1 systemd System and Service Manager
3 CHANGES WITH 253 in spe:
5 Changes in sd-boot, bootctl, and the Boot Loader Specification:
7 * systemd-boot now passes its random seed directly to the kernel's RNG
8 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
9 means the RNG gets seeded very early in boot before userspace has
12 * systemd-boot will pass a random seed when secure boot is enabled if
13 it can additionally get a random seed from EFI itself, via EFI's RNG
14 protocol or a prior seed in LINUX_EFI_RANDOM_SEED_TABLE_GUID from a
17 * The random seed stored in ESP is now refreshed whenever
18 systemd-random-seed.service is run.
20 * systemd-boot handles various seed inputs using a domain- and
21 field-separated hashing scheme.
25 Announcements of Future Feature Removals:
27 * We intend to remove cgroup v1 support from systemd release after the
28 end of 2023. If you run services that make explicit use of cgroup v1
29 features (i.e. the "legacy hierarchy" with separate hierarchies for
30 each controller), please implement compatibility with cgroup v2 (i.e.
31 the "unified hierarchy") sooner rather than later. Most of Linux
32 userspace has been ported over already.
34 * We intend to remove support for split-usr (/usr mounted separately
35 during boot) and unmerged-usr (parallel directories /bin and
36 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
37 half of 2023, in the first release that falls into that time window.
38 For more details, see:
39 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
43 * ConditionKernelVersion= checks that use the '=' or '!=' operators
44 will now do simple string comparisons (instead of version comparisons
45 á la stverscmp()). Version comparisons are still done for the
46 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
47 specified, a shell-style glob match is now done. This creates a minor
48 incompatibility compared to older systemd versions when the '*', '?',
49 '[', ']' characters are used, as these will now match as shell globs
50 instead of literally. Given that kernel version strings typically do
51 not include these characters we expect little breakage through this
54 * The service manager will now read the SELinux label used for SELinux
55 access checks from the unit file at the time it loads the file.
56 Previously, the label would be read at the moment of the access
57 check, which was problematic since at that time the unit file might
58 already have been updated or removed.
62 * systemd-measure is a new tool for calculating and signing expected
63 TPM2 PCR values for a given unified kernel image (UKI) booted via
64 sd-stub. The public key used for the signature and the signed
65 expected PCR information can be embedded inside the UKI. This
66 information can be extracted from the UKI by external tools and code
67 in the image itself and is made available to userspace in the booted
70 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
71 updated to make use of this information if available in the booted
72 kernel: when locking an encrypted volume/credential to the TPM
73 systemd-cryptenroll/systemd-creds will use the public key to bind the
74 volume/credential to any kernel that carries PCR information signed
75 by the same key pair. When unlocking such volumes/credentials
76 systemd-cryptsetup/systemd-creds will use the signature embedded in
77 the booted UKI to gain access.
79 Binding TPM-based disk encryption to public keys/signatures of PCR
80 values — instead of literal PCR values — addresses the inherent
81 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
82 disks remain accessible even if the UKI is updated, without any TPM
83 specific preparation during the OS update — as long as each UKI
84 carries the necessary PCR signature information.
86 Net effect: if you boot a properly prepared kernel, TPM-bound disk
87 encryption now defaults to be locked to kernels which carry PCR
88 signatures from the same key pair. Example: if a hypothetical distro
89 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
90 by default – bound to only FooOS kernels, and encrypted volumes bound
91 to the TPM cannot be unlocked on kernels from other sources. (But do
92 note this behaviour requires preparation/enabling in the UKI, and of
93 course users can always enroll non-TPM ways to unlock the volume.)
95 * systemd-pcrphase is a new tool that is invoked at six places during
96 system runtime, and measures additional words into TPM2 PCR 11, to
97 mark milestones of the boot process. This allows binding access to
98 specific TPM2-encrypted secrets to specific phases of the boot
99 process. (Example: LUKS2 disk encryption key only accessible in the
100 initrd, but not later.)
102 Changes in systemd itself, i.e. the manager and units
104 * The cpu controller is delegated to user manager units by default, and
105 CPUWeight= settings are applied to the top-level user slice units
106 (app.slice, background.slice, session.slice). This provides a degree
107 of resource isolation between different user services competing for
110 * Systemd can optionally do a full preset in the "first boot" condition
111 (instead of just enable-only). This behaviour is controlled by the
112 compile-time option -Dfirst-boot-full-preset. Right now it defaults
113 to 'false', but the plan is to switch it to 'true' for the subsequent
116 * Drop-ins are now allowed for transient units too.
118 * Systemd will set the taint flag 'support-ended' if it detects that
119 the OS image is past its end-of-support date. This date is declared
120 in a new /etc/os-release field SUPPORT_END= described below.
122 * Two new settings ConditionCredential= and AssertCredential= can be
123 used to skip or fail units if a certain system credential is not
126 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
128 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
129 specify the SMACK security label to use when not specified in a unit
132 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
133 specify the default timeout when waiting for device units to
136 * C.UTF-8 is used as the default locale if nothing else has been
139 * [Condition|Assert]Firmware= have been extended to support certain
140 SMBIOS fields. For example
142 ConditionFirmware=smbios-field(board_name = "Custom Board")
144 conditionalizes the unit to run only when
145 /sys/class/dmi/id/board_name contains "Custom Board" (without the
148 * ConditionFirstBoot= now correctly evaluates as true only during the
149 boot phase of the first boot. A unit executed later, after booting
150 has completed, will no longer evaluate this condition as true.
152 * Socket units will now create sockets in the SELinuxContext= of the
153 associated service unit, if any.
155 * Boot phase transitions (start initrd → exit initrd → boot complete →
156 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
157 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
158 unsealed only in the initrd.
160 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
161 also be provided to ExecStartPre= processes.
163 * Various units are now correctly ordered against
164 initrd-switch-root.target where previously a conflict without
165 ordering was configured. A stop job for those units would be queued,
166 but without the ordering it could be executed only after
167 initrd-switch-root.service, leading to units not being restarted in
168 the host system as expected.
170 * In order to fully support the IPMI watchdog driver, which has not yet
171 been ported to the new common watchdog device interface,
172 /dev/watchdog0 will be tried first and systemd will silently fallback
173 to /dev/watchdog if it is not found.
175 * New watchdog-related D-Bus properties are now published by systemd:
176 WatchdogDevice, WatchdogLastPingTimestamp,
177 WatchdogLastPingTimestampMonotonic.
179 * At shutdown, API virtual files systems (proc, sys, etc.) will be
182 * At shutdown, systemd will now log about processes blocking unmounting
185 * A new meson build option 'clock-valid-range-usec-max' was added to
186 allow disabling system time correction if RTC returns a timestamp far
189 * Propagated restart jobs will no longer be discarded while a unit is
192 * PID 1 will now import system credentials from SMBIOS Type 11 fields
193 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
194 simple, fast and generic path for supplying credentials to a VM,
195 without involving external tools such as cloud-init/ignition.
197 * The CPUWeight= setting of unit files now accepts a new special value
198 "idle", which configures "idle" level scheduling for the unit.
200 * Service processes that are activated due to a .timer or .path unit
201 triggering will now receive information about this via environment
202 variables. Note that this is information is lossy, as activation
203 might be coalesced and only one of the activating triggers will be
204 reported. This is hence more suited for debugging or tracing rather
205 than for behaviour decisions.
207 * The riscv_flush_icache(2) system call has been added to the list of
208 system calls allowed by default when SystemCallFilter= is used.
210 * The selinux context derived from the target executable, instead of
211 'init_t' used for the manager itself, is now used when creating
212 listening sockets for units that specify SELinuxContextFromNet=yes.
214 Changes in sd-boot, bootctl, and the Boot Loader Specification:
216 * The Boot Loader Specification has been cleaned up and clarified.
217 Various corner cases in version string comparisons have been fixed
218 (e.g. comparisons for empty strings). Boot counting is now part of
219 the main specification.
221 * New PCRs measurements are performed during boot: PCR 11 for the the
222 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
223 took place this is now reported to userspace via the new
224 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
226 * As before, systemd-stub will measure kernel parameters and system
227 credentials into PCR 12. It will now report this fact via the
228 StubPcrKernelParameters EFI variable to userspace.
230 * The UEFI monotonic boot counter is now included in the updated random
231 seed file maintained by sd-boot, providing some additional entropy.
233 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
234 of arranging the image manually and jumping to the kernel entry
235 point. sd-stub also installs a temporary UEFI SecurityOverride to
236 allow the (unsigned) nested image to be booted. This is safe because
237 the outer (signed) stub+kernel binary must have been verified before
238 the stub was executed.
240 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
241 is now supported by sd-boot.
243 * bootctl gained a bunch of new options: --all-architectures to install
244 binaries for all supported EFI architectures, --root= and --image=
245 options to operate on a directory or disk image, and
246 --install-source= to specify the source for binaries to install,
247 --efi-boot-option-description= to control the name of the boot entry.
249 * The sd-boot stub exports a StubFeatures flag, which is used by
250 bootctl to show features supported by the stub that was used to boot.
252 * The PE section offsets that are used by tools that assemble unified
253 kernel images have historically been hard-coded. This may lead to
254 overlapping PE sections which may break on boot. The UKI will now try
255 to detect and warn about this.
257 Any tools that assemble UKIs must update to calculate these offsets
258 dynamically. Future sd-stub versions may use offsets that will not
259 work with the currently used set of hard-coded offsets!
261 * sd-stub now accepts (and passes to the initrd and then to the full
262 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
263 signatures of expected PCR values, to allow sealing secrets via the
264 TPM2 against pre-calculated PCR measurements.
266 Changes in the hardware database:
268 * 'systemd-hwdb query' now supports the --root= option.
270 Changes in systemctl:
272 * systemctl now supports --state= and --type= options for the 'show'
275 * systemctl gained a new verb 'list-automounts' to list automount
278 * systemctl gained support for a new --image= switch to be able to
279 operate on the specified disk image (similar to the existing --root=
280 which operates relative to some directory).
282 Changes in systemd-networkd:
284 * networkd can set Linux NetLabel labels for integration with the
285 network control in security modules via a new NetLabel= option.
287 * The RapidCommit= is (re-)introduced to enable faster configuration
288 via DHCPv6 (RFC 3315).
290 * networkd gained a new option TCPCongestionControlAlgorithm= that
291 allows setting a per-route TCP algorithm.
293 * networkd gained a new option KeepFileDescriptor= to allow keeping a
294 reference (file descriptor) open on TUN/TAP interfaces, which is
295 useful to avoid link flaps while the underlying service providing the
296 interface is being serviced.
298 * RouteTable= now also accepts route table names.
300 Changes in systemd-nspawn:
302 * The --bind= and --overlay= options now support relative paths.
304 * The --bind= option now supports a 'rootidmap' value, which will
305 use id-mapped mounts to map the root user inside the container to the
306 owner of the mounted directory on the host.
308 Changes in systemd-resolved:
310 * systemd-resolved now persists DNSOverTLS in its state file too. This
311 fixes a problem when used in combination with NetworkManager, which
312 sends the setting only once, causing it to be lost if resolved was
313 restarted at any point.
315 * systemd-resolved now exposes a varlink socket at
316 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
317 root. Processed DNS requests in a JSON format will be published to
318 any clients connected to this socket.
320 resolvectl gained a 'monitor' verb to make use of this.
322 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
323 instead of returning SERVFAIL, as per RFC:
324 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
326 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
329 Changes in libsystemd and other libraries:
331 * libsystemd now exports sd_bus_error_setfv() (a convenience function
332 for setting bus errors), sd_id128_string_equal (a convenience
333 function for 128bit ID string comparisons), and
334 sd_bus_message_read_strv_extend() (a function to incrementally read
337 * libsystemd now exports sd_device_get_child_first()/_next() as a
338 high-level interface for enumerating child devices. It also supports
339 sd_device_new_child() for opening a child device given a device
342 * libsystemd now exports sd_device_monitor_set()/get_description()
343 which allow setting a custom description that will be used in log
344 messages by sd_device_monitor*.
346 * Private shared libraries (libsystemd-shared-nnn.so,
347 libsystemd-core-nnn.so) are now installed into arch-specific
348 directories to allow multi-arch installs.
350 * A new sd-gpt.h header is now published, listing GUIDs from the
351 Discoverable Partitions specification. For more details see:
352 https://systemd.io/DISCOVERABLE_PARTITIONS/
354 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
355 database given an explicit path to the file.
357 * The signal number argument to sd_event_add_signal() now can now be
358 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
359 be automatically invoked to block the specified signal. This is
360 useful to simplify invocations as the caller doesn't have to do this
363 * A new convenience call sd_event_set_signal_exit() has been added to
364 sd-event to set up signal handling so that the event loop
365 automatically terminates cleanly on SIGTERM/SIGINT.
367 Changes in other components:
369 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
370 can now be provided via the credential mechanism.
372 * systemd-analyze gained a new verb 'compare-versions' that implements
373 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
374 'dpkg --compare-versions').
376 * 'systemd-analyze dump' is extended to accept glob patterns for unit
377 names to limit the output to matching units.
379 * tmpfiles.d/ lines can read file contents to write from a credential.
380 The new modifier char '^' is used to specify that the argument is a
381 credential name. This mechanism is used to automatically populate
382 /etc/motd, /etc/issue, and /etc/hosts from credentials.
384 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
385 an inode if the specification is prefixed with ':' and the inode
388 * Default tmpfiles.d/ configuration now carries a line to automatically
389 use an 'ssh.authorized_keys.root' credential if provided to set up
390 the SSH authorized_keys file for the root user.
392 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
395 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
396 in base64. This is useful to write arbitrary binary data into files.
398 * The pkgconfig and rpm macros files now export the directory for user
399 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
401 * Detection of Apple Virtualization and detection of Parallels and
402 KubeVirt virtualization on non-x86 archs have been added.
404 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
405 user when their system will become unsupported.
407 * When performing suspend-then-hibernate, the system will estimate the
408 discharge rate and use that to set the delay until hibernation and
409 hibernate immediately instead of suspending when running from a
410 battery and the capacity is below 5%.
412 * systemd-sysctl gained a --strict option to fail when a sysctl
413 setting is unknown to the kernel.
415 * machinectl supports --force for the 'copy-to' and 'copy-from'
418 * coredumpctl gained the --root and --image options to look for journal
419 files under the specified root directory, image, or block device.
421 * 'journalctl -o' and similar commands now implement a new output mode
422 "short-delta". It is similar to "short-monotonic", but also shows the
423 time delta between subsequent messages.
425 * journalctl now respects the --quiet flag when verifying consistency
428 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
429 will indicate whether a message was logged in the 'initrd' phase or
430 in the 'system' phase of the boot process.
432 * Journal files gained a new compatibility flag
433 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
434 to the storage format that allow reducing size on disk. As with other
435 compatibility flags, older journalctl versions will not be able to
436 read journal files using this new format. The environment variable
437 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
438 disable this functionality. It is enabled by default.
440 * systemd-run's --working-directory= switch now works when used in
441 combination with --scope.
443 * portablectl gained a --force flag to skip certain sanity checks. This
444 is implemented using new flags accepted by systemd-portabled for the
445 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
446 flag now means that the attach/detach checks whether the units are
447 already present and running will be skipped. Similarly,
448 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
449 image name matches the name declared inside of the image will be
450 skipped. Callers must be sure to do those checks themselves if
453 * systemd-portabled will now use the original filename to check
454 extension-release.NAME for correctness, in case it is passed a
457 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
460 * sysext's extension-release files now support '_any' as a special
461 value for the ID= field, to allow distribution-independent extensions
462 (e.g.: fully statically compiled binaries, scripts). It also gained
463 support for a new ARCHITECTURE= field that may be used to explicitly
464 restrict an image to hosts of a specific architecture.
466 * systemd-repart now supports creating squashfs partitions. This
467 requires mksquashfs from squashfs-tools.
469 * systemd-repart gained a --split flag to also generate split
470 artifacts, i.e. a separate file for each partition. This is useful in
471 conjunction with systemd-sysupdate or other tools, or to generate
472 split dm-verity artifacts.
474 * systemd-repart is now able to generate dm-verity partitions, including
477 * systemd-repart can now set a partition UUID to zero, allowing it to
478 be filled in later, such as when using verity partitions.
480 * systemd-repart now supports drop-ins for its configuration files.
482 * Package metadata logged by systemd-coredump in the system journal is
485 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
487 * systemd-oomd now automatically links against libatomic, if available.
489 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
492 * scope units now also provide oom-kill status.
494 * systemd-pstore will now try to load only the efi_pstore kernel module
495 before running, ensuring that pstore can be used.
497 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
498 session after a preconfigure timeout.
500 * systemd-homed will now wait up to 30 seconds for workers to terminate,
501 rather than indefinitely.
503 * homectl gained a new '--luks-sector-size=' flag that allows users to
504 select the preferred LUKS sector size. Must be a power of 2 between 512
505 and 4096. systemd-userdbd records gained a corresponding field.
507 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
508 variable when generating the 'sp_lstchg' field, to ensure an image
509 build can be reproducible.
511 * 'udevadm wait' will now listen to kernel uevents too when called with
514 * When naming network devices udev will now consult the Devicetree
515 "alias" fields for the device.
517 * systemd-udev will now create infiniband/by-path and
518 infiniband/by-ibdev links for Infiniband verbs devices.
520 * systemd-udev-trigger.service will now also prioritize input devices.
522 * ConditionACPower= and systemd-ac-power will now assume the system is
523 running on AC power if no battery can be found.
525 * All features and tools using the TPM2 will now communicate with it
526 using a bind key. Beforehand, the tpm2 support used encrypted sessions
527 by creating a primary key that was used to encrypt traffic. This
528 creates a problem as the key created for encrypting the traffic could
529 be faked by an active interposer on the bus. In cases when a pin is
530 used, a bind key will be used. The pin is used as the auth value for
531 the seal key, aka the disk encryption key, and that auth value will be
532 used in the session establishment. An attacker would need the pin
533 value to create the secure session and thus an active interposer
534 without the pin cannot interpose on TPM2 traffic.
536 * systemd-growfs no longer requires udev to run.
538 * systemd-backlight now will better support systems with multiple
541 * systemd-cryptsetup's keyfile-timeout= option now also works when a
542 device is used as a keyfile.
544 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
545 unlocking key from a key file (instead of prompting the user). Note
546 that this is the key for unlocking the volume in order to be able to
547 enroll a new key, but it is not the key that is enrolled.
549 * systemd-dissect gained a new --umount switch that will safely and
550 synchronously unmount all partitions of an image previously mounted
551 with 'systemd-dissect --mount'.
553 * When using gcrypt, all systemd tools and services will now configure
554 it to prefer the OS random number generator if present.
556 * All example code shipped with documentation has been relicensed from CC0
559 * Unit tests will no longer fail when running on a system without
562 Experimental features:
564 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
567 * sd-boot can automatically enroll SecureBoot keys from files found on
568 the ESP. This enrollment can be either automatic ('force' mode) or
569 controlled by the user ('manual' mode). It is sufficient to place the
570 SecureBoot keys in the right place in the ESP and they will be picked
571 up by sd-boot and shown in the boot menu.
573 * The mkosi config in systemd gained support for automatically
574 compiling a kernel with the configuration appropriate for testing
575 systemd. This may be useful when developing or testing systemd in
576 tandem with the kernel.
578 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
579 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
580 Alexander Graf, Alexander Shopov, Alexander Wilson,
581 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
582 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
583 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
584 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
585 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
586 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
587 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
588 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
589 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
590 David Jaša, David Rheinsberg, David Seifert, David Tardon,
591 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
592 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
593 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
594 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
595 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
596 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
597 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
598 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
599 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
600 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
601 JeroenHD, jiangchuangang, João Loureiro,
602 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
603 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
604 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
605 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
606 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
607 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
608 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
609 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
610 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
611 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
612 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
613 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
614 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
615 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
616 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
617 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
618 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
619 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
620 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
621 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
622 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
623 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
624 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
625 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
627 – The Great Beyond, 2022-10-31 👻
631 Backwards-incompatible changes:
633 * The minimum kernel version required has been bumped from 3.13 to 4.15,
634 and CLOCK_BOOTTIME is now assumed to always exist.
636 * C11 with GNU extensions (aka "gnu11") is now used to build our
637 components. Public API headers are still restricted to ISO C89.
639 * In v250, a systemd-networkd feature that automatically configures
640 routes to addresses specified in AllowedIPs= was added and enabled by
641 default. However, this causes network connectivity issues in many
642 existing setups. Hence, it has been disabled by default since
643 systemd-stable 250.3. The feature can still be used by explicitly
644 configuring RouteTable= setting in .netdev files.
646 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
647 when a Condition*= check does not succeed, restoring the JobRemoved
648 signal to the behaviour it had before v250.
650 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
651 GetImageMetadataWithExtensions() have been fixed to provide an extra
652 return parameter, containing the actual extension release metadata.
653 The current implementation was judged to be broken and unusable, and
654 thus the usual procedure of adding a new set of methods was skipped,
655 and backward compatibility broken instead on the assumption that
656 nobody can be affected given the current state of this interface.
658 * All kernels supported by systemd mix bytes returned by RDRAND (or
659 similar) into the entropy pool at early boot. This means that on
660 those systems, even if /dev/urandom is not yet initialized, it still
661 returns bytes that are of at least RDRAND quality. For that reason,
662 we no longer have reason to invoke RDRAND from systemd itself, which
663 has historically been a source of bugs. Furthermore, kernels ≥5.6
664 provide the getrandom(GRND_INSECURE) interface for returning random
665 bytes before the entropy pool is initialized without warning into
666 kmsg, which is what we attempt to use if available. systemd's direct
667 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
668 running an older kernel may experience kmsg warnings that were not
669 seen with 250. For newer kernels, non-x86 systems, or older x86
670 systems, there should be no visible changes.
672 * sd-boot will now measure the kernel command line into TPM PCR 12
673 rather than PCR 8. This improves usefulness of the measurements on
674 systems where sd-boot is chainloaded from Grub. Grub measures all
675 commands its executes into PCR 8, which makes it very hard to use
676 reasonably, hence separate ourselves from that and use PCR 12
677 instead, which is what certain Ubuntu editions already do. To retain
678 compatibility with systems running older systemd systems a new meson
679 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
680 If enabled, the measurement is done twice: into the new-style PCR 12
681 *and* the old-style PCR 8. It's strongly advised to migrate all users
682 to PCR 12 for this purpose in the long run, as we intend to remove
683 this compatibility feature in two years' time.
685 * busctl capture now writes output in the newer pcapng format instead
688 * A udev rule that imported hwdb matches for USB devices with lowercase
689 hexadecimal vendor/product ID digits was added in systemd 250. This
690 has been reverted, since uppercase hexadecimal digits are supposed to
691 be used, and we already had a rule with the appropriate match.
693 Users might need to adjust their local hwdb entries.
695 * arch_prctl(2) has been moved to the @default set in the syscall filters
696 (as exposed via the SystemCallFilter= setting in service unit files).
697 It is apparently used by the linker now.
699 * The tmpfiles entries that create the /run/systemd/netif directory and
700 its subdirectories were moved from tmpfiles.d/systemd.conf to
701 tmpfiles.d/systemd-network.conf.
703 Users might need to adjust their files that override tmpfiles.d/systemd.conf
704 to account for this change.
706 * The requirement for Portable Services images to contain a well-formed
707 os-release file (i.e.: contain at least an ID field) is now enforced.
708 This applies to base images and extensions, and also to systemd-sysext.
710 Changes in the Boot Loader Specification, kernel-install and sd-boot:
712 * kernel-install's and bootctl's Boot Loader Specification Type #1
713 entry generation logic has been reworked. The user may now pick
714 explicitly by which "token" string to name the installation's boot
715 entries, via the new /etc/kernel/entry-token file or the new
716 --entry-token= switch to bootctl. By default — as before — the
717 entries are named after the local machine ID. However, in "golden
718 image" environments, where the machine ID shall be initialized on
719 first boot (as opposed to at installation time before first boot) the
720 machine ID will not be available at build time. In this case the
721 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
722 file) may be used to override the "token" for the entries, for
723 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
724 make the OS images independent of any machine ID, and ensure that the
725 images will not carry any identifiable information before first boot,
726 but on the other hand means that multiple parallel installations of
727 the very same image on the same disk cannot be supported.
729 Summary: if you are building golden images that shall acquire
730 identity information exclusively on first boot, make sure to both
731 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
732 value of the IMAGE_ID= or ID= field of /etc/os-release or another
733 suitable identifier before deploying the image.
735 * The Boot Loader Specification has been extended with
736 /loader/entries.srel file located in the EFI System Partition (ESP)
737 that disambiguates the format of the entries in the /loader/entries/
738 directory (in order to discern them from incompatible uses of this
739 directory by other projects). For entries that follow the
740 Specification, the string "type1" is stored in this file.
742 bootctl will now write this file automatically when installing the
743 systemd-boot boot loader.
745 * kernel-install supports a new initrd_generator= setting in
746 /etc/kernel/install.conf, that is exported as
747 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
748 allows choosing different initrd generators.
750 * kernel-install will now create a "staging area" (an initially-empty
751 directory to gather files for a Boot Loader Specification Type #1
752 entry). The path to this directory is exported as
753 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
754 drop files there instead of writing them directly to the final
755 location. kernel-install will move them when all files have been
756 prepared successfully.
758 * New option sort-key= has been added to the Boot Loader Specification
759 to override the sorting order of the entries in the boot menu. It is
760 read by sd-boot and bootctl, and will be written by kernel-install,
761 with the default value of IMAGE_ID= or ID= fields from
762 os-release. Together, this means that on multiboot installations,
763 entries should be grouped and sorted in a predictable way.
765 * The sort order of boot entries has been updated: entries which have
766 the new field sort-key= are sorted by it first, and all entries
767 without it are ordered later. After that, entries are sorted by
768 version so that newest entries are towards the beginning of the list.
770 * The kernel-install tool gained a new 'inspect' verb which shows the
771 paths and other settings used.
773 * sd-boot can now optionally beep when the menu is shown and menu
774 entries are selected, which can be useful on machines without a
775 working display. (Controllable via a loader.conf setting.)
777 * The --make-machine-id-directory= switch to bootctl has been replaced
778 by --make-entry-directory=, given that the entry directory is not
779 necessarily named after the machine ID, but after some other suitable
780 ID as selected via --entry-token= described above. The old name of
781 the option is still understood to maximize compatibility.
783 * 'bootctl list' gained support for a new --json= switch to output boot
784 menu entries in JSON format.
786 * 'bootctl is-installed' now supports the --graceful, and various verbs
787 omit output with the new option --quiet.
789 Changes in systemd-homed:
791 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
792 of activated home directories it manages (if the kernel and selected
793 file systems support it). So far it mapped three UID ranges: the
794 range from 0…60000, the user's own UID, and the range 60514…65534,
795 leaving everything else unmapped (in other words, the 16bit UID range
796 is mapped almost fully, with the exception of the UID subrange used
797 for systemd-homed users, with one exception: the user's own UID).
798 Unmapped UIDs may not be used for file ownership in the home
799 directory — any chown() attempts with them will fail. With this
800 release a fourth range is added to these mappings:
801 524288…1879048191. This range is the UID range intended for container
804 https://systemd.io/UIDS-GIDS
806 This range may be used for container managers that place container OS
807 trees in the home directory (which is a questionable approach, for
808 quota, permission, SUID handling and network file system
809 compatibility reasons, but nonetheless apparently commonplace). Note
810 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
811 UID assignments from the range are not managed or mapped by
812 `systemd-homed`, and must be managed with other mechanisms, in the
813 context of the local system.
815 Typically, a better approach to user namespacing in relevant
816 container managers would be to leave container OS trees on disk at
817 UID offset 0, but then map them to a dynamically allocated runtime
818 UID range via another UID mount map at container invocation
819 time. That way user namespace UID ranges become strictly a runtime
820 concept, and do not leak into persistent file systems, persistent
821 user databases or persistent configuration, thus greatly simplifying
822 handling, and improving compatibility with home directories intended
823 to be portable like the ones managed by systemd-homed.
825 Changes in shared libraries:
827 * A new libsystemd-core-<version>.so private shared library is
828 installed under /usr/lib/systemd/system, mirroring the existing
829 libsystemd-shared-<version>.so library. This allows the total
830 installation size to be reduced by binary code reuse.
832 * The <version> tag used in the name of libsystemd-shared.so and
833 libsystemd-core.so can be configured via the meson option
834 'shared-lib-tag'. Distributions may build subsequent versions of the
835 systemd package with unique tags (e.g. the full package version),
836 thus allowing multiple installations of those shared libraries to be
837 available at the same time. This is intended to fix an issue where
838 programs that link to those libraries would fail to execute because
839 they were installed earlier or later than the appropriate version of
842 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
843 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
844 format instead of as a simple series of hex characters.
846 * The sd-device API gained two new calls sd_device_new_from_devname()
847 and sd_device_new_from_path() which permit allocating an sd_device
848 object from a device node name or file system path.
850 * sd-device also gained a new call sd_device_open() which will open the
851 device node associated with a device for which an sd_device object
852 has been allocated. The call is supposed to address races around
853 device nodes being removed/recycled due to hotplug events, or media
854 change events: the call checks internally whether the major/minor of
855 the device node and the "diskseq" (in case of block devices) match
856 with the metadata loaded in the sd_device object, thus ensuring that
857 the device once opened really matches the provided sd_device object.
