd-analyze dot' gained the ability to filter for
16101 specific units via shell-style globs, to create smaller,
16102 more useful graphs. I.e. it is now possible to create simple
16103 graphs of all the dependencies between only target units, or
16104 of all units that Avahi has dependencies with.
16106 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
16107 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
16108 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
16109 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
16110 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
16111 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
16112 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
16116 * The boot-time readahead implementation for rotating media
16117 will now read the read-ahead data in multiple passes which
16118 consist of all read requests made in equidistant time
16119 intervals. This means instead of strictly reading read-ahead
16120 data in its physical order on disk we now try to find a
16121 middle ground between physical and access time order.
16123 * /etc/os-release files gained a new BUILD_ID= field for usage
16124 on operating systems that provide continuous builds of OS
16127 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
16128 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
16129 William Douglas, Zbigniew Jędrzejewski-Szmek
16133 * systemd-python gained an API exposing libsystemd-daemon.
16135 * The SMACK setup logic gained support for uploading CIPSO
16138 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
16139 ReadOnlyDirectories= and InaccessibleDirectories= has
16140 changed. The private /tmp and /var/tmp directories are now
16141 shared by all processes of a service (which means
16142 ExecStartPre= may now leave data in /tmp that ExecStart= of
16143 the same service can still access). When a service is
16144 stopped its temporary directories are immediately deleted
16145 (normal clean-up with tmpfiles is still done in addition to
16148 * By default, systemd will now set a couple of sysctl
16149 variables in the kernel: the safe sysrq options are turned
16150 on, IP route verification is turned on, and source routing
16151 disabled. The recently added hardlink and softlink
16152 protection of the kernel is turned on. These settings should
16153 be reasonably safe, and good defaults for all new systems.
16155 * The predictable network naming logic may now be turned off
16156 with a new kernel command line switch: net.ifnames=0.
16158 * A new libsystemd-bus module has been added that implements a
16159 pretty complete D-Bus client library. For details see:
16161 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
16163 * journald will now explicitly flush the journal files to disk
16164 at the latest 5min after each write. The file will then also
16165 be marked offline until the next write. This should increase
16166 reliability in case of a crash. The synchronization delay
16167 can be configured via SyncIntervalSec= in journald.conf.
16169 * There's a new remote-fs-setup.target unit that can be used
16170 to pull in specific services when at least one remote file
16171 system is to be mounted.
16173 * There are new targets timers.target and paths.target as
16174 canonical targets to pull user timer and path units in
16175 from. This complements sockets.target with a similar
16176 purpose for socket units.
16178 * libudev gained a new call udev_device_set_attribute_value()
16179 to set sysfs attributes of a device.
16181 * The udev daemon now sets the default number of worker
16182 processes executed in parallel based on the number of available
16183 CPUs instead of the amount of available RAM. This is supposed
16184 to provide a more reliable default and limit a too aggressive
16185 parallelism for setups with 1000s of devices connected.
16187 Contributions from: Auke Kok, Colin Walters, Cristian
16188 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
16189 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
16190 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
16191 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
16192 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
16193 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
16194 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
16195 Zbigniew Jędrzejewski-Szmek
16199 * Configuration of unit files may now be extended via drop-in
16200 files without having to edit/override the unit files
16201 themselves. More specifically, if the administrator wants to
16202 change one value for a service file foobar.service he can
16203 now do so by dropping in a configuration snippet into
16204 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
16205 will load all these snippets and apply them on top of the
16206 main unit configuration file, possibly extending or
16207 overriding its settings. Using these drop-in snippets is
16208 generally nicer than the two earlier options for changing
16209 unit files locally: copying the files from
16210 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
16211 them there; or creating a new file in /etc/systemd/system/
16212 that incorporates the original one via ".include". Drop-in
16213 snippets into these .d/ directories can be placed in any
16214 directory systemd looks for units in, and the usual
16215 overriding semantics between /usr/lib, /etc and /run apply
16218 * Most unit file settings which take lists of items can now be
16219 reset by assigning the empty string to them. For example,
16220 normally, settings such as Environment=FOO=BAR append a new
16221 environment variable assignment to the environment block,
16222 each time they are used. By assigning Environment= the empty
16223 string the environment block can be reset to empty. This is
16224 particularly useful with the .d/*.conf drop-in snippets
16225 mentioned above, since this adds the ability to reset list
16226 settings from vendor unit files via these drop-ins.
16228 * systemctl gained a new "list-dependencies" command for
16229 listing the dependencies of a unit recursively.
16231 * Inhibitors are now honored and listed by "systemctl
16232 suspend", "systemctl poweroff" (and similar) too, not only
16233 GNOME. These commands will also list active sessions by
16236 * Resource limits (as exposed by the various control group
16237 controllers) can now be controlled dynamically at runtime
16238 for all units. More specifically, you can now use a command
16239 like "systemctl set-cgroup-attr foobar.service cpu.shares
16240 2000" to alter the CPU shares a specific service gets. These
16241 settings are stored persistently on disk, and thus allow the
16242 administrator to easily adjust the resource usage of
16243 services with a few simple commands. This dynamic resource
16244 management logic is also available to other programs via the
16245 bus. Almost any kernel cgroup attribute and controller is
16248 * systemd-vconsole-setup will now copy all font settings to
16249 all allocated VTs, where it previously applied them only to
16252 * libsystemd-login gained the new sd_session_get_tty() API
16255 * This release drops support for a few legacy or
16256 distribution-specific LSB facility names when parsing init
16257 scripts: $x-display-manager, $mail-transfer-agent,
16258 $mail-transport-agent, $mail-transfer-agent, $smtp,
16259 $null. Also, the mail-transfer-agent.target unit backing
16260 this has been removed. Distributions which want to retain
16261 compatibility with this should carry the burden for
16262 supporting this themselves and patch support for these back
16263 in, if they really need to. Also, the facilities $syslog and
16264 $local_fs are now ignored, since systemd does not support
16265 early-boot LSB init scripts anymore, and these facilities
16266 are implied anyway for normal services. syslog.target has
16269 * There are new bus calls on PID1's Manager object for
16270 cancelling jobs, and removing snapshot units. Previously,
16271 both calls were only available on the Job and Snapshot
16272 objects themselves.
16274 * systemd-journal-gatewayd gained SSL support.
16276 * The various "environment" files, such as /etc/locale.conf
16277 now support continuation lines with a backslash ("\") as
16278 last character in the line, similarly in style (but different)
16279 to how this is supported in shells.
16281 * For normal user processes the _SYSTEMD_USER_UNIT= field is
16282 now implicitly appended to every log entry logged. systemctl
16283 has been updated to filter by this field when operating on a
16284 user systemd instance.
16286 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
16287 CAP_AUDIT_CONTROL capabilities to the capabilities set for
16288 the container. This makes it easier to boot unmodified
16289 Fedora systems in a container, which however still requires
16290 audit=0 to be passed on the kernel command line. Auditing in
16291 kernel and userspace is unfortunately still too broken in
16292 context of containers, hence we recommend compiling it out
16293 of the kernel or using audit=0. Hopefully this will be fixed
16294 one day for good in the kernel.
16296 * nspawn gained the new --bind= and --bind-ro= parameters to
16297 bind mount specific directories from the host into the
16300 * nspawn will now mount its own devpts file system instance
16301 into the container, in order not to leak pty devices from
16302 the host into the container.
16304 * systemd will now read the firmware boot time performance
16305 information from the EFI variables, if the used boot loader
16306 supports this, and takes it into account for boot performance
16307 analysis via "systemd-analyze". This is currently supported
16308 only in conjunction with Gummiboot, but could be supported
16309 by other boot loaders too. For details see:
16311 https://systemd.io/BOOT_LOADER_INTERFACE
16313 * A new generator has been added that automatically mounts the
16314 EFI System Partition (ESP) to /boot, if that directory
16315 exists, is empty, and no other file system has been
16316 configured to be mounted there.
16318 * logind will now send out PrepareForSleep(false) out
16319 unconditionally, after coming back from suspend. This may be
16320 used by applications as asynchronous notification for
16321 system resume events.
16323 * "systemctl unlock-sessions" has been added, that allows
16324 unlocking the screens of all user sessions at once, similar
16325 to how "systemctl lock-sessions" already locked all users
16326 sessions. This is backed by a new D-Bus call UnlockSessions().
16328 * "loginctl seat-status" will now show the master device of a
16329 seat. (i.e. the device of a seat that needs to be around for
16330 the seat to be considered available, usually the graphics
16333 * tmpfiles gained a new "X" line type, that allows
16334 configuration of files and directories (with wildcards) that
16335 shall be excluded from automatic cleanup ("aging").
16337 * udev default rules set the device node permissions now only
16338 at "add" events, and do not change them any longer with a
16339 later "change" event.
16341 * The log messages for lid events and power/sleep keypresses
16342 now carry a message ID.
16344 * We now have a substantially larger unit test suite, but this
16345 continues to be work in progress.
16347 * udevadm hwdb gained a new --root= parameter to change the
16348 root directory to operate relative to.
16350 * logind will now issue a background sync() request to the kernel
16351 early at shutdown, so that dirty buffers are flushed to disk early
16352 instead of at the last moment, in order to optimize shutdown
16355 * A new bootctl tool has been added that is an interface for
16356 certain boot loader operations. This is currently a preview
16357 and is likely to be extended into a small mechanism daemon
16358 like timedated, localed, hostnamed, and can be used by
16359 graphical UIs to enumerate available boot options, and
16360 request boot into firmware operations.
16362 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
16363 the rest of the package. It also has been updated to work
16364 correctly in initrds.
16366 * polkit previously has been runtime optional, and is now also
16367 compile time optional via a configure switch.
16369 * systemd-analyze has been reimplemented in C. Also "systemctl
16370 dot" has moved into systemd-analyze.
16372 * "systemctl status" with no further parameters will now print
16373 the status of all active or failed units.
16375 * Operations such as "systemctl start" can now be executed
16376 with a new mode "--irreversible" which may be used to queue
16377 operations that cannot accidentally be reversed by a later
16378 job queuing. This is by default used to make shutdown
16379 requests more robust.
16381 * The Python API of systemd now gained a new module for
16382 reading journal files.
16384 * A new tool kernel-install has been added that can install
16385 kernel images according to the Boot Loader Specification:
16387 https://systemd.io/BOOT_LOADER_SPECIFICATION
16389 * Boot time console output has been improved to provide
16390 animated boot time output for hanging jobs.
16392 * A new tool systemd-activate has been added which can be used
16393 to test socket activation with, directly from the command
16394 line. This should make it much easier to test and debug
16395 socket activation in daemons.
16397 * journalctl gained a new "--reverse" (or -r) option to show
16398 journal output in reverse order (i.e. newest line first).
16400 * journalctl gained a new "--pager-end" (or -e) option to jump
16401 to immediately jump to the end of the journal in the
16402 pager. This is only supported in conjunction with "less".
16404 * journalctl gained a new "--user-unit=" option, that works
16405 similarly to "--unit=" but filters for user units rather than
16408 * A number of unit files to ease adoption of systemd in
16409 initrds has been added. This moves some minimal logic from
16410 the various initrd implementations into systemd proper.
16412 * The journal files are now owned by a new group
16413 "systemd-journal", which exists specifically to allow access
16414 to the journal, and nothing else. Previously, we used the
16415 "adm" group for that, which however possibly covers more
16416 than just journal/log file access. This new group is now
16417 already used by systemd-journal-gatewayd to ensure this
16418 daemon gets access to the journal files and as little else
16419 as possible. Note that "make install" will also set FS ACLs
16420 up for /var/log/journal to give "adm" and "wheel" read
16421 access to it, in addition to "systemd-journal" which owns
16422 the journal files. We recommend that packaging scripts also
16423 add read access to "adm" + "wheel" to /var/log/journal, and
16424 all existing/future journal files. To normal users and
16425 administrators little changes, however packagers need to
16426 ensure to create the "systemd-journal" system group at
16427 package installation time.
16429 * The systemd-journal-gatewayd now runs as unprivileged user
16430 systemd-journal-gateway:systemd-journal-gateway. Packaging
16431 scripts need to create these system user/group at
16434 * timedated now exposes a new boolean property CanNTP that
16435 indicates whether a local NTP service is available or not.
16437 * systemd-detect-virt will now also detect xen PVs
16439 * The pstore file system is now mounted by default, if it is
16442 * In addition to the SELinux and IMA policies we will now also
16443 load SMACK policies at early boot.
16445 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
16446 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
16447 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
16448 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
16449 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
16450 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
16451 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
16452 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
16453 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
16454 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
16455 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
16456 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
16457 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
16458 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
16462 * Timer units now support calendar time events in addition to
16463 monotonic time events. That means you can now trigger a unit
16464 based on a calendar time specification such as "Thu,Fri
16465 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
16466 or fifth day of any month of the year 2013, given that it is
16467 a Thursday or a Friday. This brings timer event support
16468 considerably closer to cron's capabilities. For details on
16469 the supported calendar time specification language see
16472 * udev now supports a number of different naming policies for
16473 network interfaces for predictable names, and a combination
16474 of these policies is now the default. Please see this wiki
16475 document for details:
16477 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
16479 * Auke Kok's bootchart implementation has been added to the
16480 systemd tree. It is an optional component that can graph the
16481 boot in quite some detail. It is one of the best bootchart
16482 implementations around and minimal in its code and
16485 * nss-myhostname has been integrated into the systemd source
16486 tree. nss-myhostname guarantees that the local hostname
16487 always stays resolvable via NSS. It has been a weak
16488 requirement of systemd-hostnamed since a long time, and
16489 since its code is actually trivial we decided to just
16490 include it in systemd's source tree. It can be turned off
16491 with a configure switch.
16493 * The read-ahead logic is now capable of properly detecting
16494 whether a btrfs file system is on SSD or rotating media, in
16495 order to optimize the read-ahead scheme. Previously, it was
16496 only capable of detecting this on traditional file systems
16499 * In udev, additional device properties are now read from the
16500 IAB in addition to the OUI database. Also, Bluetooth company
16501 identities are attached to the devices as well.
16503 * In service files %U may be used as specifier that is
16504 replaced by the configured user name of the service.
16506 * nspawn may now be invoked without a controlling TTY. This
16507 makes it suitable for invocation as its own service. This
16508 may be used to set up a simple containerized server system
16509 using only core OS tools.