859 Changes in PID1, systemctl, and systemd-oomd:
861 * A new set of service monitor environment variables will be passed to
862 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
863 handler unit as OnFailure=/OnSuccess=. The variables are:
864 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
865 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
866 handler needs to watch multiple units, use a templated handler.
868 * A new ExtensionDirectories= setting in service unit files allows
869 system extensions to be loaded from a directory. (It is similar to
870 ExtensionImages=, but takes paths to directories, instead of
873 'portablectl attach --extension=' now also accepts directory paths.
875 * The user.delegate and user.invocation_id extended attributes on
876 cgroups are used in addition to trusted.delegate and
877 trusted.invocation_id. The latter pair requires privileges to set,
878 but the former doesn't and can be also set by the unprivileged user
881 (Only supported on kernels ≥5.6.)
883 * Units that were killed by systemd-oomd will now have a service result
884 of 'oom-kill'. The number of times a service was killed is tallied
885 in the 'user.oomd_ooms' extended attribute.
887 The OOMPolicy= unit file setting is now also honoured by
890 * In unit files the new %y/%Y specifiers can be used to refer to
891 normalized unit file path, which is particularly useful for symlinked
894 The new %q specifier resolves to the pretty hostname
895 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
897 The new %d specifier resolves to the credentials directory of a
898 service (same as $CREDENTIALS_DIRECTORY).
900 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
901 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
902 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
903 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
904 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
905 MountFlags= service settings now also work in unprivileged user
906 services, i.e. those run by the user's --user service manager, as long
907 as user namespaces are enabled on the system.
909 * Services with Restart=always and a failing ExecCondition= will no
910 longer be restarted, to bring ExecCondition= behaviour in line with
911 Condition*= settings.
913 * LoadCredential= now accepts a directory as the argument; all files
914 from the directory will be loaded as credentials.
916 * A new D-Bus property ControlGroupId is now exposed on service units,
917 that encapsulates the service's numeric cgroup ID that newer kernels
918 assign to each cgroup.
920 * PID 1 gained support for configuring the "pre-timeout" of watchdog
921 devices and the associated governor, via the new
922 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
923 options in /etc/systemd/system.conf.
925 * systemctl's --timestamp= option gained a new choice "unix", to show
926 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
928 * A new "taint" flag named "old-kernel" is introduced which is set when
929 the kernel systemd runs on is older then the current baseline version
930 (see above). The flag is shown in "systemctl status" output.
932 * Two additional taint flags "short-uid-range" and "short-gid-range"
933 have been added as well, which are set when systemd notices it is run
934 within a userns namespace that does not define the full 0…65535 UID
937 * A new "unmerged-usr" taint flag has been added that is set whenever
938 running on systems where /bin/ + /sbin/ are *not* symlinks to their
939 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
942 * Generators invoked by PID 1 will now have a couple of useful
943 environment variables set describing the execution context a
944 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
945 system service manager, or from the per-user service
946 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
947 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
948 systemd considers the current boot to be a "first"
949 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
950 detected and which type of hypervisor/container
951 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
954 * PID 1 will now automatically pick up system credentials from qemu's
955 fw_cfg interface, thus allowing passing arbitrary data into VM
956 systems similar to how this is already supported for passing them
957 into `systemd-nspawn` containers. Credentials may now also be passed
958 in via the new kernel command line option `systemd.set_credential=`
959 (note that kernel command line options are world-readable during
960 runtime, and only useful for credentials that require no
961 confidentiality). The credentials that can be passed to unified
962 kernels that use the `systemd-stub` UEFI stub are now similarly
963 picked up automatically. Automatic importing of system credentials
964 this way can be turned off via the new
965 `systemd.import_credentials=no` kernel command line option.
967 * LoadCredential= will now automatically look for credentials in the
968 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
969 the argument is not an absolute path. Similarly,
970 LoadCredentialEncrypted= will check the same directories plus
971 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
972 /usr/lib/credstore.encrypted/. The idea is to use those directories
973 as the system-wide location for credentials that services should pick
976 * System and service credentials are described in great detail in a new
979 https://systemd.io/CREDENTIALS
981 Changes in systemd-journald:
983 * The journal JSON export format has been added to listed of stable
984 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
986 * journalctl --list-boots now supports JSON output and the --reverse option.
988 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
989 updated, BUILDING_IMAGES is new:
991 https://systemd.io/JOURNAL_EXPORT_FORMATS
992 https://systemd.io/BUILDING_IMAGES
996 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
997 calculators, etc.), the other AV production devices (DJ tables,
998 keypads, etc.) that should accessible to the seat owner user by
1001 * udevadm trigger gained a new --prioritized-subsystem= option to
1002 process certain subsystems (and all their parent devices) earlier.
1004 systemd-udev-trigger.service now uses this new option to trigger
1005 block and TPM devices first, hopefully making the boot a bit faster.
1007 * udevadm trigger now implements --type=all, --initialized-match,
1008 --initialized-nomatch to trigger both subsystems and devices, only
1009 already-initialized devices, and only devices which haven't been
1010 initialized yet, respectively.
1012 * udevadm gained a new "wait" command for safely waiting for a specific
1013 device to show up in the udev device database. This is useful in
1014 scripts that asynchronously allocate a block device (e.g. through
1015 repartitioning, or allocating a loopback device or similar) and need
1016 to synchronize on the creation to complete.
1018 * udevadm gained a new "lock" command for locking one or more block
1019 devices while formatting it or writing a partition table to it. It is
1020 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
1021 usable in scripts dealing with block devices.
1023 * udevadm info will show a couple of additional device fields in its
1024 output, and will not apply a limited set of coloring to line types.
1026 * udevadm info --tree will now show a tree of objects (i.e. devices and
1027 suchlike) in the /sys/ hierarchy.
1029 * Block devices will now get a new set of device symlinks in
1030 /dev/disk/by-diskseq/<nr>, which may be used to reference block
1031 device nodes via the kernel's "diskseq" value. Note that this does
1032 not guarantee that opening a device by a symlink like this will
1033 guarantee that the opened device actually matches the specified
1034 diskseq value. To be safe against races, the actual diskseq value of
1035 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
1036 the one in the symlink path.
1038 * .link files gained support for setting MDI/MID-X on a link.
1040 * .link files gained support for [Match] Firmware= setting to match on
1041 the device firmware description string. By mistake, it was previously
1042 only supported in .network files.
1044 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
1045 and [SR-IOV] section to configure SR-IOV virtual functions.
1047 Changes in systemd-networkd:
1049 * The default scope for unicast routes configured through [Route]
1050 section is changed to "link", to make the behavior consistent with
1051 "ip route" command. The manual configuration of [Route] Scope= is
1054 * A new unit systemd-networkd-wait-online@<interface>.service has been
1055 added that can be used to wait for a specific network interface to be
1058 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
1059 that configures the eponymous kernel attribute on the bridge.
1061 * .netdev files now can be used to create virtual WLAN devices, and
1062 configure various settings on them, via the [WLAN] section.
1064 * .link/.network files gained support for [Match] Kind= setting to match
1065 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
1067 This value is also shown by 'networkctl status'.
1069 * The Local= setting in .netdev files for various virtual network
1070 devices gained support for specifying, in addition to the network
1071 address, the name of a local interface which must have the specified
1074 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
1075 files, to configure tunnels in external mode (a.k.a. collect metadata
1078 * [Network] L2TP= setting was removed. Please use interface specifier in
1079 Local= setting in .netdev files of corresponding L2TP interface.
1081 * New [DHCPServer] BootServerName=, BootServerAddress=, and
1082 BootFilename= settings can be used to configure the server address,
1083 server name, and file name sent in the DHCP packet (e.g. to configure
1086 Changes in systemd-resolved:
1088 * systemd-resolved is started earlier (in sysinit.target), so it
1089 available earlier and will also be started in the initrd if installed
1092 Changes in disk encryption:
1094 * systemd-cryptenroll can now control whether to require the user to
1095 enter a PIN when using TPM-based unlocking of a volume via the new
1096 --tpm2-with-pin= option.
1098 Option tpm2-pin= can be used in /etc/crypttab.
1100 * When unlocking devices via TPM, TPM2 parameter encryption is now
1101 used, to ensure that communication between CPU and discrete TPM chips
1102 cannot be eavesdropped to acquire disk encryption keys.
1104 * A new switch --fido2-credential-algorithm= has been added to
1105 systemd-cryptenroll allowing selection of the credential algorithm to
1106 use when binding encryption to FIDO2 tokens.
1108 Changes in systemd-hostnamed:
1110 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
1111 to override the values gleaned from the hwdb.
1113 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
1114 /sys/class/dmi/id) to override the chassis that is reported by
1117 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
1118 for reading the hardware serial number, as reportd by DMI. It also
1119 exposes a new method D-Bus property FirmwareVersion that encode the
1120 firmware version of the system.
1122 Changes in other components:
1124 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
1125 handling with the values that were configured during systemd build
1126 (if /etc/locale.conf has not been created through some other
1127 mechanism). This means that /etc/locale.conf should always have
1128 reasonable contents and we avoid a potential mismatch in defaults.
1130 * The userdbctl tool will now show UID range information as part of the
1131 list of known users.
1133 * A new build-time configuration setting default-user-shell= can be
1134 used to set the default shell for user records and nspawn shell
1135 invocations (instead of of the default /bin/bash).
1137 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
1138 information dynamically at runtime via IPC.
1140 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
1141 whether a functioning TPM2 infrastructure is available, i.e. if
1142 firmware, kernel driver and systemd all have TPM2 support enabled and
1145 * The systemd-creds tool gained support for generating encrypted
1146 credentials that are using an empty encryption key. While this
1147 provides no integrity nor confidentiality it's useful to implement
1148 codeflows that work the same on TPM-ful and TPM2-less systems. The
1149 service manager will only accept credentials "encrypted" that way if
1150 a TPM2 device cannot be detected, to ensure that credentials
1151 "encrypted" like that cannot be used to trick TPM2 systems.
1153 * When deciding whether to colorize output, all systemd programs now
1154 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
1157 * Meson's new install_tag feature is now in use for several components,
1158 allowing to build and install select binaries only: pam, nss, devel
1159 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
1160 $ meson build systemd-boot
1161 $ meson install --tags systemd-boot --no-rebuild
1162 https://mesonbuild.com/Installing.html#installation-tags
1164 * A new build configuration option has been added, to allow selecting the
1165 default compression algorithm used by systemd-journald and systemd-coredump.
1166 This allows to build-in support for decompressing all supported formats,
1167 but choose a specific one for compression. E.g.:
1168 $ meson -Ddefault-compression=xz
1170 Experimental features:
1172 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
1173 loader.conf that implements booting Microsoft Windows from the
1174 sd-boot in a way that first reboots the system, to reset the TPM
1175 PCRs. This improves compatibility with BitLocker's TPM use, as the
1176 PCRs will only record the Windows boot process, and not sd-boot
1177 itself, thus retaining the PCR measurements not involving sd-boot.
1178 Note that this feature is experimental for now, and is likely going
1179 to be generalized and renamed in a future release, without retaining
1180 compatibility with the current implementation.
1182 * A new systemd-sysupdate component has been added that automatically
1183 discovers, downloads, and installs A/B-style updates for the host
1184 installation itself, or container images, portable service images,
1185 and other assets. See the new systemd-sysupdate man page for updates.
1187 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
1188 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
1189 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
1190 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
1191 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
1192 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
1193 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
1194 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
1195 David, David Bond, Davide Cavalca, David Tardon, davijosw,
1196 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
1197 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
1198 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
1199 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
1200 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
1201 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
1202 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
1203 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
1204 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
1205 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
1206 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
1207 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
1208 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
1209 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
1210 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
1211 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
1212 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
1213 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
1214 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
1215 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
1216 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
1217 Simon Ellmann, Sonali Srivastava, Stefan Seering,
1218 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
1219 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
1220 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
1221 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
1222 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
1223 Zbigniew Jędrzejewski-Szmek, наб
1225 — Edinburgh, 2022-05-21
1229 * Support for encrypted and authenticated credentials has been added.
1230 This extends the credential logic introduced with v247 to support
1231 non-interactive symmetric encryption and authentication, based on a
1232 key that is stored on the /var/ file system or in the TPM2 chip (if
1233 available), or the combination of both (by default if a TPM2 chip
1234 exists the combination is used, otherwise the /var/ key only). The
1235 credentials are automatically decrypted at the moment a service is
1236 started, and are made accessible to the service itself in unencrypted
1237 form. A new tool 'systemd-creds' encrypts credentials for this
1238 purpose, and two new service file settings LoadCredentialEncrypted=
1239 and SetCredentialEncrypted= configure such credentials.
1241 This feature is useful to store sensitive material such as SSL
1242 certificates, passwords and similar securely at rest and only decrypt
1243 them when needed, and in a way that is tied to the local OS
1244 installation or hardware.
1246 * systemd-gpt-auto-generator can now automatically set up discoverable
1247 LUKS2 encrypted swap partitions.
1249 * The GPT Discoverable Partitions Specification has been substantially
1250 extended with support for root and /usr/ partitions for the majority
1251 of architectures systemd supports. This includes platforms that do
1252 not natively support UEFI, because even though GPT is specified under
1253 UEFI umbrella, it is useful on other systems too. Specifically,
1254 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
1255 Portable Services use the concept without requiring UEFI.
1257 * The GPT Discoverable Partitions Specifications has been extended with
1258 a new set of partitions that may carry PKCS#7 signatures for Verity
1259 partitions, encoded in a simple JSON format. This implements a simple
1260 mechanism for building disk images that are fully authenticated and
1261 can be tested against a set of cryptographic certificates. This is
1262 now implemented for the various systemd tools that can operate with
1263 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
1264 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
1265 The PKCS#7 signatures are passed to the kernel (where they are
1266 checked against certificates from the kernel keyring), or can be
1267 verified against certificates provided in userspace (via a simple
1268 drop-in file mechanism).
1270 * systemd-dissect's inspection logic will now report for which uses a
1271 disk image is intended. Specifically, it will display whether an
1272 image is suitable for booting on UEFI or in a container (using
1273 systemd-nspawn's --image= switch), whether it can be used as portable
1274 service, or attached as system extension.
1276 * The system-extension.d/ drop-in files now support a new field
1277 SYSEXT_SCOPE= that may encode which purpose a system extension image
1278 is for: one of "initrd", "system" or "portable". This is useful to
1279 make images more self-descriptive, and to ensure system extensions
1280 cannot be attached in the wrong contexts.
1282 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
1283 be used in portable service images to indicate which unit prefixes
1286 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
1287 now is able to decode images for non-native architectures as well.
1288 This allows systemd-nspawn to boot images of non-native architectures
1289 if the corresponding user mode emulator is installed and
1290 systemd-binfmtd is running.
1292 * systemd-logind gained new settings HandlePowerKeyLongPress=,
1293 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
1294 HandleHibernateKeyLongPress= which may be used to configure actions
1295 when the relevant keys are pressed for more than 5s. This is useful
1296 on devices that only have hardware for a subset of these keys. By
1297 default, if the reboot key is pressed long the poweroff operation is
1298 now triggered, and when the suspend key is pressed long the hibernate
1299 operation is triggered. Long pressing the other two keys currently
1300 does not trigger any operation by default.
1302 * When showing unit status updates on the console during boot and
1303 shutdown, and a service is slow to start so that the cylon animation
1304 is shown, the most recent sd_notify() STATUS= text is now shown as
1305 well. Services may use this to make the boot/shutdown output easier
1306 to understand, and to indicate what precisely a service that is slow
1307 to start or stop is waiting for. In particular, the per-user service
1308 manager instance now reports what it is doing and which service it is
1309 waiting for this way to the system service manager.
1311 * The service manager will now re-execute on reception of the
1312 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
1313 only when running as PID 1. There was no signal to request this when
1314 running as per-user service manager, i.e. as any other PID than 1.
1315 SIGRTMIN+25 works for both system and user managers.
1317 * The hardware watchdog logic in PID 1 gained support for operating
1318 with the default timeout configured in the hardware, instead of
1319 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
1320 request this behavior.
1322 * A new kernel command line option systemd.watchdog_sec= is now
1323 understood which may be used to override the hardware watchdog
1324 time-out for the boot.
1326 * A new setting DefaultOOMScoreAdjust= is now supported in
1327 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
1328 to set the default process OOM score adjustment value for processes
1329 started by the service manager. For per-user service managers this
1330 now defaults to 100, but for per-system service managers is left as
1331 is. This means that by default now services forked off the user
1332 service manager are more likely to be killed by the OOM killer than
1333 system services or the managers themselves.
1335 * A new per-service setting RestrictFileSystems= as been added that
1336 restricts the file systems a service has access to by their type.
1337 This is based on the new BPF LSM of the Linux kernel. It provides an
1338 effective way to make certain API file systems unavailable to
1339 services (and thus minimizing attack surface). A new command
1340 "systemd-analyze filesystems" has been added that lists all known
1341 file system types (and how they are grouped together under useful
1344 * Services now support a new setting RestrictNetworkInterfaces= for
1345 restricting access to specific network interfaces.
1347 * Service unit files gained new settings StartupAllowedCPUs= and
1348 StartupAllowedMemoryNodes=. These are similar to their counterparts
1349 without the "Startup" prefix and apply during the boot process
1350 only. This is useful to improve boot-time behavior of the system and
1351 assign resources differently during boot than during regular
1352 runtime. This is similar to the preexisting StartupCPUWeight=
1355 * Related to this: the various StartupXYZ= settings
1356 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
1357 during shutdown. The settings not prefixed with "Startup" hence apply
1358 during regular runtime, and those that are prefixed like that apply
1359 during boot and shutdown.
1361 * A new per-unit set of conditions/asserts
1362 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
1363 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
1364 pressure is above the configured threshold, using the kernel PSI
1365 feature. For more details see systemd.unit(5) and
1366 https://docs.kernel.org/accounting/psi.html
1368 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
1369 ProtectKernelLogs=yes can now be used.
1371 * The default maximum numbers of inodes have been raised from 64k to 1M
1372 for /dev/, and from 400k to 1M for /tmp/.
1374 * The per-user service manager learnt support for communicating with
1375 systemd-oomd to acquire OOM kill information.
1377 * A new service setting ExecSearchPath= has been added that allows
1378 changing the search path for executables for services. It affects
1379 where we look for the binaries specified in ExecStart= and similar,
1380 and the specified directories are also added the $PATH environment
1381 variable passed to invoked processes.
1383 * A new setting RuntimeRandomizedExtraSec= has been added for service
1384 and scope units that allows extending the runtime time-out as
1385 configured by RuntimeMaxSec= with a randomized amount.
1387 * The syntax of the service unit settings RuntimeDirectory=,
1388 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
1389 if the specified value is now suffixed with a colon, followed by
1390 another filename, the latter will be created as symbolic link to the
1391 specified directory. This allows creating these service directories
1392 together with alias symlinks to make them available under multiple
1395 * Service unit files gained two new settings TTYRows=/TTYColumns= for
1396 configuring rows/columns of the TTY device passed to
1397 stdin/stdout/stderr of the service. This is useful to propagate TTY
1398 dimensions to a virtual machine.
1400 * A new service unit file setting ExitType= has been added that
1401 specifies when to assume a service has exited. By default systemd
1402 only watches the main process of a service. By setting
1403 ExitType=cgroup it can be told to wait for the last process in a
1406 * Automount unit files gained a new setting ExtraOptions= that can be
1407 used to configure additional mount options to pass to the kernel when
1408 mounting the autofs instance.
1410 * "Urlification" (generation of ESC sequences that generate clickable
1411 hyperlinks in modern terminals) may now be turned off altogether
1414 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
1415 settings that default to 200 and 2 s respectively. The ratelimit
1416 ensures that a path unit cannot cause PID1 to busy-loop when it is
1417 trying to trigger a service that is skipped because of a Condition*=
1418 not being satisfied. This matches the configuration and behaviour of
1421 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
1422 as a plug-in for cryptsetup. This means the plain cryptsetup command
1423 may now be used to unlock volumes set up this way.
1425 * The TPM2 logic in cryptsetup will now automatically detect systems
1426 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
1427 updates the SHA1 banks. In such a case PCR policies will be
1428 automatically bound to the latter, not the former. This makes the PCR
1429 policies reliable, but of course do not provide the same level of
1430 trust as SHA256 banks.
1432 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
1433 RSA primary keys in addition to ECC, improving compatibility with
1434 TPM2 chips that do not support ECC. RSA keys are much slower to use
1435 than ECC, and hence are only used if ECC is not available.
1437 * /etc/crypttab gained support for a new token-timeout= setting for
1438 encrypted volumes that allows configuration of the maximum time to
1439 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
1440 the logic will query the user for a regular passphrase/recovery key
1443 * Support for activating dm-integrity volumes at boot via a new file
1444 /etc/integritytab and the tool systemd-integritysetup have been
1445 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
1446 with dm-integrity instead of dm-crypt/dm-verity.
1448 * The systemd-veritysetup-generator now understands a new usrhash=
1449 kernel command line option for specifying the Verity root hash for
1450 the partition backing the /usr/ file system. A matching set of
1451 systemd.verity_usr_* kernel command line options has been added as
1452 well. These all work similar to the corresponding options for the
1455 * The sd-device API gained a new API call sd_device_get_diskseq() to
1456 return the DISKSEQ property of a device structure. The "disk
1457 sequence" concept is a new feature recently introduced to the Linux
1458 kernel that allows detecting reuse cycles of block devices, i.e. can
1459 be used to recognize when loopback block devices are reused for a
1460 different purpose or CD-ROM drives get their media changed.
1462 * A new unit systemd-boot-update.service has been added. If enabled
1463 (the default) and the sd-boot loader is detected to be installed, it
1464 is automatically updated to the newest version when out of date. This
1465 is useful to ensure the boot loader remains up-to-date, and updates
1466 automatically propagate from the OS tree in /usr/.
1468 * sd-boot will now build with SBAT by default in order to facilitate
1469 working with recent versions of Shim that require it to be present.
1471 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
1472 This is used to robustly generate boot entry titles for Windows.
1474 * A new generic target unit factory-reset.target has been added. It is
1475 hooked into systemd-logind similar in fashion to
1476 reboot/poweroff/suspend/hibernate, and is supposed to be used to
1477 initiate a factory reset operation. What precisely this operation
1478 entails is up for the implementer to decide, the primary goal of the
1479 new unit is provide a framework where to plug in the implementation
1480 and how to trigger it.
1482 * A new meson build-time option 'clock-valid-range-usec-max' has been
1483 added which takes a time in µs and defaults to 15 years. If the RTC
1484 time is noticed to be more than the specified time ahead of the
1485 built-in epoch of systemd (which by default is the release timestamp
1486 of systemd) it is assumed that the RTC is not working correctly, and
1487 the RTC is reset to the epoch. (It already is reset to the epoch when
1488 noticed to be before it.) This should increase the chance that time
1489 doesn't accidentally jump too far ahead due to faulty hardware or
1492 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
1493 which may be used to automatically save the current system time to
1494 disk in regular intervals. This is useful to maintain a roughly
1495 monotonic clock even without RTC hardware and with some robustness
1496 against abnormal system shutdown.
1498 * systemd-analyze verify gained support for a pair of new --image= +
1499 --root= switches for verifying units below a specific root
1500 directory/image instead of on the host.
1502 * systemd-analyze verify gained support for verifying unit files under
1503 an explicitly specified unit name, independently of what the filename
1506 * systemd-analyze verify gained a new switch --recursive-errors= which
1507 controls whether to only fail on errors found in the specified units
1508 or recursively any dependent units.
1510 * systemd-analyze security now supports a new --offline mode for
1511 analyzing unit files stored on disk instead of loaded units. It may
1512 be combined with --root=/--image to analyze unit files under a root
1513 directory or disk image. It also learnt a new --threshold= parameter
1514 for specifying an exposure level threshold: if the exposure level
1515 exceeds the specified value the call will fail. It also gained a new
1516 --security-policy= switch for configuring security policies to
1517 enforce on the units. A policy is a JSON file that lists which tests
1518 shall be weighted how much to determine the overall exposure
1519 level. Altogether these new features are useful for fully automatic
1520 analysis and enforcement of security policies on unit files.
1522 * systemd-analyze security gain a new --json= switch for JSON output.
1524 * systemd-analyze learnt a new --quiet switch for reducing
1525 non-essential output. It's honored by the "dot", "syscall-filter",
1526 "filesystems" commands.
1528 * systemd-analyze security gained a --profile= option that can be used
1529 to take into account a portable profile when analyzing portable
1530 services, since a lot of the security-related settings are enabled
1533 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
1534 files, binaries and executables and prints metadata information,
1535 including the build-id and other info described on:
1536 https://systemd.io/COREDUMP_PACKAGE_METADATA/
1538 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
1539 section, for automatically propagating DNS settings from other
1542 * The static lease DHCP server logic in systemd-networkd may now serve
1543 IP addresses outside of the configured IP pool range for the server.
1545 * CAN support in systemd-networkd gained four new settings Loopback=,
1546 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
1547 control modes. It gained a number of further settings for tweaking
1550 * The [CAN] section in .network file gained new TimeQuantaNSec=,
1551 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
1552 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
1553 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
1554 DataSyncJumpWidth= settings to control bit-timing processed by the
1557 * DHCPv4 client support in systemd-networkd learnt a new Label= option
1558 for configuring the address label to apply to configure IPv4
1561 * The [IPv6AcceptRA] section of .network files gained support for a new
1562 UseMTU= setting that may be used to control whether to apply the
1563 announced MTU settings to the local interface.
1565 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
1566 setting to control whether the DHCPv4 client request and process the
1569 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
1570 [DHCPPrefixDelegation], as now the prefix delegation is also supported
1571 with DHCPv4 protocol by enabling the Use6RD= setting.
1573 * The [DHCPPrefixDelegation] section in .network file gained a new
1574 setting UplinkInterface= to specify the upstream interface.
1576 * The [DHCPv6] section in .network file gained a new setting
1577 UseDelegatedPrefix= to control whether the delegated prefixes will be
1578 propagated to the downstream interfaces.
1580 * The [IPv6AcceptRA] section of .network files now understands two new
1581 settings UseGateway=/UseRoutePrefix= for explicitly configuring
1582 whether to use the relevant fields from the IPv6 Router Advertisement
1585 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
1586 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
1587 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
1588 [IPv6AcceptRA] section to control when the DHCPv6 client is started
1589 and how the delegated prefixes are handled by the DHCPv6 client.
1591 * The IPv6Token= section in the [Network] section is deprecated, and
1592 the [IPv6AcceptRA] section gained the Token= setting for its
1593 replacement. The [IPv6Prefix] section also gained the Token= setting.
1594 The Token= setting gained 'eui64' mode to explicitly configure an
1595 address with the EUI64 algorithm based on the interface MAC address.
1596 The 'prefixstable' mode can now optionally take a secret key. The
1597 Token= setting in the [DHCPPrefixDelegation] section now supports all
1598 algorithms supported by the same settings in the other sections.
1600 * The [RoutingPolicyRule] section of .network file gained a new
1601 SuppressInterfaceGroup= setting.
1603 * The IgnoreCarrierLoss= setting in the [Network] section of .network
1604 files now allows a duration to be specified, controlling how long to
1605 wait before reacting to carrier loss.
1607 * The [DHCPServer] section of .network file gained a new Router=
1608 setting to specify the router address.
1610 * The [CAKE] section of .network files gained various new settings
1611 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
1612 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
1613 and UseRawPacketSize= for configuring CAKE.
1615 * systemd-networkd now ships with new default .network files:
1616 80-container-vb.network which matches host-side network bridge device
1617 created by systemd-nspawn's --network-bridge or --network-zone
1618 switch, and 80-6rd-tunnel.network which matches automatically created
1619 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
1621 * systemd-networkd's handling of Endpoint= resolution for WireGuard
1622 interfaces has been improved.
1624 * systemd-networkd will now automatically configure routes to addresses
1625 specified in AllowedIPs=. This feature can be controlled via
1626 RouteTable= and RouteMetric= settings in [WireGuard] or
1627 [WireGuardPeer] sections.
1629 * systemd-networkd will now once again automatically generate persistent
1630 MAC addresses for batadv and bridge interfaces. Users can disable this
1631 by using MACAddress=none in .netdev files.
1633 * systemd-networkd and systemd-udevd now support IP over InfiniBand
1634 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
1635 systemd.netdev files gained the [IPoIB] section.
1637 * systemd-networkd and systemd-udevd now support net.ifname-policy=
1638 option on the kernel command-line. This is implemented through the
1639 systemd-network-generator service that automatically generates
1640 appropriate .link, .network, and .netdev files.
1642 * The various systemd-udevd "ethtool" buffer settings now understand
1643 the special value "max" to configure the buffers to the maximum the
1646 * systemd-udevd's .link files may now configure a large variety of
1647 NIC coalescing settings, plus more hardware offload settings.
1649 * .link files gained a new WakeOnLanPassword= setting in the [Link]
1650 section that allows to specify a WoL "SecureOn" password on hardware
1653 * systemd-nspawn's --setenv= switch now supports an additional syntax:
1654 if only a variable name is specified (i.e. without being suffixed by
1655 a '=' character and a value) the current value of the environment
1656 variable is propagated to the container. e.g. --setenv=FOO will
1657 lookup the current value of $FOO in the environment, and pass it down
1658 to the container. Similar behavior has been added to homectl's,
1659 machinectl's and systemd-run's --setenv= switch.