16511 * systemd and nspawn can now accept socket file descriptors
16512 when they are started for socket activation. This enables
16513 implementation of socket activated nspawn
16514 containers. i.e. think about autospawning an entire OS image
16515 when the first SSH or HTTP connection is received. We expect
16516 that similar functionality will also be added to libvirt-lxc
16519 * journalctl will now suppress ANSI color codes when
16520 presenting log data.
16522 * systemctl will no longer show control group information for
16523 a unit if the control group is empty anyway.
16525 * logind can now automatically suspend/hibernate/shutdown the
16528 * /etc/machine-info and hostnamed now also expose the chassis
16529 type of the system. This can be used to determine whether
16530 the local system is a laptop, desktop, handset or
16531 tablet. This information may either be configured by the
16532 user/vendor or is automatically determined from ACPI and DMI
16533 information if possible.
16535 * A number of polkit actions are now bound together with "imply"
16536 rules. This should simplify creating UIs because many actions
16537 will now authenticate similar ones as well.
16539 * Unit files learnt a new condition ConditionACPower= which
16540 may be used to conditionalize a unit depending on whether an
16541 AC power source is connected or not, of whether the system
16542 is running on battery power.
16544 * systemctl gained a new "is-failed" verb that may be used in
16545 shell scripts and suchlike to check whether a specific unit
16546 is in the "failed" state.
16548 * The EnvironmentFile= setting in unit files now supports file
16549 globbing, and can hence be used to easily read a number of
16550 environment files at once.
16552 * systemd will no longer detect and recognize specific
16553 distributions. All distribution-specific #ifdeffery has been
16554 removed, systemd is now fully generic and
16555 distribution-agnostic. Effectively, not too much is lost as
16556 a lot of the code is still accessible via explicit configure
16557 switches. However, support for some distribution specific
16558 legacy configuration file formats has been dropped. We
16559 recommend distributions to simply adopt the configuration
16560 files everybody else uses now and convert the old
16561 configuration from packaging scripts. Most distributions
16562 already did that. If that's not possible or desirable,
16563 distributions are welcome to forward port the specific
16564 pieces of code locally from the git history.
16566 * When logging a message about a unit systemd will now always
16567 log the unit name in the message meta data.
16569 * localectl will now also discover system locale data that is
16570 not stored in locale archives, but directly unpacked.
16572 * logind will no longer unconditionally use framebuffer
16573 devices as seat masters, i.e. as devices that are required
16574 to be existing before a seat is considered preset. Instead,
16575 it will now look for all devices that are tagged as
16576 "seat-master" in udev. By default, framebuffer devices will
16577 be marked as such, but depending on local systems, other
16578 devices might be marked as well. This may be used to
16579 integrate graphics cards using closed source drivers (such
16580 as NVidia ones) more nicely into logind. Note however, that
16581 we recommend using the open source NVidia drivers instead,
16582 and no udev rules for the closed-source drivers will be
16583 shipped from us upstream.
16585 Contributions from: Adam Williamson, Alessandro Crismani, Auke
16586 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
16587 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
16588 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
16589 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
16590 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
16591 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
16592 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
16593 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
16594 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
16595 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
16596 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
16601 * udev gained support for loading additional device properties
16602 from an indexed database that is keyed by vendor/product IDs
16603 and similar device identifiers. For the beginning this
16604 "hwdb" is populated with data from the well-known PCI and
16605 USB database, but also includes PNP, ACPI and OID data. In
16606 the longer run this indexed database shall grow into
16607 becoming the one central database for non-essential
16608 userspace device metadata. Previously, data from the PCI/USB
16609 database was only attached to select devices, since the
16610 lookup was a relatively expensive operation due to O(n) time
16611 complexity (with n being the number of entries in the
16612 database). Since this is now O(1), we decided to add in this
16613 data for all devices where this is available, by
16614 default. Note that the indexed database needs to be rebuilt
16615 when new data files are installed. To achieve this you need
16616 to update your packaging scripts to invoke "udevadm hwdb
16617 --update" after installation of hwdb data files. For
16618 RPM-based distributions we introduced the new
16619 %udev_hwdb_update macro for this purpose.
16621 * The Journal gained support for the "Message Catalog", an
16622 indexed database to link up additional information with
16623 journal entries. For further details please check:
16625 https://www.freedesktop.org/wiki/Software/systemd/catalog
16627 The indexed message catalog database also needs to be
16628 rebuilt after installation of message catalog files. Use
16629 "journalctl --update-catalog" for this. For RPM-based
16630 distributions we introduced the %journal_catalog_update
16631 macro for this purpose.
16633 * The Python Journal bindings gained support for the standard
16634 Python logging framework.
16636 * The Journal API gained new functions for checking whether
16637 the underlying file system of a journal file is capable of
16638 properly reporting file change notifications, or whether
16639 applications that want to reflect journal changes "live"
16640 need to recheck journal files continuously in appropriate
16643 * It is now possible to set the "age" field for tmpfiles
16644 entries to 0, indicating that files matching this entry
16645 shall always be removed when the directories are cleaned up.
16647 * coredumpctl gained a new "gdb" verb which invokes gdb
16648 right-away on the selected coredump.
16650 * There's now support for "hybrid sleep" on kernels that
16651 support this, in addition to "suspend" and "hibernate". Use
16652 "systemctl hybrid-sleep" to make use of this.
16654 * logind's HandleSuspendKey= setting (and related settings)
16655 now gained support for a new "lock" setting to simply
16656 request the screen lock on all local sessions, instead of
16657 actually executing a suspend or hibernation.
16659 * systemd will now mount the EFI variables file system by
16662 * Socket units now gained support for configuration of the
16663 SMACK security label.
16665 * timedatectl will now output the time of the last and next
16666 daylight saving change.
16668 * We dropped support for various legacy and distro-specific
16669 concepts, such as insserv, early-boot SysV services
16670 (i.e. those for non-standard runlevels such as 'b' or 'S')
16671 or ArchLinux /etc/rc.conf support. We recommend the
16672 distributions who still need support this to either continue
16673 to maintain the necessary patches downstream, or find a
16674 different solution. (Talk to us if you have questions!)
16676 * Various systemd components will now bypass polkit checks for
16677 root and otherwise handle properly if polkit is not found to
16678 be around. This should fix most issues for polkit-less
16679 systems. Quite frankly this should have been this way since
16680 day one. It is absolutely our intention to make systemd work
16681 fine on polkit-less systems, and we consider it a bug if
16682 something does not work as it should if polkit is not around.
16684 * For embedded systems it is now possible to build udev and
16685 systemd without blkid and/or kmod support.
16687 * "systemctl switch-root" is now capable of switching root
16688 more than once. I.e. in addition to transitions from the
16689 initrd to the host OS it is now possible to transition to
16690 further OS images from the host. This is useful to implement
16691 offline updating tools.
16693 * Various other additions have been made to the RPM macros
16694 shipped with systemd. Use %udev_rules_update() after
16695 installing new udev rules files. %_udevhwdbdir,
16696 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
16697 %_sysctldir are now available which resolve to the right
16698 directories for packages to place various data files in.
16700 * journalctl gained the new --full switch (in addition to
16701 --all, to disable ellipsation for long messages.