1661 * systemd-nspawn gained a new switch --suppress-sync= which may be used
1662 to optionally suppress the effect of the sync()/fsync()/fdatasync()
1663 system calls for the container payload. This is useful for build
1664 system environments where safety against abnormal system shutdown is
1665 not essential as all build artifacts can be regenerated any time, but
1666 the performance win is beneficial.
1668 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
1669 same value that PID 1 uses for most forked off processes.
1671 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
1672 uidmap/nouidmap options as last parameter. If "uidmap" is used the
1673 bind mounts are created with UID mapping taking place that ensures
1674 the host's file ownerships are mapped 1:1 to container file
1675 ownerships, even if user namespacing is used. This way
1676 files/directories bound into containers will no longer show up as
1677 owned by the nobody user as they typically did if no special care was
1678 taken to shift them manually.
1680 * When discovering Windows installations sd-boot will now attempt to
1681 show the Windows version.
1683 * The color scheme to use in sd-boot may now be configured at
1686 * sd-boot gained the ability to change screen resolution during
1687 boot-time, by hitting the "r" key. This will cycle through available
1688 resolutions and save the last selection.
1690 * sd-boot learnt a new hotkey "f". When pressed the system will enter
1691 firmware setup. This is useful in environments where it is difficult
1692 to hit the right keys early enough to enter the firmware, and works
1693 on any firmware regardless which key it natively uses.
1695 * sd-boot gained support for automatically booting into the menu item
1696 selected on the last boot (using the "@saved" identifier for menu
1699 * sd-boot gained support for automatically loading all EFI drivers
1700 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
1701 Partition (ESP). These drivers are loaded before the menu entries are
1702 loaded. This is useful e.g. to load additional file system drivers
1703 for the XBOOTLDR partition.
1705 * systemd-boot will now paint the input cursor on its own instead of
1706 relying on the firmware to do so, increasing compatibility with broken
1707 firmware that doesn't make the cursor reasonably visible.
1709 * sd-boot now embeds a .osrel PE section like we expect from Boot
1710 Loader Specification Type #2 Unified Kernels. This means sd-boot
1711 itself may be used in place of a Type #2 Unified Kernel. This is
1712 useful for debugging purposes as it allows chain-loading one a
1713 (development) sd-boot instance from another.
1715 * sd-boot now supports a new "devicetree" field in Boot Loader
1716 Specification Type #1 entries: if configured the specified device
1717 tree file is installed before the kernel is invoked. This is useful
1718 for installing/applying new devicetree files without updating the
1721 * Similarly, sd-stub now can read devicetree data from a PE section
1722 ".dtb" and apply it before invoking the kernel.
1724 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
1725 gained the ability to pick up credentials and sysext files, wrap them
1726 in a cpio archive, and pass as an additional initrd to the invoked
1727 Linux kernel, in effect placing those files in the /.extra/ directory
1728 of the initrd environment. This is useful to implement trusted initrd
1729 environments which are fully authenticated but still can be extended
1730 (via sysexts) and parameterized (via encrypted/authenticated
1731 credentials, see above).
1733 Credentials can be located next to the kernel image file (credentials
1734 specific to a single boot entry), or in one of the shared directories
1735 (credentials applicable to multiple boot entries).
1737 * sd-stub now comes with a full man page, that explains its feature set
1738 and how to combine a kernel image, an initrd and the stub to build a
1739 complete EFI unified kernel image, implementing Boot Loader
1740 Specification Type #2.
1742 * sd-stub may now provide the initrd to the executed kernel via the
1743 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
1744 non-x86 architectures.
1746 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
1747 may be used to set the boot menu time-out of the boot loader (for all
1748 or just the subsequent boot).
1750 * bootctl and kernel-install will now read variables
1751 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
1752 /etc/kernel/install.conf. When set, it specifies the layout to use
1753 for installation directories on the boot partition, so that tools
1754 don't need to guess it based on the already-existing directories. The
1755 only value that is defined natively is "bls", corresponding to the
1757 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
1758 kernel-install that implement a different layout can declare other
1759 values for this variable.
1761 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
1762 assumption that if the user installed sd-boot to the ESP, they intend
1763 to use the entry layout understood by sd-boot. It'll also write
1764 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
1765 (and it wasn't specified in the config file yet). Similarly,
1766 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
1767 wasn't specified in the config file yet). Effectively, those changes
1768 mean that the machine-id used for boot loader entry installation is
1769 "frozen" upon first use and becomes independent of the actual
1772 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
1773 images created for distribution ("golden images") are built with no
1774 machine-id, so that a unique machine-id can be created on the first
1775 boot. But those images may contain boot loader entries with the
1776 machine-id used during build included in paths. Using a "frozen"
1777 value allows unambiguously identifying entries that match the
1778 specific installation, while still permitting parallel installations
1781 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
1782 kernel-install to guess the installation layout. This fixes the
1783 problem where a (possibly empty) directory in the boot partition is
1784 created from a different layout causing kernel-install plugins to
1785 assume the wrong layout. A particular example of how this may happen
1786 is the grub2 package in Fedora which includes directories under /boot
1787 directly in its file list. Various other packages pull in grub2 as a
1788 dependency, so it may be installed even if unused, breaking
1789 installations that use the bls layout.
1791 * bootctl and systemd-bless-boot can now be linked statically.
1793 * systemd-sysext now optionally doesn't insist on extension-release.d/
1794 files being placed in the image under the image's file name. If the
1795 file system xattr user.extension-release.strict is set on the
1796 extension release file, it is accepted regardless of its name. This
1797 relaxes security restrictions a bit, as system extension may be
1798 attached under a wrong name this way.
1800 * udevadm's test-builtin command learnt a new --action= switch for
1801 testing the built-in with the specified action (in place of the
1804 * udevadm info gained new switches --property=/--value for showing only
1805 specific udev properties/values instead of all.
1807 * A new hwdb database has been added that contains matches for various
1808 types of signal analyzers (protocol analyzers, logic analyzers,
1809 oscilloscopes, multimeters, bench power supplies, etc.) that should
1810 be accessible to regular users.
1812 * A new hwdb database entry has been added that carries information
1813 about types of cameras (regular or infrared), and in which direction
1814 they point (front or back).
1816 * A new rule to allow console users access to rfkill by default has been
1819 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
1820 now also owned by the system group "sgx".
1822 * A new build-time meson option "extra-net-naming-schemes=" has been
1823 added to define additional naming schemes schemes for udev's network
1824 interface naming logic. This is useful for enterprise distributions
1825 and similar which want to pin the schemes of certain distribution
1826 releases under a specific name and previously had to patch the
1827 sources to introduce new named schemes.
1829 * The predictable naming logic for network interfaces has been extended
1830 to generate stable names from Xen netfront device information.
1832 * hostnamed's chassis property can now be sourced from chassis-type
1833 field encoded in devicetree (in addition to the existing DMI
1836 * systemd-cgls now optionally displays cgroup IDs and extended
1837 attributes for each cgroup. (Controllable via the new --xattr= +
1838 --cgroup-id= switches.)
1840 * coredumpctl gained a new --all switch for operating on all
1841 Journal files instead of just the local ones.
1843 * systemd-coredump will now use libdw/libelf via dlopen() rather than
1844 directly linking, allowing users to easily opt-out of backtrace/metadata
1845 analysis of core files, and reduce image sizes when this is not needed.
1847 * systemd-coredump will now analyze core files with libdw/libelf in a
1848 forked, sandboxed process.
1850 * systemd-homed will now try to unmount an activate home area in
1851 regular intervals once the user logged out fully. Previously this was
1852 attempted exactly once but if the home directory was busy for some
1853 reason it was not tried again.
1855 * systemd-homed's LUKS2 home area backend will now create a BSD file
1856 system lock on the image file while the home area is active
1857 (i.e. mounted). If a home area is found to be locked, logins are
1858 politely refused. This should improve behavior when using home areas
1859 images that are accessible via the network from multiple clients, and
1860 reduce the chance of accidental file system corruption in that case.
1862 * Optionally, systemd-homed will now drop the kernel buffer cache once
1863 a user has fully logged out, configurable via the new --drop-caches=
1866 * systemd-homed now makes use of UID mapped mounts for the home areas.
1867 If the kernel and used file system support it, files are now
1868 internally owned by the "nobody" user (i.e. the user typically used
1869 for indicating "this ownership is not mapped"), and dynamically
1870 mapped to the UID used locally on the system via the UID mapping
1871 mount logic of recent kernels. This makes migrating home areas
1872 between different systems cheaper because recursively chown()ing file
1873 system trees is no longer necessary.
1875 * systemd-homed's CIFS backend now optionally supports CIFS service
1876 names with a directory suffix, in order to place home directories in
1877 a subdirectory of a CIFS share, instead of the top-level directory.
1879 * systemd-homed's CIFS backend gained support for specifying additional
1880 mount options in the JSON user record (cifsExtraMountOptions field,
1881 and --cifs-extra-mount-options= homectl switch). This is for example
1882 useful for configuring mount options such as "noserverino" that some
1883 SMB3 services require (use that to run a homed home directory from a
1884 FritzBox SMB3 share this way).
1886 * systemd-homed will now default to btrfs' zstd compression for home
1887 areas. This is inspired by Fedora's recent decision to switch to zstd
1890 * Additional mount options to use when mounting the file system of
1891 LUKS2 volumes in systemd-homed has been added. Via the
1892 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
1893 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
1894 systemd-homed or via the luksExtraMountOptions user record JSON
1895 property. (Exposed via homectl --luks-extra-mount-options)
1897 * homectl's resize command now takes the special size specifications
1898 "min" and "max" to shrink/grow the home area to the minimum/maximum
1899 size possible, taking disk usage/space constraints and file system
1900 limitations into account. Resizing is now generally graceful: the
1901 logic will try to get as close to the specified size as possible, but
1902 not consider it a failure if the request couldn't be fulfilled
1905 * systemd-homed gained the ability to automatically shrink home areas
1906 on logout to their minimal size and grow them again on next
1907 login. This ensures that while inactive, a home area only takes up
1908 the minimal space necessary, but once activated, it provides
1909 sufficient space for the user's needs. This behavior is only
1910 supported if btrfs is used as file system inside the home area
1911 (because only for btrfs online growing/shrinking is implemented in
1912 the kernel). This behavior is now enabled by default, but may be
1913 controlled via the new --auto-resize-mode= setting of homectl.
1915 * systemd-homed gained support for automatically re-balancing free disk
1916 space among active home areas, in case the LUKS2 backends are used,
1917 and no explicit disk size was requested. This way disk space is
1918 automatically managed and home areas resized in regular intervals and
1919 manual resizing when disk space becomes scarce should not be
1920 necessary anymore. This behavior is only supported if btrfs is used
1921 within the home areas (as only then online shrinking and growing is
1922 supported), and may be configured via the new rebalanceWeight JSON
1923 user record field (as exposed via the new --rebalance-weight= homectl
1924 setting). Re-balancing is mostly automatic, but can also be requested
1925 explicitly via "homectl rebalance", which is synchronous, and thus
1926 may be used to wait until the rebalance run is complete.
1928 * userdbctl gained a --json= switch for configured the JSON formatting
1929 to use when outputting user or group records.
1931 * userdbctl gained a new --multiplexer= switch for explicitly
1932 configuring whether to use the systemd-userdbd server side user
1933 record resolution logic.
1935 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
1936 for chaining up another command to execute after completing the
1937 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
1938 configuration of a single command to invoke, this maybe used to
1939 invoke multiple: first userdbctl's own implementation, and then any
1940 other also configured in the command line.
1942 * The sd-event API gained a new function sd_event_add_inotify_fd() that
1943 is similar to sd_event_add_inotify() but accepts a file descriptor
1944 instead of a path in the file system for referencing the inode to
1947 * The sd-event API gained a new function
1948 sd_event_source_set_ratelimit_expire_callback() that may be used to
1949 define a callback function that is called whenever an event source
1950 leaves the rate limiting phase.
1952 * New documentation has been added explaining which steps are necessary
1953 to port systemd to a new architecture:
1955 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
1957 * The x-systemd.makefs option in /etc/fstab now explicitly supports
1958 ext2, ext3, and f2fs file systems.
1960 * Mount units and units generated from /etc/fstab entries with 'noauto'
1961 are now ordered the same as other units. Effectively, they will be
1962 started earlier (if something actually pulled them in) and stopped
1963 later, similarly to normal mount units that are part of
1964 fs-local.target. This change should be invisible to users, but
1965 should prevent those units from being stopped too early during
1968 * The systemd-getty-generator now honors a new kernel command line
1969 argument systemd.getty_auto= and a new environment variable
1970 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
1971 example useful to turn off gettys inside of containers or similar
1974 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
1975 (in addition to 127.0.0.53, as before). If DNS requests are sent to
1976 this address they are propagated in "bypass" mode only, i.e. are
1977 almost not processed locally, but mostly forwarded as-is to the
1978 current upstream DNS servers. This provides a stable DNS server
1979 address that proxies all requests dynamically to the right upstream
1980 DNS servers even if these dynamically change. This stub does not do
1981 mDNS/LLMNR resolution. However, it will translate look-ups to
1982 DNS-over-TLS if necessary. This new stub is particularly useful in
1983 container/VM environments, or for tethering setups: use DNAT to
1984 redirect traffic to any IP address to this stub.
1986 * systemd-importd now honors new environment variables
1987 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
1988 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
1989 generation, btrfs quota setup and disk synchronization.
1991 * systemd-importd and systemd-resolved can now be optionally built with
1992 OpenSSL instead of libgcrypt.
1994 * systemd-repart no longer requires OpenSSL.
1996 * systemd-sysusers will no longer create the redundant 'nobody' group
1997 by default, as the 'nobody' user is already created with an
1998 appropriate primary group.
2000 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
2002 * systemctl show-environment gained support for --output=json.
2004 * pam_systemd will now first try to use the X11 abstract socket, and
2005 fallback to the socket file in /tmp/.X11-unix/ only if that does not
2008 * systemd-journald will no longer go back to volatile storage
2009 regardless of configuration when its unit is restarted.
2011 * Initial support for the LoongArch architecture has been added (system
2012 call lists, GPT partition table UUIDs, etc).
2014 * systemd-journald's own logging messages are now also logged to the
2015 journal itself when systemd-journald logs to /dev/kmsg.
2017 * systemd-journald now re-enables COW for archived journal files on
2018 filesystems that support COW. One benefit of this change is that
2019 archived journal files will now get compressed on btrfs filesystems
2020 that have compression enabled.
2022 * systemd-journald now deduplicates fields in a single log message
2023 before adding it to the journal. In archived journal files, it will
2024 also punch holes for unused parts and truncate the file as
2025 appropriate, leading to reductions in disk usage.
2027 * journalctl --verify was extended with more informative error
2030 * More of sd-journal's functions are now resistant against journal file
2033 * The shutdown command learnt a new option --show, to display the
2036 * A LICENSES/ directory is now included in the git tree. It contains a
2037 README.md file that explains the licenses used by source files in
2038 this repository. It also contains the text of all applicable
2039 licenses as they appear on spdx.org.
2041 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
2042 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
2043 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
2044 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
2045 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
2046 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
2047 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
2048 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
2049 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
2050 Christian Brauner, Christian Göttsche, Christian Wehrli,
2051 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
2052 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
2053 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
2054 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
2055 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
2056 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
2057 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
2058 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
2059 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
2060 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
2061 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
2062 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
2063 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
2064 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
2065 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
2066 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
2067 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
2068 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
2069 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
2070 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
2071 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
2072 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
2073 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
2074 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
2075 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
2076 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
2077 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
2078 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
2079 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
2080 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
2081 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
2082 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
2083 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
2084 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
2085 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
2086 Дамјан Георгиевски, наб
2088 — Warsaw, 2021-12-23
2092 * When operating on disk images via the --image= switch of various
2093 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
2094 no 'root=' parameter on the kernel command line, and multiple
2095 suitable root or /usr/ partitions exist in the image, then a simple
2096 comparison inspired by strverscmp() is done on the GPT partition
2097 label, and the newest partition is picked. This permits a simple and
2098 generic whole-file-system A/B update logic where new operating system
2099 versions are dropped into partitions whose label is then updated with
2100 a matching version identifier.
2102 * systemd-sysusers now supports querying the passwords to set for the
2103 users it creates via the "credentials" logic introduced in v247: the
2104 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
2105 credentials are consulted for the password to use (either in UNIX
2106 hashed form, or literally). By default these credentials are inherited
2107 down from PID1 (which in turn imports it from a container manager if
2108 there is one). This permits easy configuration of user passwords
2109 during first boot. Example:
2111 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
2113 Note that systemd-sysusers operates in purely additive mode: it
2114 executes no operation if the declared users already exist, and hence
2115 doesn't set any passwords as effect of the command line above if the
2116 specified root user exists already in the image. (Note that
2117 --volatile=yes ensures it doesn't, though.)
2119 * systemd-firstboot now also supports querying various system
2120 parameters via the credential subsystems. Thus, as above this may be
2121 used to initialize important system parameters on first boot of
2122 previously unprovisioned images (i.e. images with a mostly empty
2125 * PID 1 may now show both the unit name and the unit description
2126 strings in its status output during boot. This may be configured with
2127 StatusUnitFormat=combined in system.conf or
2128 systemd.status-unit-format=combined on the kernel command line.
2130 * The systemd-machine-id-setup tool now supports a --image= switch for
2131 provisioning a machine ID file into an OS disk image, similar to how
2132 --root= operates on an OS file tree. This matches the existing switch
2133 of the same name for systemd-tmpfiles, systemd-firstboot, and
2134 systemd-sysusers tools.
2136 * Similarly, systemd-repart gained support for the --image= switch too.
2137 In combination with the existing --size= option, this makes the tool
2138 particularly useful for easily growing disk images in a single
2139 invocation, following the declarative rules included in the image
2142 * systemd-repart's partition configuration files gained support for a
2143 new switch MakeDirectories= which may be used to create arbitrary
2144 directories inside file systems that are created, before registering
2145 them in the partition table. This is useful in particular for root
2146 partitions to create mount point directories for other partitions
2147 included in the image. For example, a disk image that contains a
2148 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
2149 create /home/ and /var/ as empty directories in the root file system
2150 on its creation, so that the resulting image can be mounted
2151 immediately, even in read-only mode.
2153 * systemd-repart's CopyBlocks= setting gained support for the special
2154 value "auto". If used, a suitable matching partition on the booted OS
2155 is found as source to copy blocks from. This is useful when
2156 implementing replicating installers, that are booted from one medium
2157 and then stream their own root partition onto the target medium.
2159 * systemd-repart's partition configuration files gained support for a
2160 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
2161 GPT partition flags for the created partitions: this is useful for
2162 marking newly created partitions as read-only, or as not being
2163 subject for automatic mounting from creation on.
2165 * The /etc/os-release file has been extended with two new (optional)
2166 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
2167 information for OS images that are updated comprehensively and
2168 atomically as one image. Two new specifiers %M, %A now resolve to
2169 these two fields in the various configuration options that resolve
2172 * portablectl gained a new switch --extension= for enabling portable
2173 service images with extensions that follow the extension image
2174 concept introduced with v248, and thus allows layering multiple
2175 images when setting up the root filesystem of the service.
2177 * systemd-coredump will now extract ELF build-id information from
2178 processes dumping core and include it in the coredump report.
2179 Moreover, it will look for ELF .note.package sections with
2180 distribution packaging meta-information about the crashing process.
2181 This is useful to directly embed the rpm or deb (or any other)
2182 package name and version in ELF files, making it easy to match
2183 coredump reports with the specific package for which the software was
2184 compiled. This is particularly useful on environments with ELF files
2185 from multiple vendors, different distributions and versions, as is
2186 common today in our containerized and sand-boxed world. For further
2189 https://systemd.io/COREDUMP_PACKAGE_METADATA
2191 * A new udev hardware database has been added for FireWire devices
2194 * The "net_id" built-in of udev has been updated with three
2195 backwards-incompatible changes:
2197 - PCI hotplug slot names on s390 systems are now parsed as
2198 hexadecimal numbers. They were incorrectly parsed as decimal
2199 previously, or ignored if the name was not a valid decimal
2202 - PCI onboard indices up to 65535 are allowed. Previously, numbers
2203 above 16383 were rejected. This primarily impacts s390 systems,
2204 where values up to 65535 are used.
2206 - Invalid characters in interface names are replaced with "_".
2208 The new version of the net naming scheme is "v249". The previous
2209 scheme can be selected via the "net.naming-scheme=v247" kernel
2210 command line parameter.
2212 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
2213 NULL bus object, for which they will return false. Or in other words,
2214 an unallocated bus connection is neither ready nor open.
2216 * The sd-device API acquired a new API function
2217 sd_device_get_usec_initialized() that returns the monotonic time when
2218 the udev device first appeared in the database.
2220 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
2221 sd_device_get_trigger_uuid(). The former is similar to
2222 sd_device_trigger() but returns a randomly generated UUID that is
2223 associated with the synthetic uevent generated by the call. This UUID
2224 may be read from the sd_device object a monitor eventually receives,
2225 via the sd_device_get_trigger_uuid(). This interface requires kernel
2226 4.13 or above to work, and allows tracking a synthetic uevent through
2227 the entire device management stack. The "udevadm trigger --settle"
2228 logic has been updated to make use of this concept if available to
2229 wait precisely for the uevents it generates. "udevadm trigger" also
2230 gained a new parameter --uuid that prints the UUID for each generated
2233 * sd-device also gained new APIs sd_device_new_from_ifname() and
2234 sd_device_new_from_ifindex() for allocating an sd-device object for
2235 the specified network interface. The former accepts an interface name
2236 (either a primary or an alternative name), the latter an interface
2239 * The native Journal protocol has been documented. Clients may talk
2240 this as alternative to the classic BSD syslog protocol for locally
2241 delivering log records to the Journal. The protocol has been stable
2242 for a long time and in fact been implemented already in a variety
2243 of alternative client libraries. This documentation makes the support
2246 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
2248 * A new BPFProgram= setting has been added to service files. It may be
2249 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
2250 file, or a bind mount or symlink to one. This may be used to upload
2251 and manage BPF programs externally and then hook arbitrary systemd
2254 * The "home.arpa" domain that has been officially declared as the
2255 choice for domain for local home networks per RFC 8375 has been added
2256 to the default NTA list of resolved, since DNSSEC is generally not
2257 available on private domains.
2259 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
2261 * A new ManageForeignRoutingPolicyRules= setting has been added to
2262 .network files which may be used to exclude foreign-created routing
2263 policy rules from systemd-networkd management.
2265 * systemd-network-wait-online gained two new switches -4 and -6 that
2266 may be used to tweak whether to wait for only IPv4 or only IPv6
2269 * .network files gained a new RequiredFamilyForOnline= setting to
2270 fine-tune whether to require an IPv4 or IPv6 address in order to
2271 consider an interface "online".
2273 * networkctl will now show an over-all "online" state in the per-link
2276 * In .network files a new OutgoingInterface= setting has been added to
2277 specify the output interface in bridge FDB setups.
2279 * In .network files the Multipath group ID may now be configured for
2280 [NextHop] entries, via the new Group= setting.
2282 * The DHCP server logic configured in .network files gained a new
2283 setting RelayTarget= that turns the server into a DHCP server relay.
2284 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
2285 to further tweak the DHCP relay behaviour.
2287 * The DHCP server logic also gained a new ServerAddress= setting in
2288 .network files that explicitly specifies the server IP address to
2289 use. If not specified, the address is determined automatically, as
2292 * The DHCP server logic in systemd-networkd gained support for static
2293 DHCP leases, configurable via the [DHCPServerStaticLease]
2294 section. This allows explicitly mapping specific MAC addresses to
2295 fixed IP addresses and vice versa.
2297 * The RestrictAddressFamilies= setting in service files now supports a
2298 new special value "none". If specified sockets of all address
2299 families will be made unavailable to services configured that way.
2301 * systemd-fstab-generator and systemd-repart have been updated to
2302 support booting from disks that carry only a /usr/ partition but no
2303 root partition yet, and where systemd-repart can add it in on the
2304 first boot. This is useful for implementing systems that ship with a
2305 single /usr/ file system, and whose root file system shall be set up
2306 and formatted on a LUKS-encrypted volume whose key is generated
2307 locally (and possibly enrolled in the TPM) during the first boot.
2309 * The [Address] section of .network files now accepts a new
2310 RouteMetric= setting that configures the routing metric to use for
2311 the prefix route created as effect of the address configuration.
2312 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
2313 gained matching settings for their prefix routes. (The option of the
2314 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
2315 it conceptually belongs there; the old option is still understood for
2318 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
2321 * A new udev property ID_NET_DHCP_BROADCAST on network interface
2322 devices is now honoured by systemd-networkd, controlling whether to
2323 issue DHCP offers via broadcasting. This is used to ensure that s390
2324 layer 3 network interfaces work out-of-the-box with systemd-networkd.
2326 * nss-myhostname and systemd-resolved will now synthesize address
2327 records for a new special hostname "_outbound". The name will always
2328 resolve to the local IP addresses most likely used for outbound
2329 connections towards the default routes. On multi-homed hosts this is
2330 useful to have a stable handle referring to "the" local IP address
2331 that matters most, to the point where this is defined.
2333 * The Discoverable Partition Specification has been updated with a new
2334 GPT partition flag "grow-file-system" defined for its partition
2335 types. Whenever partitions with this flag set are automatically
2336 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
2337 of systemd-nspawn or other tools; and as opposed to explicit mounting
2338 via /etc/fstab), the file system within the partition is
2339 automatically grown to the full size of the partition. If the file
2340 system size already matches the partition size this flag has no
2341 effect. Previously, this functionality has been available via the
2342 explicit x-systemd.growfs mount option, and this new flag extends
2343 this to automatically discovered mounts. A new GrowFileSystem=
2344 setting has been added to systemd-repart drop-in files that allows
2345 configuring this partition flag. This new flag defaults to on for
2346 partitions automatically created by systemd-repart, except if they
2347 are marked read-only. See the specification for further details:
2349 https://systemd.io/DISCOVERABLE_PARTITIONS
2351 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
2352 section. If enabled (which is the default), and an NTP server address
2353 is acquired through a DHCP lease on this interface an explicit route
2354 to this address is created on this interface to ensure that NTP
2355 traffic to the NTP server acquired on an interface is also routed
2356 through that interface. The pre-existing RoutesToDNS= setting that
2357 implements the same for DNS servers is now enabled by default.
2359 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
2360 been added that may be used to restrict the network interfaces
2361 sockets created by the service may be bound to. This is implemented
2364 * A new ConditionFirmware= setting has been added to unit files to
2365 conditionalize on certain firmware features. At the moment it may
2366 check whether running on an UEFI system, a device.tree system, or if
2367 the system is compatible with some specified device-tree feature.
2369 * A new ConditionOSRelease= setting has been added to unit files to
2370 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
2371 operators may be used to check if some field has some specific value
2372 or do an alphanumerical comparison. Equality comparisons are useful
2373 for fields like ID, but relative comparisons for fields like
2374 VERSION_ID or IMAGE_VERSION.
2376 * hostnamed gained a new Describe() D-Bus method that returns a JSON
2377 serialization of the host data it exposes. This is exposed via
2378 "hostnamectl --json=" to acquire a host identity description in JSON.
2379 It's our intention to add a similar features to most services and
2380 objects systemd manages, in order to simplify integration with
2381 program code that can consume JSON.
2383 * Similarly, networkd gained a Describe() method on its Manager and
2384 Link bus objects. This is exposed via "networkctl --json=".
2386 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
2387 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
2388 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
2389 that is used both to get the value (when no argument is given), and
2390 to set the value (when an argument is specified). The old names
2391 continue to be supported for compatibility.
2393 * systemd-detect-virt and ConditionVirtualization= are now able to
2394 correctly identify Amazon EC2 environments.
2396 * The LogLevelMax= setting of unit files now applies not only to log
2397 messages generated *by* the service, but also to log messages
2398 generated *about* the service by PID 1. To suppress logs concerning a
2399 specific service comprehensively, set this option to a high log
2402 * bootctl gained support for a new --make-machine-id-directory= switch
2403 that allows precise control on whether to create the top-level
2404 per-machine directory in the boot partition that typically contains
2405 Type 1 boot loader entries.
2407 * During build SBAT data to include in the systemd-boot EFI PE binaries
2408 may be specified now.
2410 * /etc/crypttab learnt a new option "headless". If specified any
2411 requests to query the user interactively for passwords or PINs will
2412 be skipped. This is useful on systems that are headless, i.e. where
2413 an interactive user is generally not present.
2415 * /etc/crypttab also learnt a new option "password-echo=" that allows
2416 configuring whether the encryption password prompt shall echo the
2417 typed password and if so, do so literally or via asterisks. (The
2418 default is the same behaviour as before: provide echo feedback via
2421 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
2422 systemd-homed has been updated to allow explicit configuration of the
2423 "user presence" and "user verification" checks, as well as whether a
2424 PIN is required for authentication, via the new switches
2425 --fido2-with-user-presence=, --fido2-with-user-verification=,
2426 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
2427 features are available, and may be enabled or disabled depends on the
2430 * systemd-nspawn's --private-user= switch now accepts the special value
2431 "identity" which configures a user namespacing environment with an
2432 identity mapping of 65535 UIDs. This means the container UID 0 is
2433 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
2434 this doesn't appear to be useful, however it does reduce the attack
2435 surface a bit, since the resulting container will possess process
2436 capabilities only within its namespace and not on the host.
2438 * systemd-nspawn's --private-user-chown switch has been replaced by a
2439 more generic --private-user-ownership= switch that accepts one of
2440 three values: "chown" is equivalent to the old --private-user-chown,
2441 and "off" is equivalent to the absence of the old switch. The value
2442 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
2443 of files and directories of the underlying image to the chosen UID
2444 range for the container. "auto" is equivalent to "map" if UID mapping
2445 mount are supported, otherwise it is equivalent to "chown". The short
2446 -U switch systemd-nspawn now implies --private-user-ownership=auto
2447 instead of the old --private-user-chown. Effectively this means: if
2448 the backing file system supports UID mapping mounts the feature is
2449 now used by default if -U is used. Generally, it's a good idea to use
2450 UID mapping mounts instead of recursive chown()ing, since it allows
2451 running containers off immutable images (since no modifications of
2452 the images need to take place), and share images between multiple
2453 instances. Moreover, the recursive chown()ing operation is slow and
2454 can be avoided. Conceptually it's also a good thing if transient UID
2455 range uses do not leak into persistent file ownership anymore. TLDR:
2456 finally, the last major drawback of user namespacing has been
2457 removed, and -U should always be used (unless you use btrfs, where
2458 UID mapped mounts do not exist; or your container actually needs
2459 privileges on the host).
2461 * nss-systemd now synthesizes user and group shadow records in addition
2462 to the main user and group records. Thus, hashed passwords managed by
2463 systemd-homed are now accessible via the shadow database.
2465 * The userdb logic (and thus nss-systemd, and so on) now read
2466 additional user/group definitions in JSON format from the drop-in
2467 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
2468 /usr/lib/userdb/. This is a simple and powerful mechanism for making
2469 additional users available to the system, with full integration into
2470 NSS including the shadow databases. Since the full JSON user/group
2471 record format is supported this may also be used to define users with
2472 resource management settings and other runtime settings that
2473 pam_systemd and systemd-logind enforce at login.
2475 * The userdbctl tool gained two new switches --with-dropin= and
2476 --with-varlink= which can be used to fine-tune the sources used for
2477 user database lookups.
2479 * systemd-nspawn gained a new switch --bind-user= for binding a host
2480 user account into the container. This does three things: the user's
2481 home directory is bind mounted from the host into the container,
2482 below the /run/userdb/home/ hierarchy. A free UID is picked in the
2483 container, and a user namespacing UID mapping to the host user's UID
2484 installed. And finally, a minimal JSON user and group record (along
2485 with its hashed password) is dropped into /run/host/userdb/. These
2486 records are picked up automatically by the userdb drop-in logic
2487 describe above, and allow the user to login with the same password as
2488 on the host. Effectively this means: if host and container run new
2489 enough systemd versions making a host user available to the container
2490 is trivially simple.
2492 * systemd-journal-gatewayd now supports the switches --user, --system,
2493 --merge, --file= that are equivalent to the same switches of
2494 journalctl, and permit exposing only the specified subset of the
2497 * The OnFailure= dependency between units is now augmented with a
2498 implicit reverse dependency OnFailureOf= (this new dependency cannot
2499 be configured directly it's only created as effect of an OnFailure=
2500 dependency in the reverse order — it's visible in "systemctl show"
2501 however). Similar, Slice= now has an reverse dependency SliceOf=,
2502 that is also not configurable directly, but useful to determine all
2503 units that are members of a slice.
2505 * A pair of new dependency types between units PropagatesStopTo= +
2506 StopPropagatedFrom= has been added, that allows propagation of unit
2507 stop events between two units. It operates similar to the existing
2508 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
2510 * A new dependency type OnSuccess= has been added (plus the reverse
2511 dependency OnSuccessOf=, which cannot be configured directly, but
2512 exists only as effect of the reverse OnSuccess=). It is similar to
2513 OnFailure=, but triggers in the opposite case: when a service exits
2514 cleanly. This allows "chaining up" of services where one or more
2515 services are started once another service has successfully completed.
2517 * A new dependency type Upholds= has been added (plus the reverse
2518 dependency UpheldBy=, which cannot be configured directly, but exists
2519 only as effect of Upholds=). This dependency type is a stronger form
2520 of Wants=: if a unit has an UpHolds= dependency on some other unit
2521 and the former is active then the latter is started whenever it is
2522 found inactive (and no job is queued for it). This is an alternative
2523 to Restart= inside service units, but less configurable, and the
2524 request to uphold a unit is not encoded in the unit itself but in
2525 another unit that intends to uphold it.
2527 * The systemd-ask-password tool now also supports reading passwords
2528 from the credentials subsystem, via the new --credential= switch.
2530 * The systemd-ask-password tool learnt a new switch --emoji= which may
2531 be used to explicit control whether the lock and key emoji (🔐) is
2532 shown in the password prompt on suitable TTYs.
2534 * The --echo switch of systemd-ask-password now optionally takes a
2535 parameter that controls character echo. It may either show asterisks
2536 (default, as before), turn echo off entirely, or echo the typed
2537 characters literally.
2539 * The systemd-ask-password tool also gained a new -n switch for
2540 suppressing output of a trailing newline character when writing the
2541 acquired password to standard output, similar to /bin/echo's -n
2544 * New documentation has been added that describes the organization of
2545 the systemd source code tree:
2547 https://systemd.io/ARCHITECTURE
2549 * Units using ConditionNeedsUpdate= will no longer be activated in
2552 * It is now possible to list a template unit in the WantedBy= or
2553 RequiredBy= settings of the [Install] section of another template
2554 unit, which will be instantiated using the same instance name.
2556 * A new MemoryAvailable property is available for units. If the unit,
2557 or the slices it is part of, have a memory limit set via MemoryMax=/
2558 MemoryHigh=, MemoryAvailable will indicate how much more memory the
2559 unit can claim before hitting the limits.
2561 * systemd-coredump will now try to stay below the cgroup memory limit
2562 placed on itself or one of the slices it runs under, if the storage
2563 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
2564 since files written on such filesystems count toward the cgroup memory
2565 limit. If there is not enough available memory in such cases to store
2566 the core file uncompressed, systemd-coredump will skip to compressed
2567 storage directly (if enabled) and it will avoid analyzing the core file
2568 to print backtrace and metadata in the journal.
2570 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
2571 of a path matches the configured expectations, and remove it if not.
2573 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
2574 specify which of the several available filesystem timestamps (access
2575 time, birth time, change time, modification time) to look at when
2576 deciding whether a path has aged enough to be cleaned.
2578 * A new IPv6StableSecretAddress= setting has been added to .network
2579 files, which takes an IPv6 address to use as secret for IPv6 address
2582 * The [DHCPServer] logic in .network files gained support for a new
2583 UplinkInterface= setting that permits configuration of the uplink
2584 interface name to propagate DHCP lease information from.
2586 * The WakeOnLan= setting in .link files now accepts a list of flags
2587 instead of a single one, to configure multiple wake-on-LAN policies.
2589 * User-space defined tracepoints (USDT) have been added to udev at
2590 strategic locations. This is useful for tracing udev behaviour and
2591 performance with bpftrace and similar tools.
2593 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
2594 setting a network timeout time.
2596 * If a system service is running in a new mount namespace (RootDirectory=
2597 and friends), all file systems will be mounted with MS_NOSUID by
2598 default, unless the system is running with SELinux enabled.
2600 * When enumerating time zones the timedatectl tool will now consult the
2601 'tzdata.zi' file shipped by the IANA time zone database package, in
2602 addition to 'zone1970.tab', as before. This makes sure time zone
2603 aliases are now correctly supported. Some distributions so far did
2604 not install this additional file, most do however. If you
2605 distribution does not install it yet, it might make sense to change
2608 * Intel HID rfkill event is no longer masked, since it's the only
2609 source of rfkill event on newer HP laptops. To have both backward and
2610 forward compatibility, userspace daemon needs to debounce duplicated
2611 events in a short time window.
2613 Contributions from: Aakash Singh, adrian5, Albert Brox,
2614 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
2615 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
2616 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
2617 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
2618 Christian Hesse, Daniel Schaefer, Dan Streetman,
2619 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
2620 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
2621 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
2622 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
2623 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
2624 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
2625 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
2626 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
2627 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
2628 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
2629 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
2630 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
2631 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
2632 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
2633 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
2634 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
2635 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
2636 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
2637 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
2638 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
2639 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
2640 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
2641 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
2642 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
2643 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
2645 — Edinburgh, 2021-07-07
2649 * A concept of system extension images is introduced. Such images may
2650 be used to extend the /usr/ and /opt/ directory hierarchies at
2651 runtime with additional files (even if the file system is read-only).
2652 When a system extension image is activated, its /usr/ and /opt/
2653 hierarchies and os-release information are combined via overlayfs
2654 with the file system hierarchy of the host OS.
2656 A new systemd-sysext tool can be used to merge, unmerge, list, and
2657 refresh system extension hierarchies. See
2658 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
2660 The systemd-sysext.service automatically merges installed system
2661 extensions during boot (before basic.target, but not in very early
2662 boot, since various file systems have to be mounted first).
2664 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
2665 supported system extension level.
2667 * A new ExtensionImages= unit setting can be used to apply the same
2668 system extension image concept from systemd-sysext to the namespaced
2669 file hierarchy of specific services, following the same rules and
2672 * Support for a new special "root=tmpfs" kernel command-line option has
2673 been added. When specified, a tmpfs is mounted on /, and mount.usr=
2674 should be used to point to the operating system implementation.
2676 * A new configuration file /etc/veritytab may be used to configure
2677 dm-verity integrity protection for block devices. Each line is in the
2678 format "volume-name data-device hash-device roothash options",
2679 similar to /etc/crypttab.
2681 * A new kernel command-line option systemd.verity.root_options= may be
2682 used to configure dm-verity behaviour for the root device.
2684 * The key file specified in /etc/crypttab (the third field) may now
2685 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
2686 acquired by connecting to that socket and reading from it. This
2687 allows the implementation of a service to provide key information
2688 dynamically, at the moment when it is needed.
2690 * When the hostname is set explicitly to "localhost", systemd-hostnamed
2691 will respect this. Previously such a setting would be mostly silently
2692 ignored. The goal is to honour configuration as specified by the
2695 * The fallback hostname that will be used by the system manager and
2696 systemd-hostnamed can now be configured in two new ways: by setting
2697 DEFAULT_HOSTNAME= in os-release(5), or by setting
2698 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
2699 also be configured during compilation. The environment variable is
2700 intended for testing and local overrides, the os-release(5) field is
2701 intended to allow customization by different variants of a
2702 distribution that share the same compiled packages.
2704 * The environment block of the manager itself may be configured through
2705 a new ManagerEnvironment= setting in system.conf or user.conf. This
2706 complements existing ways to set the environment block (the kernel
2707 command line for the system manager, the inherited environment and
2708 user@.service unit file settings for the user manager).
2710 * systemd-hostnamed now exports the default hostname and the source of
2711 the configured hostname ("static", "transient", or "default") as
2714 * systemd-hostnamed now exports the "HardwareVendor" and
2715 "HardwareModel" D-Bus properties, which are supposed to contain a
2716 pair of cleaned up, human readable strings describing the system's
2717 vendor and model. It's typically sourced from the firmware's DMI
2718 tables, but may be augmented from a new hwdb database. hostnamectl
2719 shows this in the status output.
2721 * Support has been added to systemd-cryptsetup for extracting the
2722 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
2723 metadata header. This allows the information how to open the
2724 encrypted device to be embedded directly in the device and obviates
2725 the need for configuration in an external file.
2727 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
2728 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
2729 pre-existing support for PKCS#11 security tokens).
2731 * systemd-repart may enroll encrypted partitions using TPM2
2732 hardware. This may be useful for example to create an encrypted /var
2733 partition bound to the machine on first boot.
2735 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
2736 and PKCS#11 security tokens to LUKS volumes, list and destroy
2739 http://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
2741 It also supports enrolling "recovery keys" and regular passphrases.
2743 * The libfido2 dependency is now based on dlopen(), so that the library
2744 is used at runtime when installed, but is not a hard runtime
2747 * systemd-cryptsetup gained support for two new options in
2748 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
2749 request synchronous processing of encryption/decryption IO.
2751 * The manager may be configured at compile time to use the fexecve()
2752 instead of the execve() system call when spawning processes. Using
2753 fexecve() closes a window between checking the security context of an
2754 executable and spawning it, but unfortunately the kernel displays
2755 stale information in the process' "comm" field, which impacts ps
2758 * The configuration option -Dcompat-gateway-hostname has been dropped.
2759 "_gateway" is now the only supported name.
2761 * The ConditionSecurity=tpm2 unit file setting may be used to check if
2762 the system has at least one TPM2 (tpmrm class) device.
2764 * A new ConditionCPUFeature= has been added that may be used to
2765 conditionalize units based on CPU features. For example,
2766 ConditionCPUFeature=rdrand will condition a unit so that it is only
2767 run when the system CPU supports the RDRAND opcode.
2769 * The existing ConditionControlGroupController= setting has been
2770 extended with two new values "v1" and "v2". "v2" means that the
2771 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
2772 hierarchy or the hybrid hierarchy are used.
2774 * A new PrivateIPC= setting on a unit file allows executed processes to
2775 be moved into a private IPC namespace, with separate System V IPC
2776 identifiers and POSIX message queues.
2778 A new IPCNamespacePath= allows the unit to be joined to an existing
2781 * The tables of system calls in seccomp filters are now automatically
2782 generated from kernel lists exported on
2783 https://fedora.juszkiewicz.com.pl/syscalls.html.
2785 The following architectures should now have complete lists:
2786 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
2787 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
2789 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
2790 on /run/ if it is not already a mount point. A writable /run/ has
2791 always been a requirement for a functioning system, but this was not
2792 guaranteed when using a read-only image.
2794 Users can always specify BindPaths= or InaccessiblePaths= as
2795 overrides, and they will take precedence. If the host's root mount
2796 point is used, there is no change in behaviour.
2798 * New bind mounts and file system image mounts may be injected into the
2799 mount namespace of a service (without restarting it). This is exposed
2800 respectively as 'systemctl bind <unit> <path>…' and
2801 'systemctl mount-image <unit> <image>…'.
2803 * The StandardOutput= and StandardError= settings can now specify files
2804 to be truncated for output (as "truncate:<path>").
2806 * The ExecPaths= and NoExecPaths= settings may be used to specify
2807 noexec for parts of the file system.
2809 * sd-bus has a new function sd_bus_open_user_machine() to open a
2810 connection to the session bus of a specific user in a local container
2811 or on the local host. This is exposed in the existing -M switch to
2812 systemctl and similar tools:
2814 systemctl --user -M lennart@foobar start foo
2816 This will connect to the user bus of a user "lennart" in container
2817 "foobar". If no container name is specified, the specified user on
2818 the host itself is connected to
2820 systemctl --user -M lennart@ start quux
2822 * sd-bus also gained a convenience function sd_bus_message_send() to
2823 simplify invocations of sd_bus_send(), taking only a single
2824 parameter: the message to send.
2826 * sd-event allows rate limits to be set on event sources, for dealing
2827 with high-priority event sources that might starve out others. See
2828 the new man page sd_event_source_set_ratelimit(3) for details.
2830 * systemd.link files gained a [Link] Promiscuous= switch, which allows
2831 the device to be raised in promiscuous mode.
2833 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
2834 number of TX and RX queues to be configured.
2836 New [Link] TransmitQueueLength= setting allows the size of the TX
2837 queue to be configured.
2839 New [Link] GenericSegmentOffloadMaxBytes= and
2840 GenericSegmentOffloadMaxSegments= allow capping the packet size and
2841 the number of segments accepted in Generic Segment Offload.
2843 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
2844 wireless routing protocol that operates on ISO/OSI Layer 2 only and
2845 uses ethernet frames to route/bridge packets. This encompasses a new
2846 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
2847 new settings in .netdev files, and a new BatmanAdvanced= setting in
2850 * systemd.network files gained a [Network] RouteTable= configuration
2851 switch to select the routing policy table.
2853 systemd.network files gained a [RoutingPolicyRule] Type=
2854 configuration switch (one of "blackhole, "unreachable", "prohibit").
2856 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
2857 RouteAllowList= settings to ignore/accept route advertisements from
2858 routers matching specified prefixes. The DenyList= setting has been
2859 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
2862 systemd.network files gained a [DHCPv6] UseAddress= setting to
2863 optionally ignore the address provided in the lease.
2865 systemd.network files gained a [DHCPv6PrefixDelegation]
2866 ManageTemporaryAddress= switch.
2868 systemd.network files gained a new ActivationPolicy= setting which
2869 allows configuring how the UP state of an interface shall be managed,
2870 i.e. whether the interface is always upped, always downed, or may be
2871 upped/downed by the user using "ip link set dev".
2873 * The default for the Broadcast= setting in .network files has slightly
2874 changed: the broadcast address will not be configured for wireguard
2877 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
2878 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
2879 configuration options for VLAN packet handling.
2881 * udev rules may now set log_level= option. This allows debug logs to
2882 be enabled for select events, e.g. just for a specific subsystem or
2883 even a single device.
2885 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
2886 DATA_PREPARED_ID properties for block devices with ISO9660 file
2889 * udev now exports decoded DMI information about installed memory slots
2890 as device properties under the /sys/class/dmi/id/ pseudo device.
2892 * /dev/ is not mounted noexec anymore. This didn't provide any
2893 significant security benefits and would conflict with the executable
2894 mappings used with /dev/sgx device nodes. The previous behaviour can
2895 be restored for individual services with NoExecPaths=/dev (or by allow-
2896 listing and excluding /dev from ExecPaths=).
2898 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
2899 and /dev/vhost-net are owned by the kvm group.
2901 * The hardware database has been extended with a list of fingerprint
2902 readers that correctly support USB auto-suspend using data from
2905 * systemd-resolved can now answer DNSSEC questions through the stub
2906 resolver interface in a way that allows local clients to do DNSSEC
2907 validation themselves. For a question with DO+CD set, it'll proxy the
2908 DNS query and respond with a mostly unmodified packet received from
2909 the upstream server.
2911 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
2912 resolved.conf. If true the service will provide caching even for DNS
2913 lookups made to an upstream DNS server on the 127.0.0.1/::1
2914 addresses. By default (and when the option is false) systemd-resolved
2915 will not cache such lookups, in order to avoid duplicate local
2916 caching, under the assumption the local upstream server caches
2919 * systemd-resolved now implements RFC5001 NSID in its local DNS
2920 stub. This may be used by local clients to determine whether they are
2921 talking to the DNS resolver stub or a different DNS server.
2923 * When resolving host names and other records resolvectl will now
2924 report where the data was acquired from (i.e. the local cache, the
2925 network, locally synthesized, …) and whether the network traffic it
2926 effected was encrypted or not. Moreover the tool acquired a number of
2927 new options --cache=, --synthesize=, --network=, --zone=,
2928 --trust-anchor=, --validate= that take booleans and may be used to
2929 tweak a lookup, i.e. whether it may be answered from cached
2930 information, locally synthesized information, information acquired
2931 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
2932 anchor, and whether DNSSEC validation shall be executed for the
2935 * systemd-nspawn gained a new --ambient-capability= setting
2936 (AmbientCapability= in .nspawn files) to configure ambient
2937 capabilities passed to the container payload.
2939 * systemd-nspawn gained the ability to configure the firewall using the
2940 nftables subsystem (in addition to the existing iptables
2941 support). Similarly, systemd-networkd's IPMasquerade= option now
2942 supports nftables as back-end, too. In both cases NAT on IPv6 is now
2943 supported too, in addition to IPv4 (the iptables back-end still is
2946 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
2947 retains its meaning, but has been deprecated. Please switch to either
2948 "ivp4" or "both" (if covering IPv6 is desired).
2950 * systemd-importd will now download .verity and .roothash.p7s files
2951 along with the machine image (as exposed via machinectl pull-raw).
2953 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
2954 setting to configure the time a unit's cgroup needs to exceed memory
2955 pressure limits before action will be taken, and a new
2956 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
2959 systemd-oomd is now considered fully supported (the usual
2960 backwards-compatibility promises apply). Swap is not required for
2961 operation, but it is still recommended.
2963 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
2964 configures the retry delay when trying to contact servers.
2966 * systemd-stdio-bridge gained --system/--user options to connect to the
2967 system bus (previous default) or the user session bus.
2969 * systemd-localed may now call locale-gen to generate missing locales
2970 on-demand (UTF-8-only). This improves integration with Debian-based
2971 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
2973 * systemctl --check-inhibitors=true may now be used to obey inhibitors
2974 even when invoked non-interactively. The old --ignore-inhibitors
2975 switch is now deprecated and replaced by --check-inhibitors=false.
2977 * systemctl import-environment will now emit a warning when called
2978 without any arguments (i.e. to import the full environment block of
2979 the called program). This command will usually be invoked from a
2980 shell, which means that it'll inherit a bunch of variables which are
2981 specific to that shell, and usually to the TTY the shell is connected
2982 to, and don't have any meaning in the global context of the system or
2983 user service manager. Instead, only specific variables should be
2984 imported into the manager environment block.
2986 Similarly, programs which update the manager environment block by
2987 directly calling the D-Bus API of the manager, should also push
2988 specific variables, and not the full inherited environment.
2990 * systemctl's status output now shows unit state with a more careful
2991 choice of Unicode characters: units in maintenance show a "○" symbol
2992 instead of the usual "●", failed units show "×", and services being
2995 * coredumpctl gained a --debugger-arguments= switch to pass arguments
2996 to the debugger. It also gained support for showing coredump info in
2997 a simple JSON format.
2999 * systemctl/loginctl/machinectl's --signal= option now accept a special
3000 value "list", which may be used to show a brief table with known
3001 process signals and their numbers.
3003 * networkctl now shows the link activation policy in status.
3005 * Various tools gained --pager/--no-pager/--json= switches to
3006 enable/disable the pager and provide JSON output.
3008 * Various tools now accept two new values for the SYSTEMD_COLORS
3009 environment variable: "16" and "256", to configure how many terminal
3010 colors are used in output.
3012 * less 568 or newer is now required for the auto-paging logic of the
3013 various tools. Hyperlink ANSI sequences in terminal output are now
3014 used even if a pager is used, and older versions of less are not able
3015 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
3016 disable this output again.
3018 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
3019 builds, non-merged-usr builds) are now officially deprecated. A
3020 warning is emitted during build. Support is slated to be removed in
3021 about a year (when the Debian Bookworm release development starts).
3023 * Systems with the legacy cgroup v1 hierarchy are now marked as
3024 "tainted", to make it clearer that using the legacy hierarchy is not
3027 * systemd-localed will now refuse to configure a keymap which is not
3028 installed in the file system. This is intended as a bug fix, but
3029 could break cases where systemd-localed was used to configure the
3030 keymap in advanced of it being installed. It is necessary to install
3031 the keymap file first.
3033 * The main git development branch has been renamed to 'main'.
3035 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
3036 for partitions, as in the vast majority of cases they contain none
3037 and are used internally by the bootloader (eg: uboot).
3039 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
3040 spawned processes to the PID of the process itself. This may be used
3041 by programs for detecting whether they were forked off by the service
3042 manager itself or are a process forked off further down the tree.
3044 * The sd-device API gained four new calls: sd_device_get_action() to
3045 determine the uevent add/remove/change/… action the device object has
3046 been seen for, sd_device_get_seqno() to determine the uevent sequence
3047 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
3048 object from stat(2) data of a device node, and sd_device_trigger() to
3049 write to the 'uevent' attribute of a device.
3051 * For most tools the --no-legend= switch has been replaced by
3052 --legend=no and --legend=yes, to force whether tables are shown with
3055 * Units acquired a new property "Markers" that takes a list of zero,
3056 one or two of the following strings: "needs-reload" and
3057 "needs-restart". These markers may be set via "systemctl
3058 set-property". Once a marker is set, "systemctl reload-or-restart
3059 --marked" may be invoked to execute the operation the units are
3060 marked for. This is useful for package managers that want to mark
3061 units for restart/reload while updating, but effect the actual
3062 operations at a later step at once.
3064 * The sd_bus_message_read_strv() API call of sd-bus may now also be
3065 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
3068 * bootctl will now report whether the UEFI firmware used a TPM2 device
3069 and measured the boot process into it.
3071 * systemd-tmpfiles learnt support for a new environment variable
3072 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
3073 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
3074 even if the root fs of the system is not itself a btrfs volume.
3076 * systemd-detect-virt/ConditionVirtualization= will now explicitly
3077 detect Docker/Podman environments where possible. Moreover, they
3078 should be able to generically detect any container manager as long as
3079 it assigns the container a cgroup.
3081 * portablectl gained a new "reattach" verb for detaching/reattaching a
3082 portable service image, useful for updating images on-the-fly.
3084 * Intel SGX enclave device nodes (which expose a security feature of
3085 newer Intel CPUs) will now be owned by a new system group "sgx".
3087 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
3088 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
3089 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
3090 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
3091 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
3092 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
3093 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
3094 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
3095 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
3096 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
3097 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
3098 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
3099 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
3100 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
3101 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
3102 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
3103 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
3104 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
3105 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
3106 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
3107 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
3108 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
3109 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
3110 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
3111 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
3112 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
3113 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
3114 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
3115 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
3116 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
3117 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
3118 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
3119 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
3120 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
3121 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
3122 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
3123 Zmicer Turok, Дамјан Георгиевски
3125 — Berlin, 2021-03-30
3129 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
3130 "bind" and "unbind" to the Linux device model. When this kernel
3131 change was made, systemd-udevd was only minimally updated to handle
3132 and propagate these new event types. The introduction of these new
3133 uevents (which are typically generated for USB devices and devices
3134 needing a firmware upload before being functional) resulted in a
3135 number of issues which we so far didn't address. We hoped the kernel
3136 maintainers would themselves address these issues in some form, but
3137 that did not happen. To handle them properly, many (if not most) udev
3138 rules files shipped in various packages need updating, and so do many
3139 programs that monitor or enumerate devices with libudev or sd-device,
3140 or otherwise process uevents. Please note that this incompatibility
3141 is not fault of systemd or udev, but caused by an incompatible kernel
3142 change that happened back in Linux 4.14, but is becoming more and
3143 more visible as the new uevents are generated by more kernel drivers.
3145 To minimize issues resulting from this kernel change (but not avoid
3146 them entirely) starting with systemd-udevd 247 the udev "tags"
3147 concept (which is a concept for marking and filtering devices during
3148 enumeration and monitoring) has been reworked: udev tags are now
3149 "sticky", meaning that once a tag is assigned to a device it will not
3150 be removed from the device again until the device itself is removed
3151 (i.e. unplugged). This makes sure that any application monitoring
3152 devices that match a specific tag is guaranteed to both see uevents
3153 where the device starts being relevant, and those where it stops
3154 being relevant (the latter now regularly happening due to the new
3155 "unbind" uevent type). The udev tags concept is hence now a concept
3156 tied to a *device* instead of a device *event* — unlike for example
3157 udev properties whose lifecycle (as before) is generally tied to a
3158 device event, meaning that the previously determined properties are
3159 forgotten whenever a new uevent is processed.
3161 With the newly redefined udev tags concept, sometimes it's necessary
3162 to determine which tags are the ones applied by the most recent
3163 uevent/database update, in order to discern them from those
3164 originating from earlier uevents/database updates of the same
3165 device. To accommodate for this a new automatic property CURRENT_TAGS
3166 has been added that works similar to the existing TAGS property but
3167 only lists tags set by the most recent uevent/database
3168 update. Similarly, the libudev/sd-device API has been updated with
3169 new functions to enumerate these 'current' tags, in addition to the
3170 existing APIs that now enumerate the 'sticky' ones.
3172 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
3173 essential that all udev rules files and applications are updated to
3174 handle the new events. Specifically:
3176 • All rule files that currently use a header guard similar to
3177 ACTION!="add|change",GOTO="xyz_end" should be updated to use
3178 ACTION=="remove",GOTO="xyz_end" instead, so that the
3179 properties/tags they add are also applied whenever "bind" (or
3180 "unbind") is seen. (This is most important for all physical device
3181 types — those for which "bind" and "unbind" are currently
3182 generated, for all other device types this change is still
3183 recommended but not as important — but certainly prepares for
3184 future kernel uevent type additions).
3186 • Similarly, all code monitoring devices that contains an 'if' branch
3187 discerning the "add" + "change" uevent actions from all other
3188 uevents actions (i.e. considering devices only relevant after "add"
3189 or "change", and irrelevant on all other events) should be reworked
3190 to instead negatively check for "remove" only (i.e. considering
3191 devices relevant after all event types, except for "remove", which
3192 invalidates the device). Note that this also means that devices
3193 should be considered relevant on "unbind", even though conceptually
3194 this — in some form — invalidates the device. Since the precise
3195 effect of "unbind" is not generically defined, devices should be
3196 considered relevant even after "unbind", however I/O errors
3197 accessing the device should then be handled gracefully.