16703 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
16704 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
16705 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
16706 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
16707 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
16708 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
16709 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
16710 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
16711 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
16715 * journalctl gained new --since= and --until= switches to
16716 filter by time. It also now supports nice filtering for
16717 units via --unit=/-u.
16719 * Type=oneshot services may use ExecReload= and do the
16722 * The journal daemon now supports time-based rotation and
16723 vacuuming, in addition to the usual disk-space based
16726 * The journal will now index the available field values for
16727 each field name. This enables clients to show pretty drop
16728 downs of available match values when filtering. The bash
16729 completion of journalctl has been updated
16730 accordingly. journalctl gained a new switch -F to list all
16731 values a certain field takes in the journal database.
16733 * More service events are now written as structured messages
16734 to the journal, and made recognizable via message IDs.
16736 * The timedated, localed and hostnamed mini-services which
16737 previously only provided support for changing time, locale
16738 and hostname settings from graphical DEs such as GNOME now
16739 also have a minimal (but very useful) text-based client
16740 utility each. This is probably the nicest way to changing
16741 these settings from the command line now, especially since
16742 it lists available options and is fully integrated with bash
16745 * There's now a new tool "systemd-coredumpctl" to list and
16746 extract coredumps from the journal.
16748 * We now install a README each in /var/log/ and
16749 /etc/rc.d/init.d explaining where the system logs and init
16750 scripts went. This hopefully should help folks who go to
16751 that dirs and look into the otherwise now empty void and
16752 scratch their heads.
16754 * When user-services are invoked (by systemd --user) the
16755 $MANAGERPID env var is set to the PID of systemd.
16757 * SIGRTMIN+24 when sent to a --user instance will now result
16758 in immediate termination of systemd.
16760 * gatewayd received numerous feature additions such as a
16761 "follow" mode, for live syncing and filtering.
16763 * browse.html now allows filtering and showing detailed
16764 information on specific entries. Keyboard navigation and
16765 mouse screen support has been added.
16767 * gatewayd/journalctl now supports HTML5/JSON
16768 Server-Sent-Events as output.
16770 * The SysV init script compatibility logic will now
16771 heuristically determine whether a script supports the
16772 "reload" verb, and only then make this available as
16773 "systemctl reload".
16775 * "systemctl status --follow" has been removed, use "journalctl
16778 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
16779 have been removed since they are hardly useful to be
16782 * And I'd like to take the opportunity to specifically mention
16783 Zbigniew for his great contributions. Zbigniew, you rock!
16785 Contributions from: Andrew Eikum, Christian Hesse, Colin
16786 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
16787 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
16788 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
16789 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
16790 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
16791 Jędrzejewski-Szmek, Сковорода Никита Андреевич
16795 * If /etc/vconsole.conf is non-existent or empty we will no
16796 longer load any console font or key map at boot by
16797 default. Instead the kernel defaults will be left
16798 intact. This is definitely the right thing to do, as no
16799 configuration should mean no configuration, and hard-coding
16800 font names that are different on all archs is probably a bad
16801 idea. Also, the kernel default key map and font should be
16802 good enough for most cases anyway, and mostly identical to
16803 the userspace fonts/key maps we previously overloaded them
16804 with. If distributions want to continue to default to a
16805 non-kernel font or key map they should ship a default
16806 /etc/vconsole.conf with the appropriate contents.
16808 Contributions from: Colin Walters, Daniel J Walsh, Dave
16809 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
16810 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
16814 * journalctl gained a new --cursor= switch to show entries
16815 starting from the specified location in the journal.
16817 * We now enforce a size limit on journal entry fields exported
16818 with "-o json" in journalctl. Fields larger than 4K will be
16819 assigned null. This can be turned off with --all.
16821 * An (optional) journal gateway daemon is now available as
16822 "systemd-journal-gatewayd.service". This service provides
16823 access to the journal via HTTP and JSON. This functionality
16824 will be used to implement live log synchronization in both
16825 pull and push modes, but has various other users too, such
16826 as easy log access for debugging of embedded devices. Right
16827 now it is already useful to retrieve the journal via HTTP:
16829 # systemctl start systemd-journal-gatewayd.service
16830 # wget http://localhost:19531/entries
16832 This will download the journal contents in a
16833 /var/log/messages compatible format. The same as JSON:
16835 # curl -H"Accept: application/json" http://localhost:19531/entries
16837 This service is also accessible via a web browser where a
16838 single static HTML5 app is served that uses the JSON logic
16839 to enable the user to do some basic browsing of the
16840 journal. This will be extended later on. Here's an example
16841 screenshot of this app in its current state:
16843 https://0pointer.de/public/journal-gatewayd
16845 Contributions from: Kay Sievers, Lennart Poettering, Robert
16846 Milasan, Tom Gundersen
16850 * The bash completion logic is now available for journalctl
16853 * We do not mount the "cpuset" controller anymore together with
16854 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
16855 started if no parameters are assigned to it. "cpuset" hence
16856 broke code that assumed it could create "cpu" groups and
16859 * journalctl -f will now subscribe to terminal size changes,
16860 and line break accordingly.
16862 Contributions from: Dave Reisner, Kay Sievers, Lennart
16863 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
16867 * nspawn will now create a symlink /etc/localtime in the
16868 container environment, copying the host's timezone
16869 setting. Previously this has been done via a bind mount, but
16870 since symlinks cannot be bind mounted this has now been
16871 changed to create/update the appropriate symlink.
16873 * journalctl -n's line number argument is now optional, and
16874 will default to 10 if omitted.
16876 * journald will now log the maximum size the journal files may
16877 take up on disk. This is particularly useful if the default
16878 built-in logic of determining this parameter from the file
16879 system size is used. Use "systemctl status
16880 systemd-journald.service" to see this information.
16882 * The multi-seat X wrapper tool has been stripped down. As X
16883 is now capable of enumerating graphics devices via udev in a
16884 seat-aware way the wrapper is not strictly necessary
16885 anymore. A stripped down temporary stop-gap is still shipped
16886 until the upstream display managers have been updated to
16887 fully support the new X logic. Expect this wrapper to be
16888 removed entirely in one of the next releases.
16890 * HandleSleepKey= in logind.conf has been split up into
16891 HandleSuspendKey= and HandleHibernateKey=. The old setting
16892 is not available anymore. X11 and the kernel are
16893 distinguishing between these keys and we should too. This
16894 also means the inhibition lock for these keys has been split
16897 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
16898 Poettering, Lukas Nykryn, Václav Pavlín
16902 * Whenever a unit changes state we will now log this to the
16903 journal and show along the unit's own log output in
16904 "systemctl status".
16906 * ConditionPathIsMountPoint= can now properly detect bind
16907 mount points too. (Previously, a bind mount of one file
16908 system to another place in the same file system could not be
16909 detected as mount, since they shared struct stat's st_dev
16912 * We will now mount the cgroup controllers cpu, cpuacct,
16913 cpuset and the controllers net_cls, net_prio together by
16916 * nspawn containers will now have a virtualized boot
16917 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
16918 over with a randomized ID at container initialization). This
16919 has the effect of making "journalctl -b" do the right thing
16922 * The JSON output journal serialization has been updated not
16923 to generate "endless" list objects anymore, but rather one
16924 JSON object per line. This is more in line how most JSON
16925 parsers expect JSON objects. The new output mode
16926 "json-pretty" has been added to provide similar output, but
16927 neatly aligned for readability by humans.