3199 • Any code that uses device tags for deciding whether a device is
3200 relevant or not most likely needs to be updated to use the new
3201 udev_device_has_current_tag() API (or sd_device_has_current_tag()
3202 in case sd-device is used), to check whether the tag is set at the
3203 moment an uevent is seen (as opposed to the existing
3204 udev_device_has_tag() API which checks if the tag ever existed on
3205 the device, following the API concept redefinition explained
3208 We are very sorry for this breakage and the requirement to update
3209 packages using these interfaces. We'd again like to underline that
3210 this is not caused by systemd/udev changes, but result of a kernel
3213 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
3214 packages have not retriggered devices once the udev package (or any
3215 auxiliary package installing additional udev rules) is updated. We
3216 intend to work with major distributions to change this, so that
3217 "udevadm trigger -a change" is issued on such upgrades, ensuring that
3218 the updated ruleset is applied to the devices already discovered, so
3219 that (asynchronously) after the upgrade completed the udev database
3220 is consistent with the updated rule set. This means udev rules must
3221 be ready to be retriggered with a "change" action any time, and
3222 result in correct and complete udev database entries. While the
3223 majority of udev rule files known to us currently get this right,
3224 some don't. Specifically, there are udev rules files included in
3225 various packages that only set udev properties on the "add" action,
3226 but do not handle the "change" action. If a device matching those
3227 rules is retriggered with the "change" action (as is intended here)
3228 it would suddenly lose the relevant properties. This always has been
3229 problematic, but as soon as all udev devices are triggered on relevant
3230 package upgrades this will become particularly so. It is strongly
3231 recommended to fix offending rules so that they can handle a "change"
3232 action at any time, and acquire all necessary udev properties even
3233 then. Or in other words: the header guard mentioned above
3234 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
3235 this, as it makes sure rules are rerun on "change" correctly, and
3236 accumulate the correct and complete set of udev properties. udev rule
3237 definitions that cannot handle "change" events being triggered at
3238 arbitrary times should be considered buggy.
3240 * The MountAPIVFS= service file setting now defaults to on if
3241 RootImage= and RootDirectory= are used, which means that with those
3242 two settings /proc/, /sys/ and /dev/ are automatically properly set
3243 up for services. Previous behaviour may be restored by explicitly
3244 setting MountAPIVFS=off.
3246 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
3247 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
3248 latter it takes precedence over the former, similar to how most of
3249 systemd's own configuration is handled. Given that PAM stack
3250 definitions are primarily put together by OS vendors/distributions
3251 (though possibly overridden by users), this systemd release moves its
3252 own PAM stack configuration for the "systemd-user" PAM service (i.e.
3253 for the PAM session invoked by the per-user user@.service instance)
3254 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
3255 packages' vendor versions of their PAM stack definitions from
3256 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
3257 desired the location to which systemd installs its PAM stack
3258 configuration may be changed via the -Dpamconfdir Meson option.
3260 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
3261 libpwquality and libcryptsetup have been changed to be based on
3262 dlopen(): instead of regular dynamic library dependencies declared in
3263 the binary ELF headers, these libraries are now loaded on demand
3264 only, if they are available. If the libraries cannot be found the
3265 relevant operations will fail gracefully, or a suitable fallback
3266 logic is chosen. This is supposed to be useful for general purpose
3267 distributions, as it allows minimizing the list of dependencies the
3268 systemd packages pull in, permitting building of more minimal OS
3269 images, while still making use of these "weak" dependencies should
3270 they be installed. Since many package managers automatically
3271 synthesize package dependencies from ELF shared library dependencies,
3272 some additional manual packaging work has to be done now to replace
3273 those (slightly downgraded from "required" to "recommended" or
3274 whatever is conceptually suitable for the package manager). Note that
3275 this change does not alter build-time behaviour: as before the
3276 build-time dependencies have to be installed during build, even if
3277 they now are optional during runtime.
3279 * sd-event.h gained a new call sd_event_add_time_relative() for
3280 installing timers relative to the current time. This is mostly a
3281 convenience wrapper around the pre-existing sd_event_add_time() call
3282 which installs absolute timers.
3284 * sd-event event sources may now be placed in a new "exit-on-failure"
3285 mode, which may be controlled via the new
3286 sd_event_source_get_exit_on_failure() and
3287 sd_event_source_set_exit_on_failure() functions. If enabled, any
3288 failure returned by the event source handler functions will result in
3289 exiting the event loop (unlike the default behaviour of just
3290 disabling the event source but continuing with the event loop). This
3291 feature is useful to set for all event sources that define "primary"
3292 program behaviour (where failure should be fatal) in contrast to
3293 "auxiliary" behaviour (where failure should remain local).
3295 * Most event source types sd-event supports now accept a NULL handler
3296 function, in which case the event loop is exited once the event
3297 source is to be dispatched, using the userdata pointer — converted to
3298 a signed integer — as exit code of the event loop. Previously this
3299 was supported for IO and signal event sources already. Exit event
3300 sources still do not support this (simply because it makes little
3301 sense there, as the event loop is already exiting when they are
3304 * A new per-unit setting RootImageOptions= has been added which allows
3305 tweaking the mount options for any file system mounted as effect of
3306 the RootImage= setting.
3308 * Another new per-unit setting MountImages= has been added, that allows
3309 mounting additional disk images into the file system tree accessible
3312 * Timer units gained a new FixedRandomDelay= boolean setting. If
3313 enabled, the random delay configured with RandomizedDelaySec= is
3314 selected in a way that is stable on a given system (though still
3315 different for different units).
3317 * Socket units gained a new setting Timestamping= that takes "us", "ns"
3318 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
3321 * systemd-repart now generates JSON output when requested with the new
3324 * systemd-machined's OpenMachineShell() bus call will now pass
3325 additional policy metadata data fields to the PolicyKit
3326 authentication request.
3328 * systemd-tmpfiles gained a new -E switch, which is equivalent to
3329 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
3330 --exclude=/sys. It's particularly useful in combination with --root=,
3331 when operating on OS trees that do not have any of these four runtime
3332 directories mounted, as this means no files below these subtrees are
3333 created or modified, since those mount points should probably remain
3336 * systemd-tmpfiles gained a new --image= switch which is like --root=,
3337 but takes a disk image instead of a directory as argument. The
3338 specified disk image is mounted inside a temporary mount namespace
3339 and the tmpfiles.d/ drop-ins stored in the image are executed and
3340 applied to the image. systemd-sysusers similarly gained a new
3341 --image= switch, that allows the sysusers.d/ drop-ins stored in the
3342 image to be applied onto the image.
3344 * Similarly, the journalctl command also gained an --image= switch,
3345 which is a quick one-step solution to look at the log data included
3348 * journalctl's --output=cat option (which outputs the log content
3349 without any metadata, just the pure text messages) will now make use
3350 of terminal colors when run on a suitable terminal, similarly to the
3353 * JSON group records now support a "description" string that may be
3354 used to add a human-readable textual description to such groups. This
3355 is supposed to match the user's GECOS field which traditionally
3356 didn't have a counterpart for group records.
3358 * The "systemd-dissect" tool that may be used to inspect OS disk images
3359 and that was previously installed to /usr/lib/systemd/ has now been
3360 moved to /usr/bin/, reflecting its updated status of an officially
3361 supported tool with a stable interface. It gained support for a new
3362 --mkdir switch which when combined with --mount has the effect of
3363 creating the directory to mount the image to if it is missing
3364 first. It also gained two new commands --copy-from and --copy-to for
3365 copying files and directories in and out of an OS image without the
3366 need to manually mount it. It also acquired support for a new option
3367 --json= to generate JSON output when inspecting an OS image.
3369 * The cgroup2 file system is now mounted with the
3370 "memory_recursiveprot" mount option, supported since kernel 5.7. This
3371 means that the MemoryLow= and MemoryMin= unit file settings now apply
3372 recursively to whole subtrees.
3374 * systemd-homed now defaults to using the btrfs file system — if
3375 available — when creating home directories in LUKS volumes. This may
3376 be changed with the DefaultFileSystemType= setting in homed.conf.
3377 It's now the default file system in various major distributions and
3378 has the major benefit for homed that it can be grown and shrunk while
3379 mounted, unlike the other contenders ext4 and xfs, which can both be
3380 grown online, but not shrunk (in fact xfs is the technically most
3381 limited option here, as it cannot be shrunk at all).
3383 * JSON user records managed by systemd-homed gained support for
3384 "recovery keys". These are basically secondary passphrases that can
3385 unlock user accounts/home directories. They are computer-generated
3386 rather than user-chosen, and typically have greater entropy.
3387 homectl's --recovery-key= option may be used to add a recovery key to
3388 a user account. The generated recovery key is displayed as a QR code,
3389 so that it can be scanned to be kept in a safe place. This feature is
3390 particularly useful in combination with systemd-homed's support for
3391 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
3392 security tokens are lost. Recovery keys may be entered wherever the
3393 system asks for a password.
3395 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
3396 home directory which indicates that a home directory has not been
3397 deactivated cleanly when offline. This flag is useful to identify
3398 home directories for which the offline discard logic did not run when
3399 offlining, and where it would be a good idea to log in again to catch
3402 * systemctl gained a new parameter --timestamp= which may be used to
3403 change the style in which timestamps are output, i.e. whether to show
3404 them in local timezone or UTC, or whether to show µs granularity.
3406 * Alibaba's "pouch" container manager is now detected by
3407 systemd-detect-virt, ConditionVirtualization= and similar
3408 constructs. Similar, they now also recognize IBM PowerVM machine
3411 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
3412 place to use for propagating external mounts into the
3413 container. Similarly /run/host/notify is now used as the socket path
3414 for container payloads to communicate with the container manager
3415 using sd_notify(). The container manager now uses the
3416 /run/host/inaccessible/ directory to place "inaccessible" file nodes
3417 of all relevant types which may be used by the container payload as
3418 bind mount source to over-mount inodes to make them inaccessible.
3419 /run/host/container-manager will now be initialized with the same
3420 string as the $container environment variable passed to the
3421 container's PID 1. /run/host/container-uuid will be initialized with
3422 the same string as $container_uuid. This means the /run/host/
3423 hierarchy is now the primary way to make host resources available to
3424 the container. The Container Interface documents these new files and
3427 https://systemd.io/CONTAINER_INTERFACE
3429 * Support for the "ConditionNull=" unit file condition has been
3430 deprecated and undocumented for 6 years. systemd started to warn
3431 about its use 1.5 years ago. It has now been removed entirely.
3433 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
3434 a sd_bus_error struct and a list of error names, and checks if the
3435 error matches one of these names. It's a convenience wrapper that is
3436 useful in cases where multiple errors shall be handled the same way.
3438 * A new system call filter list "@known" has been added, that contains
3439 all system calls known at the time systemd was built.
3441 * Behaviour of system call filter allow lists has changed slightly:
3442 system calls that are contained in @known will result in EPERM by
3443 default, while those not contained in it result in ENOSYS. This
3444 should improve compatibility because known system calls will thus be
3445 communicated as prohibited, while unknown (and thus newer ones) will
3446 be communicated as not implemented, which hopefully has the greatest
3447 chance of triggering the right fallback code paths in client
3450 * "systemd-analyze syscall-filter" will now show two separate sections
3451 at the bottom of the output: system calls known during systemd build
3452 time but not included in any of the filter groups shown above, and
3453 system calls defined on the local kernel but known during systemd
3456 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
3457 systemd-nspawn all system call filter violations will be logged by
3458 the kernel (audit). This is useful for tracking down system calls
3459 invoked by container payloads that are prohibited by the container's
3460 system call filter policy.
3462 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
3463 systemd-nspawn (and other programs that use seccomp) all seccomp
3464 filtering is turned off.
3466 * Two new unit file settings ProtectProc= and ProcSubset= have been
3467 added that expose the hidepid= and subset= mount options of procfs.
3468 All processes of the unit will only see processes in /proc that are
3469 are owned by the unit's user. This is an important new sandboxing
3470 option that is recommended to be set on all system services. All
3471 long-running system services that are included in systemd itself set
3472 this option now. This option is only supported on kernel 5.8 and
3473 above, since the hidepid= option supported on older kernels was not a
3474 per-mount option but actually applied to the whole PID namespace.
3476 * Socket units gained a new boolean setting FlushPending=. If enabled
3477 all pending socket data/connections are flushed whenever the socket
3478 unit enters the "listening" state, i.e. after the associated service
3481 * The unit file setting NUMAMask= gained a new "all" value: when used,
3482 all existing NUMA nodes are added to the NUMA mask.
3484 * A new "credentials" logic has been added to system services. This is
3485 a simple mechanism to pass privileged data to services in a safe and
3486 secure way. It's supposed to be used to pass per-service secret data
3487 such as passwords or cryptographic keys but also associated less
3488 private information such as user names, certificates, and similar to
3489 system services. Each credential is identified by a short user-chosen
3490 name and may contain arbitrary binary data. Two new unit file
3491 settings have been added: SetCredential= and LoadCredential=. The
3492 former allows setting a credential to a literal string, the latter
3493 sets a credential to the contents of a file (or data read from a
3494 user-chosen AF_UNIX stream socket). Credentials are passed to the
3495 service via a special credentials directory, one file for each
3496 credential. The path to the credentials directory is passed in a new
3497 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
3498 are passed in the file system they may be easily referenced in
3499 ExecStart= command lines too, thus no explicit support for the
3500 credentials logic in daemons is required (though ideally daemons
3501 would look for the bits they need in $CREDENTIALS_DIRECTORY
3502 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
3503 backed by unswappable memory if privileges allow it, immutable if
3504 privileges allow it, is accessible only to the service's UID, and is
3505 automatically destroyed when the service stops.
3507 * systemd-nspawn supports the same credentials logic. It can both
3508 consume credentials passed to it via the aforementioned
3509 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
3510 to its payload. The service manager/PID 1 has been updated to match
3511 this: it can also accept credentials from the container manager that
3512 invokes it (in fact: any process that invokes it), and passes them on
3513 to its services. Thus, credentials can be propagated recursively down
3514 the tree: from a system's service manager to a systemd-nspawn
3515 service, to the service manager that runs as container payload and to
3516 the service it runs below. Credentials may also be added on the
3517 systemd-nspawn command line, using new --set-credential= and
3518 --load-credential= command line switches that match the
3519 aforementioned service settings.
3521 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
3522 the partition drop-ins which may be used to format/LUKS
3523 encrypt/populate any created partitions. The partitions are
3524 encrypted/formatted/populated before they are registered in the
3525 partition table, so that they appear atomically: either the
3526 partitions do not exist yet or they exist fully encrypted, formatted,
3527 and populated — there is no time window where they are
3528 "half-initialized". Thus the system is robust to abrupt shutdown: if
3529 the tool is terminated half-way during its operations on next boot it
3530 will start from the beginning.
3532 * systemd-repart's --size= operation gained a new "auto" value. If
3533 specified, and operating on a loopback file it is automatically sized
3534 to the minimal size the size constraints permit. This is useful to
3535 use "systemd-repart" as an image builder for minimally sized images.
3537 * systemd-resolved now gained a third IPC interface for requesting name
3538 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
3539 interface is now supported. The nss-resolve NSS module has been
3540 modified to use this new interface instead of D-Bus. Using Varlink
3541 has a major benefit over D-Bus: it works without a broker service,
3542 and thus already during earliest boot, before the dbus daemon has
3543 been started. This means name resolution via systemd-resolved now
3544 works at the same time systemd-networkd operates: from earliest boot
3545 on, including in the initrd.
3547 * systemd-resolved gained support for a new DNSStubListenerExtra=
3548 configuration file setting which may be used to specify additional IP
3549 addresses the built-in DNS stub shall listen on, in addition to the
3550 main one on 127.0.0.53:53.
3552 * Name lookups issued via systemd-resolved's D-Bus and Varlink
3553 interfaces (and thus also via glibc NSS if nss-resolve is used) will
3554 now honour a trailing dot in the hostname: if specified the search
3555 path logic is turned off. Thus "resolvectl query foo." is now
3556 equivalent to "resolvectl query --search=off foo.".
3558 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
3559 exposes how /etc/resolv.conf is currently managed: by resolved (and
3560 in which mode if so) or another subsystem. "resolvctl" will display
3561 this property in its status output.
3563 * The resolv.conf snippets systemd-resolved provides will now set "."
3564 as the search domain if no other search domain is known. This turns
3565 off the derivation of an implicit search domain by nss-dns for the
3566 hostname, when the hostname is set to an FQDN. This change is done to
3567 make nss-dns using resolv.conf provided by systemd-resolved behave
3568 more similarly to nss-resolve.
3570 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
3571 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
3572 "birth" time (btime) of a file in addition to the atime, mtime, and
3575 * systemd-analyze gained a new verb "capability" that lists all known
3576 capabilities by the systemd build and by the kernel.
3578 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
3579 advance the system clock to it at boot if it is noticed to be before
3580 that time. Previously, PID 1 would only advance the time to an epoch
3581 time that is set during build-time. With this new file OS builders
3582 can change this epoch timestamp on individual OS images without
3583 having to rebuild systemd.
3585 * systemd-logind will now listen to the KEY_RESTART key from the Linux
3586 input layer and reboot the system if it is pressed, similarly to how
3587 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
3588 was originally defined in the Multimedia context (to restart playback
3589 of a song or film), but is now primarily used in various embedded
3590 devices for "Reboot" buttons. Accordingly, systemd-logind will now
3591 honour it as such. This may configured in more detail via the new
3592 HandleRebootKey= and RebootKeyIgnoreInhibited=.
3594 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
3595 copying OS trees, for example in "systemd-nspawn --ephemeral",
3596 "systemd-nspawn --template=", "machinectl clone" and similar. This is
3597 useful when operating with OSTree images, which use hardlinks heavily
3598 throughout, and where such copies previously resulting in "exploding"
3601 * systemd-nspawn's --console= setting gained support for a new
3602 "autopipe" value, which is identical to "interactive" when invoked on
3603 a TTY, and "pipe" otherwise.
3605 * systemd-networkd's .network files gained support for explicitly
3606 configuring the multicast membership entries of bridge devices in the
3607 [BridgeMDB] section. It also gained support for the PIE queuing
3608 discipline in the [FlowQueuePIE] sections.
3610 * systemd-networkd's .netdev files may now be used to create "BareUDP"
3611 tunnels, configured in the new [BareUDP] setting.
3613 * systemd-networkd's Gateway= setting in .network files now accepts the
3614 special values "_dhcp4" and "_ipv6ra" to configure additional,
3615 locally defined, explicit routes to the gateway acquired via DHCP or
3616 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
3617 but still accepted for backwards compatibility.
3619 * systemd-networkd's [IPv6PrefixDelegation] section and
3620 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
3621 IPv6SendRA= (the old names are still accepted for backwards
3624 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
3625 boolean setting in [Network] section. If enabled, the delegated prefix
3626 gained by another link will be configured, and an address within the
3627 prefix will be assigned.
3629 * systemd-networkd's .network files gained the Announce= boolean setting
3630 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
3631 prefix will be announced through IPv6 router advertisement (IPv6 RA).
3632 The setting is enabled by default.
3634 * VXLAN tunnels may now be marked as independent of any underlying
3635 network interface via the new Independent= boolean setting.
3637 * systemctl gained support for two new verbs: "service-log-level" and
3638 "service-log-target" may be used on services that implement the
3639 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
3640 adjust the log level and target. All of systemd's long-running
3641 services support this now, but ideally all system services would
3642 implement this interface to make the system more uniformly
3645 * The SystemCallErrorNumber= unit file setting now accepts the new
3646 "kill" and "log" actions, in addition to arbitrary error number
3647 specifications as before. If "kill" the processes are killed on the
3648 event, if "log" the offending system call is audit logged.
3650 * A new SystemCallLog= unit file setting has been added that accepts a
3651 list of system calls that shall be logged about (audit).
3653 * The OS image dissection logic (as used by RootImage= in unit files or
3654 systemd-nspawn's --image= switch) has gained support for identifying
3655 and mounting explicit /usr/ partitions, which are now defined in the
3656 discoverable partition specification. This should be useful for
3657 environments where the root file system is
3658 generated/formatted/populated dynamically on first boot and combined
3659 with an immutable /usr/ tree that is supplied by the vendor.
3661 * In the final phase of shutdown, within the systemd-shutdown binary
3662 we'll now try to detach MD devices (i.e software RAID) in addition to
3663 loopback block devices and DM devices as before. This is supposed to
3664 be a safety net only, in order to increase robustness if things go
3665 wrong. Storage subsystems are expected to properly detach their
3666 storage volumes during regular shutdown already (or in case of
3667 storage backing the root file system: in the initrd hook we return to
3670 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
3671 will now log the thread ID in their log output. This is useful when
3672 working with heavily threaded programs.
3674 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
3675 not use the RDRAND CPU instruction. This is useful in environments
3676 such as replay debuggers where non-deterministic behaviour is not
3679 * The autopaging logic in systemd's various tools (such as systemctl)
3680 has been updated to turn on "secure" mode in "less"
3681 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
3682 detected. This disables invoking external programs from the pager,
3683 via the pipe logic. This behaviour may be overridden via the new
3684 $SYSTEMD_PAGERSECURE environment variable.
3686 * Units which have resource limits (.service, .mount, .swap, .slice,
3687 .socket, and .slice) gained new configuration settings
3688 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
3689 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
3690 limits and optional action taken by systemd-oomd.
3692 * A new service systemd-oomd has been added. It monitors resource
3693 contention for selected parts of the unit hierarchy using the PSI
3694 information reported by the kernel, and kills processes when memory
3695 or swap pressure is above configured limits. This service is only
3696 enabled by default in developer mode (see below) and should be
3697 considered a preview in this release. Behaviour details and option
3698 names are subject to change without the usual backwards-compatibility
3701 * A new helper oomctl has been added to introspect systemd-oomd state.
3702 It is only enabled by default in developer mode and should be
3703 considered a preview without the usual backwards-compatibility
3706 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
3707 enabled, systemd reads the system UID boundaries from /etc/login.defs
3708 at runtime, instead of using the built-in values selected during
3709 build. This is an option to improve compatibility for upgrades from
3710 old systems. It's strongly recommended not to make use of this
3711 functionality on new systems (or even enable it during build), as it
3712 makes something runtime-configurable that is mostly an implementation
3713 detail of the OS, and permits avoidable differences in deployments
3714 that create all kinds of problems in the long run.
3716 * New meson option '-Dmode=developer|release' has been added. When
3717 'developer', additional checks and features are enabled that are
3718 relevant during upstream development, e.g. verification that
3719 semi-automatically-generated documentation has been properly updated
3720 following API changes. Those checks are considered hints for
3721 developers and are not actionable in downstream builds. In addition,
3722 extra features that are not ready for general consumption may be
3723 enabled in developer mode. It is thus recommended to set
3724 '-Dmode=release' in end-user and distro builds.
3726 * systemd-cryptsetup gained support for processing detached LUKS
3727 headers specified on the kernel command line via the header=
3728 parameter of the luks.options= kernel command line option. The same
3729 device/path syntax as for key files is supported for header files
3732 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
3733 slot index data for devices that are connected through a PCI bridge
3734 where the _SUN index is associated with the bridge instead of the
3735 network device itself. Previously this would create ambiguous device
3736 naming if multiple network interfaces were connected to the same PCI
3737 bridge. Since this is a naming scheme incompatibility on systems that
3738 possess hardware like this it has been introduced as new naming
3739 scheme "v247". The previous scheme can be selected via the
3740 "net.naming-scheme=v245" kernel command line parameter.
3742 * ConditionFirstBoot= semantics have been modified to be safe towards
3743 abnormal system power-off during first boot. Specifically, the
3744 "systemd-machine-id-commit.service" service now acts as boot
3745 milestone indicating when the first boot process is sufficiently
3746 complete in order to not consider the next following boot also a
3747 first boot. If the system is reset before this unit is reached the
3748 first time, the next boot will still be considered a first boot; once
3749 it has been reached, no further boots will be considered a first
3750 boot. The "first-boot-complete.target" unit now acts as official hook
3751 point to order against this. If a service shall be run on every boot
3752 until the first boot fully succeeds it may thus be ordered before
3753 this target unit (and pull it in) and carry ConditionFirstBoot=
3756 * bootctl's set-default and set-oneshot commands now accept the three
3757 special strings "@default", "@oneshot", "@current" in place of a boot
3758 entry id. These strings are resolved to the current default and
3759 oneshot boot loader entry, as well as the currently booted one. Thus
3760 a command "bootctl set-default @current" may be used to make the
3761 currently boot menu item the new default for all subsequent boots.
3763 * "systemctl edit" has been updated to show the original effective unit
3764 contents in commented form in the text editor.
3766 * Units in user mode are now segregated into three new slices:
3767 session.slice (units that form the core of graphical session),
3768 app.slice ("normal" user applications), and background.slice
3769 (low-priority tasks). Unless otherwise configured, user units are
3770 placed in app.slice. The plan is to add resource limits and
3771 protections for the different slices in the future.
3773 * New GPT partition types for RISCV32/64 for the root and /usr
3774 partitions, and their associated Verity partitions have been defined,
3775 and are now understood by systemd-gpt-auto-generator, and the OS
3776 image dissection logic.
3778 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
3779 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
3780 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
3781 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
3782 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
3783 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
3784 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
3785 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
3786 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
3787 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
3788 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
3789 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
3790 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
3791 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
3792 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
3793 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
3794 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
3795 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
3796 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
3797 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
3798 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
3799 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
3800 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
3801 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
3802 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
3803 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
3804 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
3805 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
3806 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
3807 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
3808 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
3809 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
3810 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
3812 – Warsaw, 2020-11-26
3816 * The service manager gained basic support for cgroup v2 freezer. Units
3817 can now be suspended or resumed either using new systemctl verbs,
3818 freeze and thaw respectively, or via D-Bus.
3820 * PID 1 may now automatically load pre-compiled AppArmor policies from
3821 /etc/apparmor/earlypolicy during early boot.
3823 * The CPUAffinity= setting in service unit files now supports a new
3824 special value "numa" that causes the CPU affinity masked to be set
3825 based on the NUMA mask.
3827 * systemd will now log about all left-over processes remaining in a
3828 unit when the unit is stopped. It will now warn about services using
3829 KillMode=none, as this is generally an unsafe thing to make use of.
3831 * Two new unit file settings
3832 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
3833 added. They may be used to check whether a specific file system path
3834 resides on a block device that is encrypted on the block level
3835 (i.e. using dm-crypt/LUKS).
3837 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
3838 has been added that may be used for simple environment checks. This
3839 is particularly useful when passing in environment variables from a
3840 container manager (or from PAM in case of the systemd --user
3843 * .service unit files now accept a new setting CoredumpFilter= which
3844 allows configuration of the memory sections coredumps of the
3845 service's processes shall include.
3847 * .mount units gained a new ReadWriteOnly= boolean option. If set
3848 it will not be attempted to mount a file system read-only if mounting
3849 in read-write mode doesn't succeed. An option x-systemd.rw-only is
3850 available in /etc/fstab to control the same.
3852 * .socket units gained a new boolean setting PassPacketInfo=. If
3853 enabled, the kernel will attach additional per-packet metadata to all
3854 packets read from the socket, as an ancillary message. This controls
3855 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
3856 depending on socket type.
3858 * .service units gained a new setting RootHash= which may be used to
3859 specify the root hash for verity enabled disk images which are
3860 specified in RootImage=. RootVerity= may be used to specify a path to
3861 the Verity data matching a RootImage= file system. (The latter is
3862 only useful for images that do not contain the Verity data embedded
3863 into the same image that carries a GPT partition table following the
3864 Discoverable Partition Specification). Similarly, systemd-nspawn
3865 gained a new switch --verity-data= that takes a path to a file with
3866 the verity data of the disk image supplied in --image=, if the image
3867 doesn't contain the verity data itself.
3869 * .service units gained a new setting RootHashSignature= which takes
3870 either a base64 encoded PKCS#7 signature of the root hash specified
3871 with RootHash=, or a path to a file to read the signature from. This
3872 allows validation of the root hash against public keys available in
3873 the kernel keyring, and is only supported on recent kernels
3874 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
3875 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
3876 this mechanism has also been added to systemd-veritysetup.
3878 * .service unit files gained two new options
3879 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
3880 tune behaviour if a start or stop timeout is hit, i.e. whether to
3881 terminate the service with SIGTERM, SIGABRT or SIGKILL.
3883 * Most options in systemd that accept hexadecimal values prefixed with
3884 0x in additional to the usual decimal notation now also support octal
3885 notation when the 0o prefix is used and binary notation if the 0b
3888 * Various command line parameters and configuration file settings that
3889 configure key or certificate files now optionally take paths to
3890 AF_UNIX sockets in the file system. If configured that way a stream
3891 connection is made to the socket and the required data read from
3892 it. This is a simple and natural extension to the existing regular
3893 file logic, and permits other software to provide keys or
3894 certificates via simple IPC services, for example when unencrypted
3895 storage on disk is not desired. Specifically, systemd-networkd's
3896 Wireguard and MACSEC key file settings as well as
3897 systemd-journal-gatewayd's and systemd-journal-remote's PEM
3898 key/certificate parameters support this now.
3900 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
3901 configuration files that support specifier expansion learnt six new
3902 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
3903 resolve to the various ID fields from /etc/os-release, %l resolves to
3904 the "short" hostname of the system, i.e. the hostname configured in
3905 the kernel truncated at the first dot.
3907 * Support for the .include syntax in unit files has been removed. The
3908 concept has been obsolete for 6 years and we started warning about
3909 its pending removal 2 years ago (also see NEWS file below). It's
3912 * StandardError= and StandardOutput= in unit files no longer support
3913 the "syslog" and "syslog-console" switches. They were long removed
3914 from the documentation, but will now result in warnings when used,
3915 and be converted to "journal" and "journal+console" automatically.
3917 * If the service setting User= is set to the "nobody" user, a warning
3918 message is now written to the logs (but the value is nonetheless
3919 accepted). Setting User=nobody is unsafe, since the primary purpose
3920 of the "nobody" user is to own all files whose owner cannot be mapped
3921 locally. It's in particular used by the NFS subsystem and in user
3922 namespacing. By running a service under this user's UID it might get
3923 read and even write access to all these otherwise unmappable files,
3924 which is quite likely a major security problem.
3926 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
3927 and others) now have a size and inode limits applied (50% of RAM for
3928 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
3929 that the implicit kernel default is 50% too, so there is no change
3930 in the size limit for /tmp and /dev/shm.
3932 * nss-mymachines lost support for resolution of users and groups, and
3933 now only does resolution of hostnames. This functionality is now
3934 provided by nss-systemd. Thus, the 'mymachines' entry should be
3935 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
3936 (and 'systemd' added if it is not already there).
3938 * A new kernel command line option systemd.hostname= has been added
3939 that allows controlling the hostname that is initialized early during
3942 * A kernel command line option "udev.blockdev_read_only" has been
3943 added. If specified all hardware block devices that show up are
3944 immediately marked as read-only by udev. This option is useful for
3945 making sure that a specific boot under no circumstances modifies data
3946 on disk. Use "blockdev --setrw" to undo the effect of this, per
3949 * A new boolean kernel command line option systemd.swap= has been
3950 added, which may be used to turn off automatic activation of swap
3951 devices listed in /etc/fstab.
3953 * New kernel command line options systemd.condition-needs-update= and
3954 systemd.condition-first-boot= have been added, which override the
3955 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
3958 * A new kernel command line option systemd.clock-usec= has been added
3959 that allows setting the system clock to the specified time in µs
3960 since Jan 1st, 1970 early during boot. This is in particular useful
3961 in order to make test cases more reliable.
3963 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
3964 systemd-coredump to save core files for suid processes. When saving
3965 the core file, systemd-coredump will use the effective uid and gid of
3966 the process that faulted.
3968 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
3969 now automatically set to "Y" at boot, in order to enable pstore
3970 generation for collection with systemd-pstore.
3972 * We provide a set of udev rules to enable auto-suspend on PCI and USB
3973 devices that were tested to correctly support it. Previously, this
3974 was distributed as a set of udev rules, but has now been replaced by
3975 by a set of hwdb entries (and a much shorter udev rule to take action
3976 if the device modalias matches one of the new hwdb entries).
3978 As before, entries are periodically imported from the database
3979 maintained by the ChromiumOS project. If you have a device that
3980 supports auto-suspend correctly and where it should be enabled by
3981 default, please submit a patch that adds it to the database (see
3982 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
3984 * systemd-udevd gained the new configuration option timeout_signal= as well
3985 as a corresponding kernel command line option udev.timeout_signal=.
3986 The option can be used to configure the UNIX signal that the main
3987 daemon sends to the worker processes on timeout. Setting the signal
3988 to SIGABRT is useful for debugging.
3990 * .link files managed by systemd-udevd gained options RxFlowControl=,
3991 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
3992 order to configure various flow control parameters. They also gained
3993 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
3994 frame ring buffer sizes.
3996 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
3997 enabled systemd-networkd manages all routes configured by other tools.
3999 * .network files managed by systemd-networkd gained a new section
4000 [SR-IOV], in order to configure SR-IOV capable network devices.
4002 * systemd-networkd's [IPv6Prefix] section in .network files gained a
4003 new boolean setting Assign=. If enabled an address from the prefix is
4004 automatically assigned to the interface.
4006 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
4007 controls delegated prefixes assigned by DHCPv6 client. The section
4008 has three settings: SubnetID=, Assign=, and Token=. The setting
4009 SubnetID= allows explicit configuration of the preferred subnet that
4010 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
4011 Assign= is enabled (which is the default) an address from any acquired
4012 delegated prefix is automatically chosen and assigned to the
4013 interface. The setting Token= specifies an optional address generation
4016 * systemd-networkd's [Network] section gained a new setting
4017 IPv4AcceptLocal=. If enabled the interface accepts packets with local
4020 * systemd-networkd gained support for configuring the HTB queuing
4021 discipline in the [HierarchyTokenBucket] and
4022 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
4023 be configured in the [PFIFO] section, "GRED" in
4024 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
4025 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
4026 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
4027 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
4028 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
4029 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
4031 * systemd-networkd gained support for a new Termination= setting in the
4032 [CAN] section for configuring the termination resistor. It also
4033 gained a new ListenOnly= setting for controlling whether to only
4034 listen on CAN interfaces, without interfering with traffic otherwise
4035 (which is useful for debugging/monitoring CAN network
4036 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
4037 been added to configure various CAN-FD aspects.
4039 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
4040 When enabled, DHCPv6 will be attempted right-away without requiring an
4041 Router Advertisement packet suggesting it first (i.e. without the 'M'
4042 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
4043 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
4044 the RA packets suggest it.
4046 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
4047 which may be used to turn off use of the gateway information provided
4048 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
4049 used to configure how to process leases that lack a lifetime option.
4051 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
4052 setting SendVendorOption= allowing configuration of additional vendor
4053 options to send in the DHCP requests/responses. The [DHCPv6] section
4054 gained a new SendOption= setting for sending arbitrary DHCP
4055 options. RequestOptions= has been added to request arbitrary options
4056 from the server. UserClass= has been added to set the DHCP user class
4059 * systemd-networkd's [DHCPServer] section gained a new set of options
4060 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
4061 information about these three protocols in the DHCP lease. It also
4062 gained support for including "MUD" URLs ("Manufacturer Usage
4063 Description"). Support for "MUD" URLs was also added to the LLDP
4064 stack, configurable in the [LLDP] section in .network files.
4066 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
4067 mode. Also, the sections now support a new setting SourceMACAddress=.
4069 * systemd-networkd's .netdev files now support a new setting
4070 VLANProtocol= in the [Bridge] section that allows configuration of
4071 the VLAN protocol to use.
4073 * systemd-networkd supports a new Group= setting in the [Link] section
4074 of the .network files, to control the link group.
4076 * systemd-networkd's [Network] section gained a new
4077 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
4078 link local address is generated.
4080 * A new default .network file is now shipped that matches TUN/TAP
4081 devices that begin with "vt-" in their name. Such interfaces will
4082 have IP routing onto the host links set up automatically. This is
4083 supposed to be used by VM managers to trivially acquire a network
4084 interface which is fully set up for host communication, simply by
4085 carefully picking an interface name to use.
4087 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
4088 which sets the route priority for routes specified by the DHCP server.
4090 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
4091 which configures the vendor class information sent to DHCP server.
4093 * The BlackList= settings in .network files' [DHCPv4] and
4094 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
4095 are still understood to provide compatibility.
4097 * networkctl gained the new "forcerenew" command for forcing all DHCP
4098 server clients to renew their lease. The interface "status" output
4099 will now show numerous additional fields of information about an
4100 interface. There are new "up" and "down" commands to bring specific
4101 interfaces up or down.
4103 * systemd-resolved's DNS= configuration option now optionally accepts a
4104 port number (after ":") and a host name (after "#"). When the host
4105 name is specified, the DNS-over-TLS certificate is validated to match
4106 the specified hostname. Additionally, in case of IPv6 addresses, an
4107 interface may be specified (after "%").
4109 * systemd-resolved may be configured to forward single-label DNS names.
4110 This is not standard-conformant, but may make sense in setups where
4111 public DNS servers are not used.
4113 * systemd-resolved's DNS-over-TLS support gained SNI validation.
4115 * systemd-nspawn's --resolv-conf= switch gained a number of new
4116 supported values. Specifically, options starting with "replace-" are
4117 like those prefixed "copy-" but replace any existing resolv.conf
4118 file. And options ending in "-uplink" and "-stub" can now be used to
4119 propagate other flavours of resolv.conf into the container (as
4120 defined by systemd-resolved).
4122 * The various programs included in systemd can now optionally output
4123 their log messages on stderr prefixed with a timestamp, controlled by
4124 the $SYSTEMD_LOG_TIME environment variable.
4126 * systemctl gained a new "-P" switch that is a shortcut for "--value
4129 * "systemctl list-units" and "systemctl list-machines" no longer hide
4130 their first output column with --no-legend. To hide the first column,
4133 * "systemctl reboot" takes the option "--reboot-argument=".
4134 The optional positional argument to "systemctl reboot" is now
4135 being deprecated in favor of this option.
4137 * systemd-run gained a new switch --slice-inherit. If specified the
4138 unit it generates is placed in the same slice as the systemd-run
4141 * systemd-journald gained support for zstd compression of large fields
4142 in journal files. The hash tables in journal files have been hardened
4143 against hash collisions. This is an incompatible change and means
4144 that journal files created with new systemd versions are not readable
4145 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
4146 environment variable for systemd-journald.service is set to 0 this
4147 new hardening functionality may be turned off, so that generated
4148 journal files remain compatible with older journalctl
4151 * journalctl will now include a clickable link in the default output for
4152 each log message for which an URL with further documentation is
4153 known. This is only supported on terminal emulators that support
4154 clickable hyperlinks, and is turned off if a pager is used (since
4155 "less" still doesn't support hyperlinks,
4156 unfortunately). Documentation URLs may be included in log messages
4157 either by including a DOCUMENTATION= journal field in it, or by
4158 associating a journal message catalog entry with the log message's
4159 MESSAGE_ID, which then carries a "Documentation:" tag.
4161 * journald.conf gained a new boolean setting Audit= that may be used to
4162 control whether systemd-journald will enable audit during
4165 * when systemd-journald's log stream is broken up into multiple lines
4166 because the PID of the sender changed this is indicated in the
4167 generated log records via the _LINE_BREAK=pid-change field.
4169 * journalctl's "-o cat" output mode will now show one or more journal
4170 fields specified with --output-fields= instead of unconditionally
4171 MESSAGE=. This is useful to retrieve a very specific set of fields
4172 without any decoration.
4174 * The sd-journal.h API gained two new functions:
4175 sd_journal_enumerate_available_unique() and
4176 sd_journal_enumerate_available_data() that operate like their
4177 counterparts that lack the _available_ in the name, but skip items
4178 that cannot be read and processed by the local implementation
4179 (i.e. are compressed in an unsupported format or such),
4181 * coredumpctl gained a new --file= switch, matching the same one in
4182 journalctl: a specific journal file may be specified to read the
4185 * coredumps collected by systemd-coredump may now be compressed using
4188 * systemd-binfmt gained a new switch --unregister for unregistering all
4189 registered entries at once. This is now invoked automatically at
4190 shutdown, so that binary formats registered with the "F" flag will
4191 not block clean file system unmounting.
4193 * systemd-notify's --pid= switch gained new values: "parent", "self",
4194 "auto" for controlling which PID to send to the service manager: the
4195 systemd-notify process' PID, or the one of the process invoking it.
4197 * systemd-logind's Session bus object learnt a new method call
4198 SetType() for temporarily updating the session type of an already
4199 allocated session. This is useful for upgrading tty sessions to
4200 graphical ones once a compositor is invoked.
4202 * systemd-socket-proxy gained a new switch --exit-idle-time= for
4203 configuring an exit-on-idle time.
4205 * systemd-repart's --empty= setting gained a new value "create". If
4206 specified a new empty regular disk image file is created under the
4207 specified name. Its size may be specified with the new --size=
4208 option. The latter is also supported without the "create" mode, in
4209 order to grow existing disk image files to the specified size. These
4210 two new options are useful when creating or manipulating disk images
4211 instead of operating on actual block devices.
4213 * systemd-repart drop-ins now support a new UUID= setting to control
4214 the UUID to assign to a newly created partition.
4216 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
4219 * systemd-repart's Label= setting now support the usual, simple
4220 specifier expansion.
4222 * systemd-homed's LUKS backend gained the ability to discard empty file
4223 system blocks automatically when the user logs out. This is enabled
4224 by default to ensure that home directories take minimal space when
4225 logged out but get full size guarantees when logged in. This may be
4226 controlled with the new --luks-offline-discard= switch to homectl.
4228 * If systemd-homed detects that /home/ is encrypted as a whole it will
4229 now default to the directory or subvolume backends instead of the
4230 LUKS backend, in order to avoid double encryption. The default
4231 storage and file system may now be configured explicitly, too, via
4232 the new /etc/systemd/homed.conf configuration file.
4234 * systemd-homed now supports unlocking home directories with FIDO2
4235 security tokens that support the 'hmac-secret' extension, in addition
4236 to the existing support for PKCS#11 security token unlocking
4237 support. Note that many recent hardware security tokens support both
4238 interfaces. The FIDO2 support is accessible via homectl's
4239 --fido2-device= option.
4241 * homectl's --pkcs11-uri= setting now accepts two special parameters:
4242 if "auto" is specified and only one suitable PKCS#11 security token
4243 is plugged in, its URL is automatically determined and enrolled for
4244 unlocking the home directory. If "list" is specified a brief table of
4245 suitable PKCS#11 security tokens is shown. Similar, the new
4246 --fido2-device= option also supports these two special values, for
4247 automatically selecting and listing suitable FIDO2 devices.
4249 * The /etc/crypttab tmp option now optionally takes an argument
4250 selecting the file system to use. Moreover, the default is now
4251 changed from ext2 to ext4.
4253 * There's a new /etc/crypttab option "keyfile-erase". If specified the
4254 key file listed in the same line is removed after use, regardless if
4255 volume activation was successful or not. This is useful if the key
4256 file is only acquired transiently at runtime and shall be erased
4257 before the system continues to boot.
4259 * There's also a new /etc/crypttab option "try-empty-password". If
4260 specified, before asking the user for a password it is attempted to
4261 unlock the volume with an empty password. This is useful for
4262 installing encrypted images whose password shall be set on first boot
4263 instead of at installation time.
4265 * systemd-cryptsetup will now attempt to load the keys to unlock
4266 volumes with automatically from files in
4267 /etc/cryptsetup-keys.d/<volume>.key and
4268 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
4270 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
4271 /etc/crypttab, during boot.
4273 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
4274 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
4277 * A new generator systemd-xdg-autostart-generator has been added. It
4278 generates systemd unit files from XDG autostart .desktop files, and
4279 may be used to let the systemd user instance manage services that are
4280 started automatically as part of the desktop session.
4282 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
4283 to query and change the firmware's 'reboot into firmware' setup flag.
4285 * systemd-firstboot gained a new switch --kernel-command-line= that may
4286 be used to initialize the /etc/kernel/cmdline file of the image. It
4287 also gained a new switch --root-password-hashed= which is like
4288 --root-password= but accepts a pre-hashed UNIX password as
4289 argument. The new option --delete-root-password may be used to unset
4290 any password for the root user (dangerous!). The --root-shell= switch
4291 may be used to control the shell to use for the root account. A new
4292 --force option may be used to override any already set settings with
4293 the parameters specified on the command line (by default, the tool
4294 will not override what has already been set before, i.e. is purely
4297 * systemd-firstboot gained support for a new --image= switch, which is
4298 similar to --root= but accepts the path to a disk image file, on
4299 which it then operates.
4301 * A new sd-path.h API has been added to libsystemd. It provides a
4302 simple API for retrieving various search paths and primary
4303 directories for various resources.
4305 * A new call sd_notify_barrier() has been added to the sd-daemon.h
4306 API. The call will block until all previously sent sd_notify()
4307 messages have been processed by the service manager. This is useful
4308 to remove races caused by a process already having disappeared at the
4309 time a notification message is processed by the service manager,
4310 making correct attribution impossible. The systemd-notify tool will
4311 now make use of this call implicitly, but this can be turned off again
4312 via the new --no-block switch.
4314 * When sending a file descriptor (fd) to the service manager to keep
4315 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
4316 may be specified. If passed the service manager will refrain from
4317 poll()ing on the file descriptor. Traditionally (and when the
4318 parameter is not specified), the service manager will poll it for
4319 POLLHUP or POLLERR events, and immediately close the fds in that
4322 * The service manager (PID1) gained a new D-Bus method call
4323 SetShowStatus() which may be used to control whether it shall show
4324 boot-time status output on the console. This method has a similar
4325 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
4327 * The sd-bus API gained a number of convenience functions that take
4328 va_list arguments rather than "...". For example, there's now
4329 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
4330 it easier to build wrappers that accept variadic arguments and want
4331 to pass a ready va_list structure to sd-bus.
4333 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
4334 flag which alters how the userdata pointer to pass to the callbacks
4335 is determined. When the flag is set, the offset field is converted
4336 as-is into a pointer, without adding it to the object pointer the
4337 vtable is associated with.
4339 * sd-bus now exposes four new functions:
4340 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
4341 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
4342 validate strings to check if they qualify as various D-Bus concepts.
4344 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
4345 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
4346 that simplify adding argument names to D-Bus methods and signals.
4348 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
4350 * Various D-Bus APIs of systemd daemons now have man pages that
4351 document the methods, signals and properties.
4353 * The expectations on user/group name syntax are now documented in
4354 detail; documentation on how classic home directories may be
4355 converted into home directories managed by homed has been added;
4356 documentation regarding integration of homed/userdb functionality in
4357 desktops has been added:
4359 https://systemd.io/USER_NAMES
4360 https://systemd.io/CONVERTING_TO_HOMED
4361 https://systemd.io/USERDB_AND_DESKTOPS
4363 * Documentation for the on-disk Journal file format has been updated
4364 and has now moved to:
4366 https://systemd.io/JOURNAL_FILE_FORMAT
4368 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
4369 has been extended by a set of environment variables that expose
4370 select fields from the host's os-release file to the container
4371 payload. Similarly, host's os-release files can be mounted into the
4372 container underneath /run/host. Together, those mechanisms provide a
4373 standardized way to expose information about the host to the
4374 container payload. Both interfaces are implemented in systemd-nspawn.
4376 * All D-Bus services shipped in systemd now implement the generic
4377 LogControl1 D-Bus API which allows clients to change log level +
4378 target of the service during runtime.
4380 * Only relevant for developers: the mkosi.default symlink has been
4381 dropped from version control. Please create a symlink to one of the
4382 distribution-specific defaults in .mkosi/ based on your preference.
4384 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
4385 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
4386 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
4387 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
4388 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
4389 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
4390 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
4391 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
4392 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
4393 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
4394 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
4395 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
4396 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
4397 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
4398 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
4399 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
4400 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
4401 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
4402 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
4403 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
4404 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
4405 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
4406 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
4407 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
4408 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
4409 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
4410 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
4411 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
4412 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
4413 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
4414 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
4415 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
4416 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
4417 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
4418 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
4419 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
4420 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
4421 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
4423 – Warsaw, 2020-07-30
4427 * A new tool "systemd-repart" has been added, that operates as an
4428 idempotent declarative repartitioner for GPT partition tables.
4429 Specifically, a set of partitions that must or may exist can be
4430 configured via drop-in files, and during every boot the partition
4431 table on disk is compared with these files, creating missing
4432 partitions or growing existing ones based on configurable relative
4433 and absolute size constraints. The tool is strictly incremental,
4434 i.e. does not delete, shrink or move partitions, but only adds and
4435 grows them. The primary use-case is OS images that ship in minimized
4436 form, that on first boot are grown to the size of the underlying
4437 block device or augmented with additional partitions. For example,
4438 the root partition could be extended to cover the whole disk, or a
4439 swap or /home partitions could be added on first boot. It can also be
4440 used for systems that use an A/B update scheme but ship images with
4441 just the A partition, with B added on first boot. The tool is
4442 primarily intended to be run in the initrd, shortly before
4443 transitioning into the host OS, but can also be run after the
4444 transition took place. It automatically discovers the disk backing
4445 the root file system, and should hence not require any additional
4446 configuration besides the partition definition drop-ins. If no
4447 configuration drop-ins are present, no action is taken.
4449 * A new component "userdb" has been added, along with a small daemon
4450 "systemd-userdbd.service" and a client tool "userdbctl". The framework
4451 allows defining rich user and group records in a JSON format,
4452 extending on the classic "struct passwd" and "struct group"
4453 structures. Various components in systemd have been updated to
4454 process records in this format, including systemd-logind and
4455 pam-systemd. The user records are intended to be extensible, and
4456 allow setting various resource management, security and runtime
4457 parameters that shall be applied to processes and sessions of the
4458 user as they log in. This facility is intended to allow associating
4459 such metadata directly with user/group records so that they can be
4460 produced, extended and consumed in unified form. We hope that
4461 eventually frameworks such as sssd will generate records this way, so
4462 that for the first time resource management and various other
4463 per-user settings can be configured in LDAP directories and then
4464 provided to systemd (specifically to systemd-logind and pam-system)
4465 to apply on login. For further details see:
4467 https://systemd.io/USER_RECORD
4468 https://systemd.io/GROUP_RECORD
4469 https://systemd.io/USER_GROUP_API
4471 * A small new service systemd-homed.service has been added, that may be
4472 used to securely manage home directories with built-in encryption.
4473 The complete user record data is unified with the home directory,
4474 thus making home directories naturally migratable. Its primary
4475 back-end is based on LUKS volumes, but fscrypt, plain directories,
4476 and other storage schemes are also supported. This solves a couple of
4477 problems we saw with traditional ways to manage home directories, in
4478 particular when it comes to encryption. For further discussion of
4479 this, see the video of Lennart's talk at AllSystemsGo! 2019:
4481 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
4483 For further details about the format and expectations on home
4484 directories this new daemon makes, see:
4486 https://systemd.io/HOME_DIRECTORY
4488 * systemd-journald is now multi-instantiable. In addition to the main
4489 instance systemd-journald.service there's now a template unit
4490 systemd-journald@.service, with each instance defining a new named
4491 log 'namespace' (whose name is specified via the instance part of the
4492 unit name). A new unit file setting LogNamespace= has been added,
4493 taking such a namespace name, that assigns services to the specified
4494 log namespaces. As each log namespace is serviced by its own
4495 independent journal daemon, this functionality may be used to improve
4496 performance and increase isolation of applications, at the price of
4497 losing global message ordering. Each instance of journald has a
4498 separate set of configuration files, with possibly different disk
4499 usage limitations and other settings.
4501 journalctl now takes a new option --namespace= to show logs from a
4502 specific log namespace. The sd-journal.h API gained
4503 sd_journal_open_namespace() for opening the log stream of a specific
4504 log namespace. systemd-journald also gained the ability to exit on
4505 idle, which is useful in the context of log namespaces, as this means
4506 log daemons for log namespaces can be activated automatically on
4507 demand and will stop automatically when no longer used, minimizing
4510 * When systemd-tmpfiles copies a file tree using the 'C' line type it
4511 will now label every copied file according to the SELinux database.
4513 * When systemd/PID 1 detects it is used in the initrd it will now boot
4514 into initrd.target rather than default.target by default. This should
4515 make it simpler to build initrds with systemd as for many cases the
4516 only difference between a host OS image and an initrd image now is
4517 the presence of the /etc/initrd-release file.
4519 * A new kernel command line option systemd.cpu_affinity= is now
4520 understood. It's equivalent to the CPUAffinity= option in
4521 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
4522 itself and the default for all other processes.
4524 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
4525 equivalent), the SELinux database is now reloaded, ensuring that
4526 sockets and other file system objects are generated taking the new
4527 database into account.
4529 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
4530 "quiet" has been changed to imply that instead of
4531 "systemd.show-status=auto". In this mode, only messages about errors
4532 and significant delays in boot are shown on the console.
4534 * The sd-event.h API gained native support for the new Linux "pidfd"
4535 concept. This permits watching processes using file descriptors
4536 instead of PID numbers, which fixes a number of races and makes
4537 process supervision more robust and efficient. All of systemd's
4538 components will now use pidfds if the kernel supports it for process
4539 watching, with the exception of PID 1 itself, unfortunately. We hope
4540 to move PID 1 to exclusively using pidfds too eventually, but this
4541 requires some more kernel work first. (Background: PID 1 watches
4542 processes using waitid() with the P_ALL flag, and that does not play
4543 together nicely with pidfds yet.)
4545 * Closely related to this, the sd-event.h API gained two new calls
4546 sd_event_source_send_child_signal() (for sending a signal to a
4547 watched process) and sd_event_source_get_child_process_own() (for
4548 marking a process so that it is killed automatically whenever the
4549 event source watching it is freed).
4551 * systemd-networkd gained support for configuring Token Bucket Filter
4552 (TBF) parameters in its qdisc configuration support. Similarly,
4553 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
4554 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
4556 * systemd-networkd gained support for Intermediate Functional Block
4557 (IFB) network devices.
4559 * systemd-networkd gained support for configuring multi-path IP routes,
4560 using the new MultiPathRoute= setting in the [Route] section.
4562 * systemd-networkd's DHCPv4 client has been updated to support a new
4563 SendDecline= option. If enabled, duplicate address detection is done
4564 after a DHCP offer is received from the server. If a conflict is
4565 detected, the address is declined. The DHCPv4 client also gained
4566 support for a new RouteMTUBytes= setting that allows to configure the
4567 MTU size to be used for routes generated from DHCPv4 leases.
4569 * The PrefixRoute= setting in systemd-networkd's [Address] section of
4570 .network files has been deprecated, and replaced by AddPrefixRoute=,
4571 with its sense inverted.
4573 * The Gateway= setting of [Route] sections of .network files gained
4574 support for a special new value "_dhcp". If set, the configured
4575 static route uses the gateway host configured via DHCP.
4577 * New User= and SuppressPrefixLength= settings have been implemented
4578 for the [RoutingPolicyRule] section of .network files to configure
4579 source routing based on UID ranges and prefix length, respectively.
4581 * The Type= match property of .link files has been generalized to
4582 always match the device type shown by 'networkctl status', even for
4583 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
4586 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
4587 D-Bus message object as "sensitive". Those objects are erased from
4588 memory when they are freed. This concept is intended to be used for
4589 messages that contain security sensitive data. A new flag
4590 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
4591 in sd-bus vtables, causing any incoming and outgoing messages of
4592 those methods to be implicitly marked as "sensitive".
4594 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
4595 contents of a message (or parts thereof) to standard output for
4598 * systemd-sysusers gained support for creating users with the primary
4599 group named differently than the user.
4601 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
4602 gained support for growing XFS partitions. Previously it supported
4603 only ext4 and btrfs partitions.
4605 * The support for /etc/crypttab gained a new x-initrd.attach option. If
4606 set, the specified encrypted volume is unlocked already in the
4607 initrd. This concept corresponds to the x-initrd.mount option in
4610 * systemd-cryptsetup gained native support for unlocking encrypted
4611 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
4612 encryption of volumes to YubiKeys. This is exposed in the new
4613 pkcs11-uri= option in /etc/crypttab.
4615 * The /etc/fstab support in systemd now supports two new mount options
4616 x-systemd.{required,wanted}-by=, for explicitly configuring the units
4617 that the specified mount shall be pulled in by, in place of
4618 the usual local-fs.target/remote-fs.target.
4620 * The https://systemd.io/ web site has been relaunched, directly
4621 populated with most of the documentation included in the systemd
4622 repository. systemd also acquired a new logo, thanks to Tobias
4625 * systemd-udevd gained support for managing "alternative" network
4626 interface names, as supported by new Linux kernels. For the first
4627 time this permits assigning multiple (and longer!) names to a network
4628 interface. systemd-udevd will now by default assign the names
4629 generated via all supported naming schemes to each interface. This
4630 may be further tweaked with .link files and the AlternativeName= and
4631 AlternativeNamesPolicy= settings. Other components of systemd have
4632 been updated to support the new alternative names wherever
4633 appropriate. For example, systemd-nspawn will now generate
4634 alternative interface names for the host-facing side of container
4635 veth links based on the full container name without truncation.
4637 * systemd-nspawn interface naming logic has been updated in another way
4638 too: if the main interface name (i.e. as opposed to new-style
4639 "alternative" names) based on the container name is truncated, a
4640 simple hashing scheme is used to give different interface names to
4641 multiple containers whose names all begin with the same prefix. Since
4642 this changes the primary interface names pointing to containers if
4643 truncation happens, the old scheme may still be requested by
4644 selecting an older naming scheme, via the net.naming-scheme= kernel
4645 command line option.
4647 * PrivateUsers= in service files now works in services run by the
4648 systemd --user per-user instance of the service manager.
4650 * A new per-service sandboxing option ProtectClock= has been added that
4651 locks down write access to the system clock. It takes away device
4652 node access to /dev/rtc as well as the system calls that set the
4653 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
4654 Note that this option does not affect access to auxiliary services
4655 that allow changing the clock, for example access to
4658 * The systemd-id128 tool gained a new "show" verb for listing or
4659 resolving a number of well-known UUIDs/128bit IDs, currently mostly
4660 GPT partition table types.
4662 * The Discoverable Partitions Specification has been updated to support
4663 /var and /var/tmp partition discovery. Support for this has been
4664 added to systemd-gpt-auto-generator. For details see:
4666 https://systemd.io/DISCOVERABLE_PARTITIONS
4668 * "systemctl list-unit-files" has been updated to show a new column
4669 with the suggested enablement state based on the vendor preset files
4670 for the respective units.