16929 * We dropped all explicit sync() invocations in the shutdown
16930 code. The kernel does this implicitly anyway in the kernel
16931 reboot() syscall. halt(8)'s -n option is now a compatibility
16934 * We now support virtualized reboot() in containers, as
16935 supported by newer kernels. We will fall back to exit() if
16936 CAP_SYS_REBOOT is not available to the container. Also,
16937 nspawn makes use of this now and will actually reboot the
16938 container if the containerized OS asks for that.
16940 * journalctl will only show local log output by default
16941 now. Use --merge (-m) to show remote log output, too.
16943 * libsystemd-journal gained the new sd_journal_get_usage()
16944 call to determine the current disk usage of all journal
16945 files. This is exposed in the new "journalctl --disk-usage"
16948 * journald gained a new configuration setting SplitMode= in
16949 journald.conf which may be used to control how user journals
16950 are split off. See journald.conf(5) for details.
16952 * A new condition type ConditionFileNotEmpty= has been added.
16954 * tmpfiles' "w" lines now support file globbing, to write
16955 multiple files at once.
16957 * We added Python bindings for the journal submission
16958 APIs. More Python APIs for a number of selected APIs will
16959 likely follow. Note that we intend to add native bindings
16960 only for the Python language, as we consider it common
16961 enough to deserve bindings shipped within systemd. There are
16962 various projects outside of systemd that provide bindings
16963 for languages such as PHP or Lua.
16965 * Many conditions will now resolve specifiers such as %i. In
16966 addition, PathChanged= and related directives of .path units
16967 now support specifiers as well.
16969 * There's now a new RPM macro definition for the system preset
16972 * journald will now warn if it ca not forward a message to the
16973 syslog daemon because its socket is full.
16975 * timedated will no longer write or process /etc/timezone,
16976 except on Debian. As we do not support late mounted /usr
16977 anymore /etc/localtime always being a symlink is now safe,
16978 and hence the information in /etc/timezone is not necessary
16981 * logind will now always reserve one VT for a text getty (VT6
16982 by default). Previously if more than 6 X sessions where
16983 started they took up all the VTs with auto-spawned gettys,
16984 so that no text gettys were available anymore.
16986 * udev will now automatically inform the btrfs kernel logic
16987 about btrfs RAID components showing up. This should make
16988 simple hotplug based btrfs RAID assembly work.
16990 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
16991 (but not for its children which will stay at the kernel
16992 default). This should allow setups with a lot more listening
16995 * systemd will now always pass the configured timezone to the
16996 kernel at boot. timedated will do the same when the timezone
16999 * logind's inhibition logic has been updated. By default,
17000 logind will now handle the lid switch, the power and sleep
17001 keys all the time, even in graphical sessions. If DEs want
17002 to handle these events on their own they should take the new
17003 handle-power-key, handle-sleep-key and handle-lid-switch
17004 inhibitors during their runtime. A simple way to achieve
17005 that is to invoke the DE wrapped in an invocation of:
17007 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
17009 * Access to unit operations is now checked via SELinux taking
17010 the unit file label and client process label into account.
17012 * systemd will now notify the administrator in the journal
17013 when he over-mounts a non-empty directory.
17015 * There are new specifiers that are resolved in unit files,
17016 for the hostname (%H), the machine ID (%m) and the boot ID
17019 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
17020 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
17021 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
17022 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
17023 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
17024 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
17025 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
17029 * Support for reading structured kernel messages from
17030 /dev/kmsg has now been added and is enabled by default.
17032 * Support for reading kernel messages from /proc/kmsg has now
17033 been removed. If you want kernel messages in the journal
17034 make sure to run a recent kernel (>= 3.5) that supports
17035 reading structured messages from /dev/kmsg (see
17036 above). /proc/kmsg is now exclusive property of classic
17037 syslog daemons again.
17039 * The libudev API gained the new
17040 udev_device_new_from_device_id() call.
17042 * The logic for file system namespace (ReadOnlyDirectory=,
17043 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
17044 require pivot_root() anymore. This means fewer temporary
17045 directories are created below /tmp for this feature.
17047 * nspawn containers will now see and receive all submounts
17048 made on the host OS below the root file system of the
17051 * Forward Secure Sealing is now supported for Journal files,
17052 which provide cryptographical sealing of journal files so
17053 that attackers cannot alter log history anymore without this
17054 being detectable. Lennart will soon post a blog story about
17055 this explaining it in more detail.
17057 * There are two new service settings RestartPreventExitStatus=
17058 and SuccessExitStatus= which allow configuration of exit
17059 status (exit code or signal) which will be excepted from the
17060 restart logic, resp. consider successful.
17062 * journalctl gained the new --verify switch that can be used
17063 to check the integrity of the structure of journal files and
17064 (if Forward Secure Sealing is enabled) the contents of
17067 * nspawn containers will now be run with /dev/stdin, /dev/fd/
17068 and similar symlinks pre-created. This makes running shells
17069 as container init process a lot more fun.
17071 * The fstab support can now handle PARTUUID= and PARTLABEL=
17074 * A new ConditionHost= condition has been added to match
17075 against the hostname (with globs) and machine ID. This is
17076 useful for clusters where a single OS image is used to
17077 provision a large number of hosts which shall run slightly
17078 different sets of services.
17080 * Services which hit the restart limit will now be placed in a
17083 Contributions from: Bertram Poettering, Dave Reisner, Huang
17084 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
17085 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
17089 * When running in --user mode systemd will now become a
17090 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
17091 tree a lot more organized.
17093 * A new PartOf= unit dependency type has been introduced that
17094 may be used to group services in a natural way.
17096 * "systemctl enable" may now be used to enable instances of
17099 * journalctl now prints error log levels in red, and
17100 warning/notice log levels in bright white. It also supports
17101 filtering by log level now.
17103 * cgtop gained a new -n switch (similar to top), to configure
17104 the maximum number of iterations to run for. It also gained
17105 -b, to run in batch mode (accepting no input).
17107 * The suffix ".service" may now be omitted on most systemctl
17108 command lines involving service unit names.
17110 * There's a new bus call in logind to lock all sessions, as
17111 well as a loginctl verb for it "lock-sessions".
17113 * libsystemd-logind.so gained a new call sd_journal_perror()
17114 that works similar to libc perror() but logs to the journal
17115 and encodes structured information about the error number.
17117 * /etc/crypttab entries now understand the new keyfile-size=
17120 * shutdown(8) now can send a (configurable) wall message when
17121 a shutdown is cancelled.
17123 * The mount propagation mode for the root file system will now
17124 default to "shared", which is useful to make containers work
17125 nicely out-of-the-box so that they receive new mounts from
17126 the host. This can be undone locally by running "mount
17127 --make-rprivate /" if needed.
17129 * The prefdm.service file has been removed. Distributions
17130 should maintain this unit downstream if they intend to keep
17131 it around. However, we recommend writing normal unit files
17132 for display managers instead.