4672 * "systemctl" gained a new option "--with-dependencies". If specified
4673 commands such as "systemctl status" or "systemctl cat" will now show
4674 all specified units along with all units they depend on.
4676 * networkctl gained support for showing per-interface logs in its
4679 * systemd-networkd-wait-online gained support for specifying the maximum
4680 operational state to wait for, and to wait for interfaces to
4683 * The [Match] section of .link and .network files now supports a new
4684 option PermanentMACAddress= which may be used to check against the
4685 permanent MAC address of a network device even if a randomized MAC
4688 * The [TrafficControlQueueingDiscipline] section in .network files has
4689 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
4690 dropped from the individual setting names.
4692 * Any .link and .network files that have an empty [Match] section (this
4693 also includes empty and commented-out files) will now be
4694 rejected. systemd-udev and systemd-networkd started warning about
4695 such files in version 243.
4697 * systemd-logind will now validate access to the operation of changing
4698 the virtual terminal via a polkit action. By default, only users
4699 with at least one session on a local VT are granted permission.
4701 * When systemd sets up PAM sessions that invoked service processes
4702 shall run in, the pam_setcred() API is now invoked, thus permitting
4703 PAM modules to set additional credentials for the processes.
4705 * portablectl attach/detach verbs now accept --now and --enable options
4706 to combine attachment with enablement and invocation, or detachment
4707 with stopping and disablement.
4709 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
4710 fixed, which in turn exposed bugs in unit configuration of services
4711 which have Type=oneshot and should only run once, but do not have
4712 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
4713 service may be started again after exiting successfully, for example
4714 as a dependency in another transaction. Affected services included
4715 some internal systemd services (most notably
4716 systemd-vconsole-setup.service, which was updated to have
4717 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
4718 plymouth has been suitably updated or patched before upgrading to
4719 this systemd release. See
4720 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
4721 additional discussion.
4723 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
4724 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
4725 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
4726 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
4727 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
4728 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
4729 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
4730 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
4731 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
4732 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
4733 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
4734 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
4735 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
4736 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
4737 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
4738 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
4739 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
4740 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
4741 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
4742 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
4743 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
4744 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
4745 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
4746 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
4749 – Warsaw, 2020-03-06
4753 * Support for the cpuset cgroups v2 controller has been added.
4754 Processes may be restricted to specific CPUs using the new
4755 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
4756 AllowedMemoryNodes= setting.
4758 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
4759 now be configured using a new RestartKillSignal= setting. This
4760 allows units which signals to request termination to implement
4761 different behaviour when stopping in preparation for a restart.
4763 * "systemctl clean" may now be used also for socket, mount, and swap
4766 * systemd will also read configuration options from the EFI variable
4767 SystemdOptions. This may be used to configure systemd behaviour when
4768 modifying the kernel command line is inconvenient, but configuration
4769 on disk is read too late, for example for the options related to
4770 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
4771 set the EFI variable.
4773 * systemd will now disable printk ratelimits in early boot. This should
4774 allow us to capture more logs from the early boot phase where normal
4775 storage is not available and the kernel ring buffer is used for
4776 logging. Configuration on the kernel command line has higher priority
4777 and overrides the systemd setting.
4779 systemd programs which log to /dev/kmsg directly use internal
4780 ratelimits to prevent runaway logging. (Normally this is only used
4781 during early boot, so in practice this change has very little
4784 * Unit files now support top level dropin directories of the form
4785 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
4786 that affects all corresponding unit files.
4788 * systemctl gained support for 'stop --job-mode=triggering' which will
4789 stop the specified unit and any units which could trigger it.
4791 * Unit status display now includes units triggering and triggered by
4792 the unit being shown.
4794 * The RuntimeMaxSec= setting is now supported by scopes, not just
4795 .service units. This is particularly useful for PAM sessions which
4796 create a scope unit for the user login. systemd.runtime_max_sec=
4797 setting may used with the pam_systemd module to limit the duration
4798 of the PAM session, for example for time-limited logins.
4800 * A new @pkey system call group is now defined to make it easier to
4801 allow-list memory protection syscalls for containers and services
4802 which need to use them.
4804 * systemd-udevd: removed the 30s timeout for killing stale workers on
4805 exit. systemd-udevd now waits for workers to finish. The hard-coded
4806 exit timeout of 30s was too short for some large installations, where
4807 driver initialization could be prematurely interrupted during initrd
4808 processing if the root file system had been mounted and init was
4809 preparing to switch root. If udevd is run without systemd and workers
4810 are hanging while udevd receives an exit signal, udevd will now exit
4811 when udev.event_timeout is reached for the last hanging worker. With
4812 systemd, the exit timeout can additionally be configured using
4813 TimeoutStopSec= in systemd-udevd.service.
4815 * udev now provides a program (fido_id) that identifies FIDO CTAP1
4816 ("U2F")/CTAP2 security tokens based on the usage declared in their
4817 report and descriptor and outputs suitable environment variables.
4818 This replaces the externally maintained allow lists of all known
4819 security tokens that were used previously.
4821 * Automatically generated autosuspend udev rules for allow-listed
4822 devices have been imported from the Chromium OS project. This should
4823 improve power saving with many more devices.
4825 * udev gained a new "CONST{key}=value" setting that allows matching
4826 against system-wide constants without forking a helper binary.
4827 Currently "arch" and "virt" keys are supported.
4829 * udev now opens CDROMs in non-exclusive mode when querying their
4830 capabilities. This should fix issues where other programs trying to
4831 use the CDROM cannot gain access to it, but carries a risk of
4832 interfering with programs writing to the disk, if they did not open
4833 the device in exclusive mode as they should.
4835 * systemd-networkd does not create a default route for IPv4 link local
4836 addressing anymore. The creation of the route was unexpected and was
4837 breaking routing in various cases, but people who rely on it being
4838 created implicitly will need to adjust. Such a route may be requested
4839 with DefaultRouteOnDevice=yes.
4841 Similarly, systemd-networkd will not assign a link-local IPv6 address
4842 when IPv6 link-local routing is not enabled.
4844 * Receive and transmit buffers may now be configured on links with
4845 the new RxBufferSize= and TxBufferSize= settings.
4847 * systemd-networkd may now advertise additional IPv6 routes. A new
4848 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
4851 * systemd-networkd may now configure "next hop" routes using the
4852 [NextHop] section and Gateway= and Id= settings.
4854 * systemd-networkd will now retain DHCP config on restarts by default
4855 (but this may be overridden using the KeepConfiguration= setting).
4856 The default for SendRelease= has been changed to true.
4858 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
4859 received from the server.
4861 The client will use the received SIP server list if UseSIP=yes is
4864 The client may be configured to request specific options from the
4865 server using a new RequestOptions= setting.
4867 The client may be configured to send arbitrary options to the server
4868 using a new SendOption= setting.
4870 A new IPServiceType= setting has been added to configure the "IP
4871 service type" value used by the client.
4873 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
4874 request prefix hints in the DHCPv6 solicitation.
4876 * The DHCPv4 server may be configured to send arbitrary options using
4877 a new SendOption= setting.
4879 * The DHCPv4 server may now be configured to emit SIP server list using
4880 the new EmitSIP= and SIP= settings.
4882 * systemd-networkd and networkctl may now renew DHCP leases on demand.
4883 networkctl has a new 'networkctl renew' verb.
4885 * systemd-networkd may now reconfigure links on demand. networkctl
4886 gained two new verbs: "reload" will reload the configuration, and
4887 "reconfigure DEVICE…" will reconfigure one or more devices.
4889 * .network files may now match on SSID and BSSID of a wireless network,
4890 i.e. the access point name and hardware address using the new SSID=
4891 and BSSID= options. networkctl will display the current SSID and
4892 BSSID for wireless links.
4894 .network files may also match on the wireless network type using the
4895 new WLANInterfaceType= option.
4897 * systemd-networkd now includes default configuration that enables
4898 link-local addressing when connected to an ad-hoc wireless network.
4900 * systemd-networkd may configure the Traffic Control queueing
4901 disciplines in the kernel using the new
4902 [TrafficControlQueueingDiscipline] section and Parent=,
4903 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
4904 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
4905 NetworkEmulatorDuplicateRate= settings.
4907 * systemd-tmpfiles gained a new w+ setting to append to files.
4909 * systemd-analyze dump will now report when the memory configuration in
4910 the kernel does not match what systemd has configured (usually,
4911 because some external program has modified the kernel configuration
4914 * systemd-analyze gained a new --base-time= switch instructs the
4915 'calendar' verb to resolve times relative to that timestamp instead
4916 of the present time.
4918 * journalctl --update-catalog now produces deterministic output (making
4919 reproducible image builds easier).
4921 * A new devicetree-overlay setting is now documented in the Boot Loader
4924 * The default value of the WatchdogSec= setting used in systemd
4925 services (the ones bundled with the project itself) may be set at
4926 configuration time using the -Dservice-watchdog= setting. If set to
4927 empty, the watchdogs will be disabled.
4929 * systemd-resolved validates IP addresses in certificates now when GnuTLS
4932 * libcryptsetup >= 2.0.1 is now required.
4934 * A configuration option -Duser-path= may be used to override the $PATH
4935 used by the user service manager. The default is again to use the same
4936 path as the system manager.
4938 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
4939 outputting the 128bit IDs in UUID format (i.e. in the "canonical
4942 * Service units gained a new sandboxing option ProtectKernelLogs= which
4943 makes sure the program cannot get direct access to the kernel log
4944 buffer anymore, i.e. the syslog() system call (not to be confused
4945 with the API of the same name in libc, which is not affected), the
4946 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
4947 inaccessible to the service. It's recommended to enable this setting
4948 for all services that should not be able to read from or write to the
4949 kernel log buffer, which are probably almost all.
4951 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
4952 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
4953 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
4954 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
4955 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
4956 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
4957 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
4958 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
4959 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
4960 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
4961 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
4962 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
4963 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
4964 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
4965 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
4966 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
4967 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
4968 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
4969 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
4970 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
4971 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
4972 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
4973 Zach Smith, Zbigniew Jędrzejewski-Szmek
4975 – Warsaw, 2019-11-29
4979 * This release enables unprivileged programs (i.e. requiring neither
4980 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
4981 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
4982 kernel for the whole UNIX group range, i.e. all processes. This
4983 change should be reasonably safe, as the kernel support for it was
4984 specifically implemented to allow safe access to ICMP Echo for
4985 processes lacking any privileges. If this is not desirable, it can be
4986 disabled again by setting the parameter to "1 0".
4988 * Previously, filters defined with SystemCallFilter= would have the
4989 effect that any calling of an offending system call would terminate
4990 the calling thread. This behaviour never made much sense, since
4991 killing individual threads of unsuspecting processes is likely to
4992 create more problems than it solves. With this release the default
4993 action changed from killing the thread to killing the whole
4994 process. For this to work correctly both a kernel version (>= 4.14)
4995 and a libseccomp version (>= 2.4.0) supporting this new seccomp
4996 action is required. If an older kernel or libseccomp is used the old
4997 behaviour continues to be used. This change does not affect any
4998 services that have no system call filters defined, or that use
4999 SystemCallErrorNumber= (and thus see EPERM or another error instead
5000 of being killed when calling an offending system call). Note that
5001 systemd documentation always claimed that the whole process is
5002 killed. With this change behaviour is thus adjusted to match the
5005 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
5006 4194304 by default, i.e. the full 22bit range the kernel allows, up
5007 from the old 16bit range. This should improve security and
5008 robustness, as PID collisions are made less likely (though certainly
5009 still possible). There are rumours this might create compatibility
5010 problems, though at this moment no practical ones are known to
5011 us. Downstream distributions are hence advised to undo this change in
5012 their builds if they are concerned about maximum compatibility, but
5013 for everybody else we recommend leaving the value bumped. Besides
5014 improving security and robustness this should also simplify things as
5015 the maximum number of allowed concurrent tasks was previously bounded
5016 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
5017 only a single knob is left ("kernel.threads-max"). There have been
5018 concerns that usability is affected by this change because larger PID
5019 numbers are harder to type, but we believe the change from 5 digits
5020 to 7 digits doesn't hamper usability.
5022 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
5023 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
5024 hierarchically set default memory protection values for a particular
5025 subtree of the unit hierarchy.
5027 * Memory protection directives can now take a value of zero, allowing
5028 explicit opting out of a default value propagated by an ancestor.
5030 * systemd now defaults to the "unified" cgroup hierarchy setup during
5031 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
5032 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
5033 change reflects the fact that cgroupsv2 support has matured
5034 substantially in both systemd and in the kernel, and is clearly the
5035 way forward. Downstream production distributions might want to
5036 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
5037 their builds as unfortunately the popular container managers have not
5038 caught up with the kernel API changes.
5040 * Man pages are not built by default anymore (html pages were already
5041 disabled by default), to make development builds quicker. When
5042 building systemd for a full installation with documentation, meson
5043 should be called with -Dman=true and/or -Dhtml=true as appropriate.
5044 The default was changed based on the assumption that quick one-off or
5045 repeated development builds are much more common than full optimized
5046 builds for installation, and people need to pass various other
5047 options to when doing "proper" builds anyway, so the gain from making
5048 development builds quicker is bigger than the one time disruption for
5051 Two scripts are created in the *build* directory to generate and
5052 preview man and html pages on demand, e.g.:
5054 build/man/man systemctl
5055 build/man/html systemd.index
5057 * libidn2 is used by default if both libidn2 and libidn are installed.
5058 Please use -Dlibidn=true if libidn is preferred.
5060 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
5061 big-endian machines. Before, bytes were written and read in native
5062 machine order as exposed by the native libc __cpu_mask interface.
5063 Now, little-endian order is always used (CPUs 0–7 are described by
5064 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
5065 This change fixes D-Bus calls that cross endianness boundary.
5067 The presentation format used for CPUAffinity= by "systemctl show" and
5068 "systemd-analyze dump" is changed to present CPU indices instead of
5069 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
5070 shown as CPUAffinity=03000000000000000000000000000… (on
5071 little-endian) or CPUAffinity=00000000000000300000000000000… (on
5072 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
5073 input format. The maximum integer that will be printed in the new
5074 format is 8191 (four digits), while the old format always used a very
5075 long number (with the length varying by architecture), so they can be
5076 unambiguously distinguished.
5078 * /usr/sbin/halt.local is no longer supported. Implementation in
5079 distributions was inconsistent and it seems this functionality was
5082 To replace this functionality, users should:
5083 - either define a new unit and make it a dependency of final.target
5084 (systemctl add-wants final.target my-halt-local.service)
5085 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
5086 and ensure that it accepts "halt", "poweroff", "reboot", and
5087 "kexec" as an argument, see the description in systemd-shutdown(8).
5089 * When a [Match] section in .link or .network file is empty (contains
5090 no match patterns), a warning will be emitted. Please add any "match
5091 all" pattern instead, e.g. OriginalName=* or Name=* in case all
5092 interfaces should really be matched.
5094 * A new setting NUMAPolicy= may be used to set process memory
5095 allocation policy. This setting can be specified in
5096 /etc/systemd/system.conf and hence will set the default policy for
5097 PID1. The default policy can be overridden on a per-service
5098 basis. The related setting NUMAMask= is used to specify NUMA node
5099 mask that should be associated with the selected policy.
5101 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
5102 generates when processes it manages are reaching their memory limits,
5103 and will place their units in a special state, and optionally kill or
5104 stop the whole unit.
5106 * The service manager will now expose bus properties for the IO
5107 resources used by units. This information is also shown in "systemctl
5108 status" now (for services that have IOAccounting=yes set). Moreover,
5109 the IO accounting data is included in the resource log message
5110 generated whenever a unit stops.
5112 * Units may now configure an explicit timeout to wait for when killed
5113 with SIGABRT, for example when a service watchdog is hit. Previously,
5114 the regular TimeoutStopSec= timeout was applied in this case too —
5115 now a separate timeout may be set using TimeoutAbortSec=.
5117 * Services may now send a special WATCHDOG=trigger message with
5118 sd_notify() to trigger an immediate "watchdog missed" event, and thus
5119 trigger service termination. This is useful both for testing watchdog
5120 handling, but also for defining error paths in services, that shall
5121 be handled the same way as watchdog events.
5123 * There are two new per-unit settings IPIngressFilterPath= and
5124 IPEgressFilterPath= which allow configuration of a BPF program
5125 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
5126 to apply to the IP packet ingress/egress path of all processes of a
5127 unit. This is useful to allow running systemd services with BPF
5128 programs set up externally.
5130 * systemctl gained a new "clean" verb for removing the state, cache,
5131 runtime or logs directories of a service while it is terminated. The
5132 new verb may also be used to remove the state maintained on disk for
5133 timer units that have Persistent= configured.
5135 * During the last phase of shutdown systemd will now automatically
5136 increase the log level configured in the "kernel.printk" sysctl so
5137 that any relevant loggable events happening during late shutdown are
5138 made visible. Previously, loggable events happening so late during
5139 shutdown were generally lost if the "kernel.printk" sysctl was set to
5140 high thresholds, as regular logging daemons are terminated at that
5141 time and thus nothing is written to disk.
5143 * If processes terminated during the last phase of shutdown do not exit
5144 quickly systemd will now show their names after a short time, to make
5145 debugging easier. After a longer timeout they are forcibly killed,
5148 * journalctl (and the other tools that display logs) will now highlight
5149 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
5150 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
5151 are now shown in blue color, to separate them visually from regular
5152 logs. References to configuration files are now turned into clickable
5153 links on terminals that support that.
5155 * systemd-journald will now stop logging to /var/log/journal during
5156 shutdown when /var/ is on a separate mount, so that it can be
5157 unmounted safely during shutdown.
5159 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
5161 * systemd-resolved "Cache=" configuration option in resolved.conf has
5162 been extended to also accept the 'no-negative' value. Previously,
5163 only a boolean option was allowed (yes/no), having yes as the
5164 default. If this option is set to 'no-negative', negative answers are
5165 not cached while the old cache heuristics are used positive answers.
5166 The default remains unchanged.
5168 * The predictable naming scheme for network devices now supports
5169 generating predictable names for "netdevsim" devices.
5171 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
5174 Those two changes form a new net.naming-policy-scheme= entry.
5175 Distributions which want to preserve naming stability may want to set
5176 the -Ddefault-net-naming-scheme= configuration option.
5178 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
5179 interfaces natively.
5181 * systemd-networkd's bridge FDB support now allows configuration of a
5182 destination address for each entry (Destination=), as well as the
5183 VXLAN VNI (VNI=), as well as an option to declare what an entry is
5184 associated with (AssociatedWith=).
5186 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
5187 option for configuring the maximum number of DHCP lease requests. It
5188 also learnt a new BlackList= option for deny-listing DHCP servers (a
5189 similar setting has also been added to the IPv6 RA client), as well
5190 as a SendRelease= option for configuring whether to send a DHCP
5191 RELEASE message when terminating.
5193 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
5194 separately in the [DHCPv4] and [DHCPv6] sections.
5196 * systemd-networkd's DHCP support will now optionally create an
5197 implicit host route to the DNS server specified in the DHCP lease, in
5198 addition to the routes listed explicitly in the lease. This should
5199 ensure that in multi-homed systems DNS traffic leaves the systems on
5200 the interface that acquired the DNS server information even if other
5201 routes such as default routes exist. This behaviour may be turned on
5202 with the new RoutesToDNS= option.
5204 * systemd-networkd's VXLAN support gained a new option
5205 GenericProtocolExtension= for enabling VXLAN Generic Protocol
5206 Extension support, as well as IPDoNotFragment= for setting the IP
5207 "Don't fragment" bit on outgoing packets. A similar option has been
5208 added to the GENEVE support.
5210 * In systemd-networkd's [Route] section you may now configure
5211 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
5212 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
5213 propagation. The Type= setting now supports local, broadcast,
5214 anycast, multicast, any, xresolve routes, too.
5216 * systemd-networkd's [Network] section learnt a new option
5217 DefaultRouteOnDevice= for automatically configuring a default route
5218 onto the network device.
5220 * systemd-networkd's bridging support gained two new options ProxyARP=
5221 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
5222 MulticastRouter= for configuring multicast routing behaviour. A new
5223 option MulticastIGMPVersion= may be used to change bridge's multicast
5224 Internet Group Management Protocol (IGMP) version.
5226 * systemd-networkd's FooOverUDP support gained the ability to configure
5227 local and peer IP addresses via Local= and Peer=. A new option
5228 PeerPort= may be used to configure the peer's IP port.
5230 * systemd-networkd's TUN support gained a new setting VnetHeader= for
5231 tweaking Generic Segment Offload support.
5233 * The address family for policy rules may be specified using the new
5234 Family= option in the [RoutingPolicyRule] section.
5236 * networkctl gained a new "delete" command for removing virtual network
5237 devices, as well as a new "--stats" switch for showing device
5240 * networkd.conf gained a new setting SpeedMeter= and
5241 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
5242 measured speed may be shown by 'networkctl status'.
5244 * "networkctl status" now displays MTU and queue lengths, and more
5245 detailed information about VXLAN and bridge devices.
5247 * systemd-networkd's .network and .link files gained a new Property=
5248 setting in the [Match] section, to match against devices with
5249 specific udev properties.
5251 * systemd-networkd's tunnel support gained a new option
5252 AssignToLoopback= for selecting whether to use the loopback device
5253 "lo" as underlying device.
5255 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
5256 been renamed to LinkLayerAddress=, and it now allows configuration of
5259 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
5260 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
5261 IPv6 configuration (static or DHCPv6) was found for a given
5262 interface. It will not touch the sysctl otherwise.
5264 * The order of entries is $PATH used by the user manager instance was
5265 changed to put bin/ entries before the corresponding sbin/ entries.
5266 It is recommended to not rely on this order, and only ever have one
5267 binary with a given name in the system paths under /usr.
5269 * A new tool systemd-network-generator has been added that may generate
5270 .network, .netdev and .link files from IP configuration specified on
5271 the kernel command line in the format used by Dracut.
5273 * The CriticalConnection= setting in .network files is now deprecated,
5274 and replaced by a new KeepConfiguration= setting which allows more
5275 detailed configuration of the IP configuration to keep in place.
5277 * systemd-analyze gained a few new verbs:
5279 - "systemd-analyze timestamp" parses and converts timestamps. This is
5280 similar to the existing "systemd-analyze calendar" command which
5281 does the same for recurring calendar events.
5283 - "systemd-analyze timespan" parses and converts timespans (i.e.
5284 durations as opposed to points in time).
5286 - "systemd-analyze condition" will parse and test ConditionXYZ=
5289 - "systemd-analyze exit-status" will parse and convert exit status
5290 codes to their names and back.
5292 - "systemd-analyze unit-files" will print a list of all unit
5293 file paths and unit aliases.
5295 * SuccessExitStatus=, RestartPreventExitStatus=, and
5296 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
5297 is equivalent to "65"). Those exit status name mappings may be
5298 displayed with the systemd-analyze exit-status verb describe above.
5300 * systemd-logind now exposes a per-session SetBrightness() bus call,
5301 which may be used to securely change the brightness of a kernel
5302 brightness device, if it belongs to the session's seat. By using this
5303 call unprivileged clients can make changes to "backlight" and "leds"
5304 devices securely with strict requirements on session membership.
5305 Desktop environments may use this to generically make brightness
5306 changes to such devices without shipping private SUID binaries or
5307 udev rules for that purpose.
5309 * "udevadm info" gained a --wait-for-initialization switch to wait for
5310 a device to be initialized.
5312 * systemd-hibernate-resume-generator will now look for resumeflags= on
5313 the kernel command line, which is similar to rootflags= and may be
5314 used to configure device timeout for the hibernation device.
5316 * sd-event learnt a new API call sd_event_source_disable_unref() for
5317 disabling and unref'ing an event source in a single function. A
5318 related call sd_event_source_disable_unrefp() has been added for use
5319 with gcc's cleanup extension.
5321 * The sd-id128.h public API gained a new definition
5322 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
5325 * "busctl introspect" gained a new switch --xml-interface for dumping
5326 XML introspection data unmodified.
5328 * PID 1 may now show the unit name instead of the unit description
5329 string in its status output during boot. This may be configured in
5330 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
5331 kernel command line option systemd.status_unit_format=.
5333 * PID 1 now understands a new option KExecWatchdogSec= in
5334 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
5335 Previously watchdog functionality was only available for regular
5336 reboots. The new setting defaults to off, because we don't know in
5337 the general case if the watchdog will be reset after kexec (some
5338 drivers do reset it, but not all), and the new userspace might not be
5339 configured to handle the watchdog.
5341 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
5342 RebootWatchdogSec= to more clearly communicate what it is about. The
5343 old name is still accepted for compatibility.
5345 * The systemd.debug_shell kernel command line option now optionally
5346 takes a tty name to spawn the debug shell on, which allows a
5347 different tty to be selected than the built-in default.
5349 * Service units gained a new ExecCondition= setting which will run
5350 before ExecStartPre= and either continue execution of the unit (for
5351 clean exit codes), stop execution without marking the unit failed
5352 (for exit codes 1 through 254), or stop execution and fail the unit
5353 (for exit code 255 or abnormal termination).
5355 * A new service systemd-pstore.service has been added that pulls data
5356 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
5359 * timedatectl gained new verbs for configuring per-interface NTP
5360 service configuration for systemd-timesyncd.
5362 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
5363 2019. (You can set non-UTF-8 locales though, if you know their name.)
5365 * If variable assignments in sysctl.d/ files are prefixed with "-" any
5366 failures to apply them are now ignored.
5368 * systemd-random-seed.service now optionally credits entropy when
5369 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
5370 true for the service to enable this behaviour, but please consult the
5371 documentation first, since this comes with a couple of caveats.
5373 * systemd-random-seed.service is now a synchronization point for full
5374 initialization of the kernel's entropy pool. Services that require
5375 /dev/urandom to be correctly initialized should be ordered after this
5378 * The systemd-boot boot loader has been updated to optionally maintain
5379 a random seed file in the EFI System Partition (ESP). During the boot
5380 phase, this random seed is read and updated with a new seed
5381 cryptographically derived from it. Another derived seed is passed to
5382 the OS. The latter seed is then credited to the kernel's entropy pool
5383 very early during userspace initialization (from PID 1). This allows
5384 systems to boot up with a fully initialized kernel entropy pool from
5385 earliest boot on, and thus entirely removes all entropy pool
5386 initialization delays from systems using systemd-boot. Special care
5387 is taken to ensure different seeds are derived on system images
5388 replicated to multiple systems. "bootctl status" will show whether
5389 a seed was received from the boot loader.
5391 * bootctl gained two new verbs:
5393 - "bootctl random-seed" will generate the file in ESP and an EFI
5394 variable to allow a random seed to be passed to the OS as described
5397 - "bootctl is-installed" checks whether systemd-boot is currently
5400 * bootctl will warn if it detects that boot entries are misconfigured
5401 (for example if the kernel image was removed without purging the
5404 * A new document has been added describing systemd's use and support
5405 for the kernel's entropy pool subsystem:
5407 https://systemd.io/RANDOM_SEEDS
5409 * When the system is hibernated the swap device to write the
5410 hibernation image to is now automatically picked from all available
5411 swap devices, preferring the swap device with the highest configured
5412 priority over all others, and picking the device with the most free
5413 space if there are multiple devices with the highest priority.
5415 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
5416 option that permits selecting the timeout how long to wait for a
5417 device with an encryption key before asking for the password.
5419 * IOWeight= has learnt to properly set the IO weight when using the
5420 BFQ scheduler officially found in kernels 5.0+.
5422 * A new mailing list has been created for reporting of security issues:
5423 systemd-security@redhat.com. For mode details, see
5424 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
5426 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
5427 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
5428 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
5429 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
5430 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
5431 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
5432 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
5433 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
5434 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
5435 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
5436 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
5437 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
5438 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
5439 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
5440 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
5441 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
5442 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
5443 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
5444 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
5445 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
5446 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
5447 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
5448 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
5449 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
5450 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
5451 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
5452 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
5453 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
5454 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
5455 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
5457 – Camerino, 2019-09-03
5461 * In .link files, MACAddressPolicy=persistent (the default) is changed
5462 to cover more devices. For devices like bridges, tun, tap, bond, and
5463 similar interfaces that do not have other identifying information,
5464 the interface name is used as the basis for persistent seed for MAC
5465 and IPv4LL addresses. The way that devices that were handled
5466 previously is not changed, and this change is about covering more
5467 devices then previously by the "persistent" policy.
5469 MACAddressPolicy=random may be used to force randomized MACs and
5470 IPv4LL addresses for a device if desired.
5472 Hint: the log output from udev (at debug level) was enhanced to
5473 clarify what policy is followed and which attributes are used.
5474 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
5475 may be used to view this.
5477 Hint: if a bridge interface is created without any slaves, and gains
5478 a slave later, then now the bridge does not inherit slave's MAC.
5479 To inherit slave's MAC, for example, create the following file:
5481 # /etc/systemd/network/98-bridge-inherit-mac.link
5486 MACAddressPolicy=none
5489 * The .device units generated by systemd-fstab-generator and other
5490 generators do not automatically pull in the corresponding .mount unit
5491 as a Wants= dependency. This means that simply plugging in the device
5492 will not cause the mount unit to be started automatically. But please
5493 note that the mount unit may be started for other reasons, in
5494 particular if it is part of local-fs.target, and any unit which
5495 (transitively) depends on local-fs.target is started.
5497 * networkctl list/status/lldp now accept globbing wildcards for network
5498 interface names to match against all existing interfaces.
5500 * The $PIDFILE environment variable is set to point the absolute path
5501 configured with PIDFile= for processes of that service.
5503 * The fallback DNS server list was augmented with Cloudflare public DNS
5504 servers. Use `-Ddns-servers=` to set a different fallback.
5506 * A new special target usb-gadget.target will be started automatically
5507 when a USB Device Controller is detected (which means that the system
5508 is a USB peripheral).
5510 * A new unit setting CPUQuotaPeriodSec= assigns the time period
5511 relatively to which the CPU time quota specified by CPUQuota= is
5514 * A new unit setting ProtectHostname= may be used to prevent services
5515 from modifying hostname information (even if they otherwise would
5516 have privileges to do so).
5518 * A new unit setting NetworkNamespacePath= may be used to specify a
5519 namespace for service or socket units through a path referring to a
5520 Linux network namespace pseudo-file.
5522 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
5523 have an effect on .socket units: when used the listening socket is
5524 created within the configured network namespace instead of the host
5527 * ExecStart= command lines in unit files may now be prefixed with ':'
5528 in which case environment variable substitution is
5529 disabled. (Supported for the other ExecXYZ= settings, too.)
5531 * .timer units gained two new boolean settings OnClockChange= and
5532 OnTimezoneChange= which may be used to also trigger a unit when the
5533 system clock is changed or the local timezone is
5534 modified. systemd-run has been updated to make these options easily
5535 accessible from the command line for transient timers.
5537 * Two new conditions for units have been added: ConditionMemory= may be
5538 used to conditionalize a unit based on installed system
5539 RAM. ConditionCPUs= may be used to conditionalize a unit based on
5540 installed CPU cores.
5542 * The @default system call filter group understood by SystemCallFilter=
5543 has been updated to include the new rseq() system call introduced in
5546 * A new time-set.target has been added that indicates that the system
5547 time has been set from a local source (possibly imprecise). The
5548 existing time-sync.target is stronger and indicates that the time has
5549 been synchronized with a precise external source. Services where
5550 approximate time is sufficient should use the new target.
5552 * "systemctl start" (and related commands) learnt a new
5553 --show-transaction option. If specified brief information about all
5554 jobs queued because of the requested operation is shown.
5556 * systemd-networkd recognizes a new operation state 'enslaved', used
5557 (instead of 'degraded' or 'carrier') for interfaces which form a
5558 bridge, bond, or similar, and an new 'degraded-carrier' operational
5559 state used for the bond or bridge master interface when one of the
5560 enslaved devices is not operational.
5562 * .network files learnt the new IgnoreCarrierLoss= option for leaving
5563 networks configured even if the carrier is lost.
5565 * The RequiredForOnline= setting in .network files may now specify a
5566 minimum operational state required for the interface to be considered
5567 "online" by systemd-networkd-wait-online. Related to this
5568 systemd-networkd-wait-online gained a new option --operational-state=
5569 to configure the same, and its --interface= option was updated to
5570 optionally also take an operational state specific for an interface.
5572 * systemd-networkd-wait-online gained a new setting --any for waiting
5573 for only one of the requested interfaces instead of all of them.
5575 * systemd-networkd now implements L2TP tunnels.
5577 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
5578 may be used to cause autonomous and onlink prefixes received in IPv6
5579 Router Advertisements to be ignored.
5581 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
5582 file settings may be used to tweak bridge behaviour.
5584 * The new TripleSampling= option in .network files may be used to
5585 configure CAN triple sampling.
5587 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
5588 used to point to private or preshared key for a WireGuard interface.
5590 * /etc/crypttab now supports the same-cpu-crypt and
5591 submit-from-crypt-cpus options to tweak encryption work scheduling
5594 * systemd-tmpfiles will now take a BSD file lock before operating on a
5595 contents of directory. This may be used to temporarily exclude
5596 directories from aging by taking the same lock (useful for example
5597 when extracting a tarball into /tmp or /var/tmp as a privileged user,
5598 which might create files with really old timestamps, which
5599 nevertheless should not be deleted). For further details, see:
5601 https://systemd.io/TEMPORARY_DIRECTORIES
5603 * systemd-tmpfiles' h line type gained support for the
5604 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
5605 controlling project quota inheritance.
5607 * sd-boot and bootctl now implement support for an Extended Boot Loader
5608 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
5609 addition to the ESP partition mounted to /efi or /boot/efi.
5610 Configuration file fragments, kernels, initrds and other EFI images
5611 to boot will be loaded from both the ESP and XBOOTLDR partitions.
5612 The XBOOTLDR partition was previously described by the Boot Loader
5613 Specification, but implementation was missing in sd-boot. Support for
5614 this concept allows using the sd-boot boot loader in more
5615 conservative scenarios where the boot loader itself is placed in the
5616 ESP but the kernels to boot (and their metadata) in a separate
5619 * A system may now be booted with systemd.volatile=overlay on the
5620 kernel command line, which causes the root file system to be set up
5621 an overlayfs mount combining the root-only root directory with a
5622 writable tmpfs. In this setup, the underlying root device is not
5623 modified, and any changes are lost at reboot.
5625 * Similar, systemd-nspawn can now boot containers with a volatile
5626 overlayfs root with the new --volatile=overlay switch.
5628 * systemd-nspawn can now consume OCI runtime bundles using a new
5629 --oci-bundle= option. This implementation is fully usable, with most
5630 features in the specification implemented, but since this a lot of
5631 new code and functionality, this feature should most likely not
5632 be used in production yet.
5634 * systemd-nspawn now supports various options described by the OCI
5635 runtime specification on the command-line and in .nspawn files:
5636 --inaccessible=/Inaccessible= may be used to mask parts of the file
5637 system tree, --console=/--pipe may be used to configure how standard
5638 input, output, and error are set up.
5640 * busctl learned the `emit` verb to generate D-Bus signals.
5642 * systemd-analyze cat-config may be used to gather and display
5643 configuration spread over multiple files, for example system and user
5644 presets, tmpfiles.d, sysusers.d, udev rules, etc.
5646 * systemd-analyze calendar now takes an optional new parameter
5647 --iterations= which may be used to show a maximum number of iterations
5648 the specified expression will elapse next.
5650 * The sd-bus C API gained support for naming method parameters in the
5653 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
5654 the reboot() system call expects.
5656 * journalctl learnt a new --cursor-file= option that points to a file
5657 from which a cursor should be loaded in the beginning and to which
5658 the updated cursor should be stored at the end.
5660 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
5661 detected by systemd-detect-virt (and may also be used in
5662 ConditionVirtualization=).
5664 * The behaviour of systemd-logind may now be modified with environment
5665 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
5666 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
5667 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
5668 skip the relevant operation completely (when set to false), or to
5669 create a flag file in /run/systemd (when set to true), instead of
5670 actually commencing the real operation when requested. The presence
5671 of /run/systemd/reboot-to-firmware-setup,
5672 /run/systemd/reboot-to-boot-loader-menu, and
5673 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
5674 boot loader implementations to replace some steps logind performs
5675 during reboot with their own operations.
5677 * systemctl can be used to request a reboot into the boot loader menu
5678 or a specific boot loader entry with the new --boot-load-menu= and
5679 --boot-loader-entry= options to a reboot command. (This requires a
5680 boot loader that supports this, for example sd-boot.)
5682 * kernel-install will no longer unconditionally create the output
5683 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
5684 snippets, but will do only if the machine-specific parent directory
5685 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
5686 to create this parent directory during sd-boot installation.
5688 This makes it easier to use kernel-install with plugins which support
5689 a different layout of the bootloader partitions (for example grub2).
5691 * During package installation (with `ninja install`), we would create
5692 symlinks for getty@tty1.service, systemd-networkd.service,
5693 systemd-networkd.socket, systemd-resolved.service,
5694 remote-cryptsetup.target, remote-fs.target,
5695 systemd-networkd-wait-online.service, and systemd-timesyncd.service
5696 in /etc, as if `systemctl enable` was called for those units, to make
5697 the system usable immediately after installation. Now this is not
5698 done anymore, and instead calling `systemctl preset-all` is
5699 recommended after the first installation of systemd.
5701 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
5702 is built on seccomp. When turned on creation of SUID/SGID files is
5705 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
5706 implied if DynamicUser= is turned on for a service. This hardens
5707 these services, so that they neither can benefit from nor create
5708 SUID/SGID executables. This is a minor compatibility breakage, given
5709 that when DynamicUser= was first introduced SUID/SGID behaviour was
5710 unaffected. However, the security benefit of these two options is
5711 substantial, and the setting is still relatively new, hence we opted
5712 to make it mandatory for services with dynamic users.
5714 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
5715 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
5716 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
5717 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
5718 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
5719 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
5720 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
5721 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
5722 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
5723 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
5724 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
5725 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
5726 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
5727 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
5728 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
5729 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
5730 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
5731 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5733 — Warsaw, 2019-04-11
5737 * The default locale can now be configured at compile time. Otherwise,
5738 a suitable default will be selected automatically (one of C.UTF-8,
5739 en_US.UTF-8, and C).
5741 * The version string shown by systemd and other tools now includes the
5742 git commit hash when built from git. An override may be specified
5743 during compilation, which is intended to be used by distributions to
5744 include the package release information.
5746 * systemd-cat can now filter standard input and standard error streams
5747 for different syslog priorities using the new --stderr-priority=
5750 * systemd-journald and systemd-journal-remote reject entries which
5751 contain too many fields (CVE-2018-16865) and set limits on the
5752 process' command line length (CVE-2018-16864).
5754 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
5757 * A new network device NamePolicy "keep" is implemented for link files,
5758 and used by default in 99-default.link (the fallback configuration
5759 provided by systemd). With this policy, if the network device name
5760 was already set by userspace, the device will not be renamed again.
5761 This matches the naming scheme that was implemented before
5762 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
5763 is also enabled by default, even if not specified. Effectively, this
5764 means that if naming-scheme >= 240 is specified, network devices will
5765 be renamed according to the configuration, even if they have been
5766 renamed already, if "keep" is not specified as the naming policy in
5767 the .link file. The 99-default.link file provided by systemd includes
5768 "keep" for backwards compatibility, but it is recommended for user
5769 installed .link files to *not* include it.
5771 The "kernel" policy, which keeps kernel names declared to be
5772 "persistent", now works again as documented.
5774 * kernel-install script now optionally takes the paths to one or more
5775 initrd files, and passes them to all plugins.
5777 * The mincore() system call has been dropped from the @system-service
5778 system call filter group, as it is pretty exotic and may potentially
5779 used for side-channel attacks.
5781 * -fPIE is dropped from compiler and linker options. Please specify
5782 -Db_pie=true option to meson to build position-independent
5783 executables. Note that the meson option is supported since meson-0.49.
5785 * The fs.protected_regular and fs.protected_fifos sysctls, which were
5786 added in Linux 4.19 to make some data spoofing attacks harder, are
5787 now enabled by default. While this will hopefully improve the
5788 security of most installations, it is technically a backwards
5789 incompatible change; to disable these sysctls again, place the
5790 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
5792 fs.protected_regular = 0
5793 fs.protected_fifos = 0
5795 Note that the similar hardlink and symlink protection has been
5796 enabled since v199, and may be disabled likewise.
5798 * The files read from the EnvironmentFile= setting in unit files now
5799 parse backslashes inside quotes literally, matching the behaviour of
5802 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
5803 now automatically become NOPs when run in a chroot() environment.
5805 * The tmpfiles.d/ "C" line type will now copy directory trees not only
5806 when the destination is so far missing, but also if it already exists
5807 as a directory and is empty. This is useful to cater for systems
5808 where directory trees are put together from multiple separate mount
5809 points but otherwise empty.
5811 * A new function sd_bus_close_unref() (and the associated
5812 sd_bus_close_unrefp()) has been added to libsystemd, that combines
5813 sd_bus_close() and sd_bus_unref() in one.
5815 * udevadm control learnt a new option for --ping for testing whether a
5816 systemd-udevd instance is running and reacting.
5818 * udevadm trigger learnt a new option for --wait-daemon for waiting
5819 systemd-udevd daemon to be initialized.
5821 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
5822 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
5823 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
5824 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
5825 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
5826 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
5827 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
5828 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
5829 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
5830 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
5831 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
5832 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
5833 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
5834 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
5835 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
5836 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
5837 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
5839 — Berlin, 2019-02-14
5843 * NoNewPrivileges=yes has been set for all long-running services
5844 implemented by systemd. Previously, this was problematic due to
5845 SELinux (as this would also prohibit the transition from PID1's label
5846 to the service's label). This restriction has since been lifted, but
5847 an SELinux policy update is required.
5848 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
5850 * DynamicUser=yes is dropped from systemd-networkd.service,
5851 systemd-resolved.service and systemd-timesyncd.service, which was
5852 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
5853 and since v236 for systemd-timesyncd.service. The users and groups
5854 systemd-network, systemd-resolve and systemd-timesync are created
5855 by systemd-sysusers again. Distributors or system administrators
5856 may need to create these users and groups if they not exist (or need
5857 to re-enable DynamicUser= for those units) while upgrading systemd.
5858 Also, the clock file for systemd-timesyncd may need to move from
5859 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
5861 * When unit files are loaded from disk, previously systemd would
5862 sometimes (depending on the unit loading order) load units from the
5863 target path of symlinks in .wants/ or .requires/ directories of other
5864 units. This meant that unit could be loaded from different paths
5865 depending on whether the unit was requested explicitly or as a
5866 dependency of another unit, not honouring the priority of directories
5867 in search path. It also meant that it was possible to successfully
5868 load and start units which are not found in the unit search path, as
5869 long as they were requested as a dependency and linked to from
5870 .wants/ or .requires/. The target paths of those symlinks are not
5871 used for loading units anymore and the unit file must be found in
5874 * A new service type has been added: Type=exec. It's very similar to
5875 Type=simple but ensures the service manager will wait for both fork()
5876 and execve() of the main service binary to complete before proceeding
5877 with follow-up units. This is primarily useful so that the manager
5878 propagates any errors in the preparation phase of service execution
5879 back to the job that requested the unit to be started. For example,
5880 consider a service that has ExecStart= set to a file system binary
5881 that doesn't exist. With Type=simple starting the unit would be
5882 considered instantly successful, as only fork() has to complete
5883 successfully and the manager does not wait for execve(), and hence
5884 its failure is seen "too late". With the new Type=exec service type
5885 starting the unit will fail, as the manager will wait for the
5886 execve() and notice its failure, which is then propagated back to the
5889 NOTE: with the next release 241 of systemd we intend to change the
5890 systemd-run tool to default to Type=exec for transient services
5891 started by it. This should be mostly safe, but in specific corner
5892 cases might result in problems, as the systemd-run tool will then
5893 block on NSS calls (such as user name look-ups due to User=) done
5894 between the fork() and execve(), which under specific circumstances
5895 might cause problems. It is recommended to specify "-p Type=simple"
5896 explicitly in the few cases where this applies. For regular,
5897 non-transient services (i.e. those defined with unit files on disk)
5898 we will continue to default to Type=simple.
5900 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
5901 userspace processes is set to 1024 (soft) and 4096
5902 (hard). Previously, systemd passed this on unmodified to all
5903 processes it forked off. With this systemd release the hard limit
5904 systemd passes on is increased to 512K, overriding the kernel's
5905 defaults and substantially increasing the number of simultaneous file
5906 descriptors unprivileged userspace processes can allocate. Note that
5907 the soft limit remains at 1024 for compatibility reasons: the
5908 traditional UNIX select() call cannot deal with file descriptors >=
5909 1024 and increasing the soft limit globally might thus result in
5910 programs unexpectedly allocating a high file descriptor and thus
5911 failing abnormally when attempting to use it with select() (of
5912 course, programs shouldn't use select() anymore, and prefer
5913 poll()/epoll, but the call unfortunately remains undeservedly popular
5914 at this time). This change reflects the fact that file descriptor
5915 handling in the Linux kernel has been optimized in more recent
5916 kernels and allocating large numbers of them should be much cheaper
5917 both in memory and in performance than it used to be. Programs that
5918 want to take benefit of the increased limit have to "opt-in" into
5919 high file descriptors explicitly by raising their soft limit. Of
5920 course, when they do that they must acknowledge that they cannot use
5921 select() anymore (and neither can any shared library they use — or
5922 any shared library used by any shared library they use and so on).
5923 Which default hard limit is most appropriate is of course hard to
5924 decide. However, given reports that ~300K file descriptors are used
5925 in real-life applications we believe 512K is sufficiently high as new
5926 default for now. Note that there are also reports that using very
5927 high hard limits (e.g. 1G) is problematic: some software allocates
5928 large arrays with one element for each potential file descriptor
5929 (Java, …) — a high hard limit thus triggers excessively large memory
5930 allocations in these applications. Hopefully, the new default of 512K
5931 is a good middle ground: higher than what real-life applications
5932 currently need, and low enough for avoid triggering excessively large
5933 allocations in problematic software. (And yes, somebody should fix
5936 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
5937 to the highest possible values, as separate accounting of file
5938 descriptors is no longer necessary, as memcg tracks them correctly as
5939 part of the memory accounting anyway. Thus, from the four limits on
5940 file descriptors currently enforced (fs.file-max, fs.nr_open,
5941 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
5942 and keep only the latter two. A set of build-time options
5943 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
5944 has been added to revert this change in behaviour, which might be
5945 an option for systems that turn off memcg in the kernel.
5947 * When no /etc/locale.conf file exists (and hence no locale settings
5948 are in place), systemd will now use the "C.UTF-8" locale by default,
5949 and set LANG= to it. This locale is supported by various
5950 distributions including Fedora, with clear indications that upstream
5951 glibc is going to make it available too. This locale enables UTF-8
5952 mode by default, which appears appropriate for 2018.
5954 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
5955 default. This effectively switches the RFC3704 Reverse Path filtering
5956 from Strict mode to Loose mode. This is more appropriate for hosts
5957 that have multiple links with routes to the same networks (e.g.
5958 a client with a Wi-Fi and Ethernet both connected to the internet).
5960 Consult the kernel documentation for details on this sysctl:
5961 https://docs.kernel.org/networking/ip-sysctl.html
5963 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
5966 * CPUAccounting=yes no longer enables the CPU controller when using
5967 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
5968 statistics are now provided independently from the CPU controller.
5970 * Support for disabling a particular cgroup controller within a sub-tree
5971 has been added through the DisableControllers= directive.
5973 * cgroup_no_v1=all on the kernel command line now also implies
5974 using the unified cgroup hierarchy, unless one explicitly passes
5975 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
5977 * The new "MemoryMin=" unit file property may now be used to set the
5978 memory usage protection limit of processes invoked by the unit. This
5979 controls the cgroup v2 memory.min attribute. Similarly, the new
5980 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
5981 cgroup v2 io.latency cgroup property for configuring per-service I/O
5984 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
5985 to the cgroup v1 "devices" cgroup controller.
5987 * systemd-escape now is able to combine --unescape with --template. It
5988 also learnt a new option --instance for extracting and unescaping the
5989 instance part of a unit name.
5991 * sd-bus now provides the sd_bus_message_readv() which is similar to
5992 sd_bus_message_read() but takes a va_list object. The pair
5993 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
5994 has been added for configuring the default method call timeout to
5995 use. sd_bus_error_move() may be used to efficiently move the contents
5996 from one sd_bus_error structure to another, invalidating the
5997 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
5998 be used to control whether a bus connection object is automatically
5999 flushed when an sd-event loop is exited.
6001 * When processing classic BSD syslog log messages, journald will now
6002 save the original time-stamp string supplied in the new
6003 SYSLOG_TIMESTAMP= journal field. This permits consumers to
6004 reconstruct the original BSD syslog message more correctly.
6006 * StandardOutput=/StandardError= in service files gained support for
6007 new "append:…" parameters, for connecting STDOUT/STDERR of a service
6008 to a file, and appending to it.
6010 * The signal to use as last step of killing of unit processes is now
6011 configurable. Previously it was hard-coded to SIGKILL, which may now
6012 be overridden with the new KillSignal= setting. Note that this is the
6013 signal used when regular termination (i.e. SIGTERM) does not suffice.
6014 Similarly, the signal used when aborting a program in case of a
6015 watchdog timeout may now be configured too (WatchdogSignal=).
6017 * The XDG_SESSION_DESKTOP environment variable may now be configured in
6018 the pam_systemd argument line, using the new desktop= switch. This is
6019 useful to initialize it properly from a display manager without
6020 having to touch C code.
6022 * Most configuration options that previously accepted percentage values
6023 now also accept permille values with the '‰' suffix (instead of '%').
6025 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
6028 * systemd-resolved's configuration file resolved.conf gained a new
6029 option ReadEtcHosts= which may be used to turn off processing and
6030 honoring /etc/hosts entries.
6032 * The "--wait" switch may now be passed to "systemctl
6033 is-system-running", in which case the tool will synchronously wait
6034 until the system finished start-up.
6036 * hostnamed gained a new bus call to determine the DMI product UUID.
6038 * On x86-64 systemd will now prefer using the RDRAND processor
6039 instruction over /dev/urandom whenever it requires randomness that
6040 neither has to be crypto-grade nor should be reproducible. This
6041 should substantially reduce the amount of entropy systemd requests
6042 from the kernel during initialization on such systems, though not
6043 reduce it to zero. (Why not zero? systemd still needs to allocate
6044 UUIDs and such uniquely, which require high-quality randomness.)
6046 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
6047 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
6048 for forcing the "Other Information" bit in IPv6 RA messages. The
6049 bonding logic gained four new options AdActorSystemPriority=,
6050 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
6051 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
6052 shuffling of flows. The tunnel logic gained a new
6053 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
6054 Deployment. The policy rule logic gained four new options IPProtocol=,
6055 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
6056 support for the MulticastToUnicast= option. networkd also gained
6057 support for configuring static IPv4 ARP or IPv6 neighbor entries.
6059 * .preset files (as read by 'systemctl preset') may now be used to
6060 instantiate services.
6062 * /etc/crypttab now understands the sector-size= option to configure
6063 the sector size for an encrypted partition.
6065 * Key material for encrypted disks may now be placed on a formatted
6066 medium, and referenced from /etc/crypttab by the UUID of the file
6067 system, followed by "=" suffixed by the path to the key file.
6069 * The "collect" udev component has been removed without replacement, as
6070 it is neither used nor maintained.
6072 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
6073 LogsDirectory=, ConfigurationDirectory= settings are used in a
6074 service the executed processes will now receive a set of environment
6075 variables containing the full paths of these directories.
6076 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
6077 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
6078 are used. Note that these options may be used multiple times per
6079 service in which case the resulting paths will be concatenated and
6080 separated by colons.
6082 * Predictable interface naming has been extended to cover InfiniBand
6083 NICs. They will be exposed with an "ib" prefix.
6085 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
6086 which case the respective line failing is ignored.
6088 * .link files may now be used to configure the equivalent to the
6089 "ethtool advertise" commands.
6091 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
6092 alternative to libudev.h. Previously, the latter was just an internal
6093 wrapper around the former, but now these two APIs are exposed
6096 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
6097 which calculates an app-specific boot ID similar to how
6098 sd_id128_get_machine_app_specific() generates an app-specific machine
6101 * A new tool systemd-id128 has been added that can be used to determine
6102 and generate various 128bit IDs.
6104 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
6107 * systemd-hibernate-resume-generator will now honor the "noresume"
6108 kernel command line option, in which case it will bypass resuming
6109 from any hibernated image.
6111 * The systemd-sleep.conf configuration file gained new options
6112 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
6113 AllowHybridSleep= for prohibiting specific sleep modes even if the
6114 kernel exports them.
6116 * portablectl is now officially supported and has thus moved to
6119 * bootctl learnt the two new commands "set-default" and "set-oneshot"
6120 for setting the default boot loader item to boot to (either
6121 persistently or only for the next boot). This is currently only
6122 compatible with sd-boot, but may be implemented on other boot loaders
6123 too, that follow the boot loader interface. The updated interface is
6124 now documented here:
6126 https://systemd.io/BOOT_LOADER_INTERFACE
6128 * A new kernel command line option systemd.early_core_pattern= is now
6129 understood which may be used to influence the core_pattern PID 1
6130 installs during early boot.
6132 * busctl learnt two new options -j and --json= for outputting method
6133 call replies, properties and monitoring output in JSON.
6135 * journalctl's JSON output now supports simple ANSI coloring as well as
6136 a new "json-seq" mode for generating RFC7464 output.
6138 * Unit files now support the %g/%G specifiers that resolve to the UNIX
6139 group/GID of the service manager runs as, similar to the existing
6140 %u/%U specifiers that resolve to the UNIX user/UID.
6142 * systemd-logind learnt a new global configuration option
6143 UserStopDelaySec= that may be set in logind.conf. It specifies how
6144 long the systemd --user instance shall remain started after a user
6145 logs out. This is useful to speed up repetitive re-connections of the
6146 same user, as it means the user's service manager doesn't have to be
6147 stopped/restarted on each iteration, but can be reused between
6148 subsequent options. This setting defaults to 10s. systemd-logind also
6149 exports two new properties on its Manager D-Bus objects indicating
6150 whether the system's lid is currently closed, and whether the system
6153 * systemd gained support for a generic boot counting logic, which
6154 generically permits automatic reverting to older boot loader entries
6155 if newer updated ones don't work. The boot loader side is implemented
6156 in sd-boot, but is kept open for other boot loaders too. For details
6159 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
6161 * The SuccessAction=/FailureAction= unit file settings now learnt two
6162 new parameters: "exit" and "exit-force", which result in immediate
6163 exiting of the service manager, and are only useful in systemd --user
6164 and container environments.
6166 * Unit files gained support for a pair of options
6167 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
6168 exit status to use as service manager exit status when
6169 SuccessAction=/FailureAction= is set to exit or exit-force.
6171 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
6172 options may now be used to configure the log rate limiting applied by
6173 journald per-service.
6175 * systemd-analyze gained a new verb "timespan" for parsing and
6176 normalizing time span values (i.e. strings like "5min 7s 8us").
6178 * systemd-analyze also gained a new verb "security" for analyzing the
6179 security and sand-boxing settings of services in order to determine an
6180 "exposure level" for them, indicating whether a service would benefit
6181 from more sand-boxing options turned on for them.
6183 * "systemd-analyze syscall-filter" will now also show system calls
6184 supported by the local kernel but not included in any of the defined
6187 * .nspawn files now understand the Ephemeral= setting, matching the
6188 --ephemeral command line switch.
6190 * sd-event gained the new APIs sd_event_source_get_floating() and
6191 sd_event_source_set_floating() for controlling whether a specific
6192 event source is "floating", i.e. destroyed along with the even loop
6195 * Unit objects on D-Bus gained a new "Refs" property that lists all
6196 clients that currently have a reference on the unit (to ensure it is
6199 * The JoinControllers= option in system.conf is no longer supported, as
6200 it didn't work correctly, is hard to support properly, is legacy (as
6201 the concept only exists on cgroup v1) and apparently wasn't used.
6203 * Journal messages that are generated whenever a unit enters the failed
6204 state are now tagged with a unique MESSAGE_ID. Similarly, messages
6205 generated whenever a service process exits are now made recognizable,
6206 too. A tagged message is also emitted whenever a unit enters the
6207 "dead" state on success.
6209 * systemd-run gained a new switch --working-directory= for configuring
6210 the working directory of the service to start. A shortcut -d is
6211 equivalent, setting the working directory of the service to the
6212 current working directory of the invoking program. The new --shell
6213 (or just -S) option has been added for invoking the $SHELL of the
6214 caller as a service, and implies --pty --same-dir --wait --collect
6215 --service-type=exec. Or in other words, "systemd-run -S" is now the
6216 quickest way to quickly get an interactive in a fully clean and
6217 well-defined system service context.
6219 * machinectl gained a new verb "import-fs" for importing an OS tree
6220 from a directory. Moreover, when a directory or tarball is imported
6221 and single top-level directory found with the OS itself below the OS
6222 tree is automatically mangled and moved one level up.
6224 * systemd-importd will no longer set up an implicit btrfs loop-back
6225 file system on /var/lib/machines. If one is already set up, it will
6226 continue to be used.
6228 * A new generator "systemd-run-generator" has been added. It will
6229 synthesize a unit from one or more program command lines included in
6230 the kernel command line. This is very useful in container managers
6233 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
6235 This will run "systemd-nspawn" on an image, invoke the specified
6236 command line and immediately shut down the container again, returning
6237 the command line's exit code.
6239 * The block device locking logic is now documented:
6241 https://systemd.io/BLOCK_DEVICE_LOCKING
6243 * loginctl and machinectl now optionally output the various tables in
6244 JSON using the --output= switch. It is our intention to add similar
6245 support to systemctl and all other commands.
6247 * udevadm's query and trigger verb now optionally take a .device unit
6250 * systemd-udevd's network naming logic now understands a new
6251 net.naming