17134 * Since systemd is a crucial part of the OS we will now
17135 default to a number of compiler switches that improve
17136 security (hardening) such as read-only relocations, stack
17137 protection, and suchlike.
17139 * The TimeoutSec= setting for services is now split into
17140 TimeoutStartSec= and TimeoutStopSec= to allow configuration
17141 of individual time outs for the start and the stop phase of
17144 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
17145 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
17146 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
17147 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
17148 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
17149 Gundersen, Zbigniew Jędrzejewski-Szmek
17153 * The journal and id128 C APIs are now fully documented as man
17156 * Extra safety checks have been added when transitioning from
17157 the initial RAM disk to the main system to avoid accidental
17160 * /etc/crypttab entries now understand the new keyfile-offset=
17163 * systemctl -t can now be used to filter by unit load state.
17165 * The journal C API gained the new sd_journal_wait() call to
17166 make writing synchronous journal clients easier.
17168 * journalctl gained the new -D switch to show journals from a
17169 specific directory.
17171 * journalctl now displays a special marker between log
17172 messages of two different boots.
17174 * The journal is now explicitly flushed to /var via a service
17175 systemd-journal-flush.service, rather than implicitly simply
17176 by seeing /var/log/journal to be writable.
17178 * journalctl (and the journal C APIs) can now match for much
17179 more complex expressions, with alternatives and
17182 * When transitioning from the initial RAM disk to the main
17183 system we will now kill all processes in a killing spree to
17184 ensure no processes stay around by accident.
17186 * Three new specifiers may be used in unit files: %u, %h, %s
17187 resolve to the user name, user home directory resp. user
17188 shell. This is useful for running systemd user instances.
17190 * We now automatically rotate journal files if their data
17191 object hash table gets a fill level > 75%. We also size the
17192 hash table based on the configured maximum file size. This
17193 together should lower hash collisions drastically and thus
17194 speed things up a bit.
17196 * journalctl gained the new "--header" switch to introspect
17197 header data of journal files.
17199 * A new setting SystemCallFilters= has been added to services which may
17200 be used to apply deny lists or allow lists to system calls. This is
17201 based on SECCOMP Mode 2 of Linux 3.5.
17203 * nspawn gained a new --link-journal= switch (and quicker: -j)
17204 to link the container journal with the host. This makes it
17205 very easy to centralize log viewing on the host for all
17206 guests while still keeping the journal files separated.
17208 * Many bugfixes and optimizations
17210 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
17211 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
17212 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
17217 * Several tools now understand kernel command line arguments,
17218 which are only read when run in an initial RAM disk. They
17219 usually follow closely their normal counterparts, but are
17222 * There's a new tool to analyze the readahead files that are
17223 automatically generated at boot. Use:
17225 /usr/lib/systemd/systemd-readahead analyze /.readahead
17227 * We now provide an early debug shell on tty9 if this enabled. Use:
17229 systemctl enable debug-shell.service
17231 * All plymouth related units have been moved into the Plymouth
17232 package. Please make sure to upgrade your Plymouth version
17235 * systemd-tmpfiles now supports getting passed the basename of
17236 a configuration file only, in which case it will look for it
17237 in all appropriate directories automatically.
17239 * udevadm info now takes a /dev or /sys path as argument, and
17240 does the right thing. Example:
17242 udevadm info /dev/sda
17243 udevadm info /sys/class/block/sda
17245 * systemctl now prints a warning if a unit is stopped but a
17246 unit that might trigger it continues to run. Example: a
17247 service is stopped but the socket that activates it is left
17250 * "systemctl status" will now mention if the log output was
17251 shortened due to rotation since a service has been started.
17253 * The journal API now exposes functions to determine the
17254 "cutoff" times due to rotation.
17256 * journald now understands SIGUSR1 and SIGUSR2 for triggering
17257 immediately flushing of runtime logs to /var if possible,
17258 resp. for triggering immediate rotation of the journal
17261 * It is now considered an error if a service is attempted to
17262 be stopped that is not loaded.
17264 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
17266 * systemd-analyze now supports Python 3
17268 * tmpfiles now supports cleaning up directories via aging
17269 where the first level dirs are always kept around but
17270 directories beneath it automatically aged. This is enabled
17271 by prefixing the age field with '~'.
17273 * Seat objects now expose CanGraphical, CanTTY properties
17274 which is required to deal with very fast bootups where the
17275 display manager might be running before the graphics drivers
17276 completed initialization.
17278 * Seat objects now expose a State property.
17280 * We now include RPM macros for service enabling/disabling
17281 based on the preset logic. We recommend RPM based
17282 distributions to make use of these macros if possible. This
17283 makes it simpler to reuse RPM spec files across
17286 * We now make sure that the collected systemd unit name is
17287 always valid when services log to the journal via
17290 * There's a new man page kernel-command-line(7) detailing all
17291 command line options we understand.
17293 * The fstab generator may now be disabled at boot by passing
17294 fstab=0 on the kernel command line.
17296 * A new kernel command line option modules-load= is now understood
17297 to load a specific kernel module statically, early at boot.
17299 * Unit names specified on the systemctl command line are now
17300 automatically escaped as needed. Also, if file system or
17301 device paths are specified they are automatically turned
17302 into the appropriate mount or device unit names. Example:
17304 systemctl status /home
17305 systemctl status /dev/sda
17307 * The SysVConsole= configuration option has been removed from
17308 system.conf parsing.
17310 * The SysV search path is no longer exported on the D-Bus
17313 * The Names= option has been removed from unit file parsing.
17315 * There's a new man page bootup(7) detailing the boot process.
17317 * Every unit and every generator we ship with systemd now
17318 comes with full documentation. The self-explanatory boot is
17321 * A couple of services gained "systemd-" prefixes in their
17322 name if they wrap systemd code, rather than only external
17323 code. Among them fsck@.service which is now
17324 systemd-fsck@.service.
17326 * The HaveWatchdog property has been removed from the D-Bus
17329 * systemd.confirm_spawn= on the kernel command line should now
17332 * There's a new man page crypttab(5) which details all options
17333 we actually understand.
17335 * systemd-nspawn gained a new --capability= switch to pass
17336 additional capabilities to the container.
17338 * timedated will now read known NTP implementation unit names
17339 from /usr/lib/systemd/ntp-units.d/*.list,
17340 systemd-timedated-ntp.target has been removed.
17342 * journalctl gained a new switch "-b" that lists log data of
17343 the current boot only.
17345 * The notify socket is in the abstract namespace again, in
17346 order to support daemons which chroot() at start-up.
17348 * There is a new Storage= configuration option for journald
17349 which allows configuration of where log data should go. This
17350 also provides a way to disable journal logging entirely, so
17351 that data collected is only forwarded to the console, the
17352 kernel log buffer or another syslog implementation.
17354 * Many bugfixes and optimizations
17356 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
17357 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
17358 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
17359 Shawn Landden, Tom Gundersen
17363 * "systemctl help <unit>" now shows the man page if one is
17366 * Several new man pages have been added.
17368 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
17369 MaxLevelConsole= can now be specified in
17370 journald.conf. These options allow reducing the amount of
17371 data stored on disk or forwarded by the log level.
17373 * TimerSlackNSec= can now be specified in system.conf for
17374 PID1. This allows system-wide power savings.
17376 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
17377 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
17382 * logind is now capable of (optionally) handling power and
17383 sleep keys as well as the lid switch.
17385 * journalctl now understands the syntax "journalctl
17386 /usr/bin/avahi-daemon" to get all log output of a specific
17389 * CapabilityBoundingSet= in system.conf now also influences
17390 the capability bound set of usermode helpers of the kernel.
17392 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
17393 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
17394 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
17395 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
17399 * Note that we skipped 139 releases here in order to set the
17400 new version to something that is greater than both udev's
17401 and systemd's most recent version number.
17403 * udev: all udev sources are merged into the systemd source tree now.
17404 All future udev development will happen in the systemd tree. It
17405 is still fully supported to use the udev daemon and tools without
17406 systemd running, like in initramfs or other init systems. Building
17407 udev though, will require the *build* of the systemd tree, but
17408 udev can be properly *run* without systemd.
17410 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
17411 should be used to create dead device nodes as workarounds for broken
17414 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
17415 no longer supported. udev_monitor_new_from_netlink() needs to be
17416 used to subscribe to events.
17418 * udev: when udevd is started by systemd, processes which are left
17419 behind by forking them off of udev rules, are unconditionally cleaned
17420 up and killed now after the event handling has finished. Services or
17421 daemons must be started as systemd services. Services can be
17422 pulled-in by udev to get started, but they can no longer be directly
17423 forked by udev rules.
17425 * udev: the daemon binary is called systemd-udevd now and installed
17426 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
17427 to adapt to that, create symlink, or rename the binary after building
17430 * libudev no longer provides these symbols:
17431 udev_monitor_from_socket()
17432 udev_queue_get_failed_list_entry()
17433 udev_get_{dev,sys,run}_path()
17434 The versions number was bumped and symbol versioning introduced.
17436 * systemd-loginctl and systemd-journalctl have been renamed
17437 to loginctl and journalctl to match systemctl.
17439 * The config files: /etc/systemd/systemd-logind.conf and
17440 /etc/systemd/systemd-journald.conf have been renamed to
17441 logind.conf and journald.conf. Package updates should rename
17442 the files to the new names on upgrade.
17444 * For almost all files the license is now LGPL2.1+, changed
17445 from the previous GPL2.0+. Exceptions are some minor stuff
17446 of udev (which will be changed to LGPL2.1 eventually, too),
17447 and the MIT licensed sd-daemon.[ch] library that is suitable
17448 to be used as drop-in files.
17450 * systemd and logind now handle system sleep states, in
17451 particular suspending and hibernating.
17453 * logind now implements a sleep/shutdown/idle inhibiting logic
17454 suitable for a variety of uses. Soonishly Lennart will blog
17455 about this in more detail.
17457 * var-run.mount and var-lock.mount are no longer provided
17458 (which previously bind mounted these directories to their new
17459 places). Distributions which have not converted these
17460 directories to symlinks should consider stealing these files
17461 from git history and add them downstream.
17463 * We introduced the Documentation= field for units and added
17464 this to all our shipped units. This is useful to make it
17465 easier to explore the boot and the purpose of the various
17468 * All smaller setup units (such as
17469 systemd-vconsole-setup.service) now detect properly if they
17470 are run in a container and are skipped when
17471 appropriate. This guarantees an entirely noise-free boot in
17472 Linux container environments such as systemd-nspawn.
17474 * A framework for implementing offline system updates is now
17475 integrated, for details see:
17476 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
17478 * A new service type Type=idle is available now which helps us
17479 avoiding ugly interleaving of getty output and boot status
17482 * There's now a system-wide CapabilityBoundingSet= option to
17483 globally reduce the set of capabilities for the
17484 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
17485 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
17486 even CAP_NET_ADMIN system-wide for secure systems.
17488 * There are now system-wide DefaultLimitXXX= options to
17489 globally change the defaults of the various resource limits
17490 for all units started by PID 1.
17492 * Harald Hoyer's systemd test suite has been integrated into
17493 systemd which allows easy testing of systemd builds in qemu
17494 and nspawn. (This is really awesome! Ask us for details!)
17496 * The fstab parser is now implemented as generator, not inside
17499 * systemctl will now warn you if .mount units generated from
17500 /etc/fstab are out of date due to changes in fstab that
17501 have not been read by systemd yet.
17503 * systemd is now suitable for usage in initrds. Dracut has
17504 already been updated to make use of this. With this in place
17505 initrds get a slight bit faster but primarily are much
17506 easier to introspect and debug since "systemctl status" in
17507 the host system can be used to introspect initrd services,
17508 and the journal from the initrd is kept around too.
17510 * systemd-delta has been added, a tool to explore differences
17511 between user/admin configuration and vendor defaults.
17513 * PrivateTmp= now affects both /tmp and /var/tmp.
17515 * Boot time status messages are now much prettier and feature
17516 proper english language. Booting up systemd has never been
17519 * Read-ahead pack files now include the inode number of all
17520 files to pre-cache. When the inode changes the pre-caching
17521 is not attempted. This should be nicer to deal with updated
17522 packages which might result in changes of read-ahead
17525 * We now temporaritly lower the kernel's read_ahead_kb variable
17526 when collecting read-ahead data to ensure the kernel's
17527 built-in read-ahead does not add noise to our measurements
17528 of necessary blocks to pre-cache.
17530 * There's now RequiresMountsFor= to add automatic dependencies
17531 for all mounts necessary for a specific file system path.
17533 * MountAuto= and SwapAuto= have been removed from
17534 system.conf. Mounting file systems at boot has to take place
17537 * nspawn now learned a new switch --uuid= to set the machine
17538 ID on the command line.
17540 * nspawn now learned the -b switch to automatically search
17541 for an init system.
17543 * vt102 is now the default TERM for serial TTYs, upgraded from
17546 * systemd-logind now works on VT-less systems.
17548 * The build tree has been reorganized. The individual
17549 components now have directories of their own.
17551 * A new condition type ConditionPathIsReadWrite= is now available.
17553 * nspawn learned the new -C switch to create cgroups for the
17554 container in other hierarchies.
17556 * We now have support for hardware watchdogs, configurable in
17559 * The scheduled shutdown logic now has a public API.
17561 * We now mount /tmp as tmpfs by default, but this can be
17562 masked and /etc/fstab can override it.
17564 * Since udisks does not make use of /media anymore we are not
17565 mounting a tmpfs on it anymore.
17567 * journalctl gained a new --local switch to only interleave
17568 locally generated journal files.
17570 * We can now load the IMA policy at boot automatically.
17572 * The GTK tools have been split off into a systemd-ui.
17574 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
17575 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
17576 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
17577 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
17578 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
17579 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
17580 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
17581 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
17582 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
17587 * This is mostly a bugfix release
17589 * Support optional initialization of the machine ID from the
17590 KVM or container configured UUID.
17592 * Support immediate reboots with "systemctl reboot -ff"
17594 * Show /etc/os-release data in systemd-analyze output
17596 * Many bugfixes for the journal, including endianness fixes and
17597 ensuring that disk space enforcement works
17599 * sd-login.h is C++ compatible again
17601 * Extend the /etc/os-release format on request of the Debian
17604 * We now refuse non-UTF8 strings used in various configuration
17605 and unit files. This is done to ensure we do not pass invalid
17606 data over D-Bus or expose it elsewhere.
17608 * Register Mimo USB Screens as suitable for automatic seat
17611 * Read SELinux client context from journal clients in a race
17614 * Reorder configuration file lookup order. /etc now always
17615 overrides /run in order to allow the administrator to always
17616 and unconditionally override vendor-supplied or
17617 automatically generated data.
17619 * The various user visible bits of the journal now have man
17620 pages. We still lack man pages for the journal API calls
17623 * We now ship all man pages in HTML format again in the
17626 Contributions from: Dave Reisner, Dirk Eibach, Frederic
17627 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
17628 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
17633 * This is mostly a bugfix release
17635 * systems lacking /etc/os-release are no longer supported.
17637 * Various functionality updates to libsystemd-login.so
17639 * Track class of PAM logins to distinguish greeters from
17640 normal user logins.
17642 Contributions from: Kay Sievers, Lennart Poettering, Michael
17647 * This is an important bugfix release for v41.
17649 * Building man pages is now optional which should be useful
17650 for those building systemd from git but unwilling to install
17653 * Watchdog support for supervising services is now usable. In
17654 a future release support for hardware watchdogs
17655 (i.e. /dev/watchdog) will be added building on this.
17657 * Service start rate limiting is now configurable and can be
17658 turned off per service. When a start rate limit is hit a
17659 reboot can automatically be triggered.
17661 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
17663 Contributions from: Benjamin Franzke, Bill Nottingham,
17664 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
17665 Schmidt, Michał Górny, Piotr Drąg
17669 * The systemd binary is installed /usr/lib/systemd/systemd now;
17670 An existing /sbin/init symlink needs to be adapted with the
17673 * The code that loads kernel modules has been ported to invoke
17674 libkmod directly, instead of modprobe. This means we do not
17675 support systems with module-init-tools anymore.
17677 * Watchdog support is now already useful, but still not
17680 * A new kernel command line option systemd.setenv= is
17681 understood to set system wide environment variables
17682 dynamically at boot.
17684 * We now limit the set of capabilities of systemd-journald.
17686 * We now set SIGPIPE to ignore by default, since it only is
17687 useful in shell pipelines, and has little use in general
17688 code. This can be disabled with IgnoreSIPIPE=no in unit
17691 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
17692 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
17697 * This is mostly a bugfix release
17699 * We now expose the reason why a service failed in the
17700 "Result" D-Bus property.
17702 * Rudimentary service watchdog support (will be completed over
17703 the next few releases.)
17705 * When systemd forks off in order execute some service we will
17706 now immediately changes its argv[0] to reflect which process
17707 it will execute. This is useful to minimize the time window
17708 with a generic argv[0], which makes bootcharts more useful
17710 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
17711 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
17712 Mike Kazantsev, Ray Strode
17716 * This is mostly a test release, but incorporates many
17719 * New systemd-cgtop tool to show control groups by their
17722 * Linking against libacl for ACLs is optional again. If
17723 disabled, support tracking device access for active logins
17724 goes becomes unavailable, and so does access to the user
17725 journals by the respective users.
17727 * If a group "adm" exists, journal files are automatically
17728 owned by them, thus allow members of this group full access
17729 to the system journal as well as all user journals.
17731 * The journal now stores the SELinux context of the logging
17732 client for all entries.
17734 * Add C++ inclusion guards to all public headers
17736 * New output mode "cat" in the journal to print only text
17737 messages, without any meta data like date or time.
17739 * Include tiny X server wrapper as a temporary stop-gap to
17740 teach XOrg udev display enumeration. This is used by display
17741 managers such as gdm, and will go away as soon as XOrg
17742 learned native udev hotplugging for display devices.
17744 * Add new systemd-cat tool for executing arbitrary programs
17745 with STDERR/STDOUT connected to the journal. Can also act as
17746 BSD logger replacement, and does so by default.
17748 * Optionally store all locally generated coredumps in the
17749 journal along with meta data.
17751 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
17752 writing short strings to files (for usage for /sys), and for
17753 creating symlinks, character and block device nodes.
17755 * New unit file option ControlGroupPersistent= to make cgroups
17756 persistent, following the mechanisms outlined in
17757 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
17759 * Support multiple local RTCs in a sane way
17761 * No longer monopolize IO when replaying readahead data on
17762 rotating disks, since we might starve non-file-system IO to
17763 death, since fanotify() will not see accesses done by blkid,
17766 * Do not show kernel threads in systemd-cgls anymore, unless
17767 requested with new -k switch.
17769 Contributions from: Dan Horák, Kay Sievers, Lennart
17770 Poettering, Michal Schmidt
17774 * This is mostly a test release, but incorporates many
17777 * The git repository moved to:
17778 git://anongit.freedesktop.org/systemd/systemd
17779 ssh://git.freedesktop.org/git/systemd/systemd
17781 * First release with the journal
17782 https://0pointer.de/blog/projects/the-journal.html
17784 * The journal replaces both systemd-kmsg-syslogd and
17785 systemd-stdout-bridge.
17787 * New sd_pid_get_unit() API call in libsystemd-logind
17789 * Many systemadm clean-ups
17791 * Introduce remote-fs-pre.target which is ordered before all
17792 remote mounts and may be used to start services before all
17795 * Added Mageia support
17797 * Add bash completion for systemd-loginctl
17799 * Actively monitor PID file creation for daemons which exit in
17800 the parent process before having finished writing the PID
17801 file in the daemon process. Daemons which do this need to be
17802 fixed (i.e. PID file creation must have finished before the
17803 parent exits), but we now react a bit more gracefully to them.
17805 * Add colourful boot output, mimicking the well-known output
17806 of existing distributions.
17808 * New option PassCredentials= for socket units, for
17809 compatibility with a recent kernel ABI breakage.
17811 * /etc/rc.local is now hooked in via a generator binary, and
17812 thus will no longer act as synchronization point during
17815 * systemctl list-unit-files now supports --root=.
17817 * systemd-tmpfiles now understands two new commands: z, Z for
17818 relabelling files according to the SELinux database. This is
17819 useful to apply SELinux labels to specific files in /sys,
17820 among other things.
17822 * Output of SysV services is now forwarded to both the console
17823 and the journal by default, not only just the console.
17825 * New man pages for all APIs from libsystemd-login.
17827 * The build tree got reorganized and the build system is a
17828 lot more modular allowing embedded setups to specifically
17829 select the components of systemd they are interested in.
17831 * Support for Linux systems lacking the kernel VT subsystem is
17834 * configure's --with-rootdir= got renamed to
17835 --with-rootprefix= to follow the naming used by udev and
17838 * Unless specified otherwise we will now install to /usr instead
17839 of /usr/local by default.
17841 * Processes with '@' in argv[0][0] are now excluded from the
17842 final shut-down killing spree, following the logic explained
17844 https://systemd.io/ROOT_STORAGE_DAEMONS/
17846 * All processes remaining in a service cgroup when we enter
17847 the START or START_PRE states are now killed with
17848 SIGKILL. That means it is no longer possible to spawn
17849 background processes from ExecStart= lines (which was never
17850 supported anyway, and bad style).
17852 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
17853 reloading of units together.
17855 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
17856 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
17857 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
17858 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
17859 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek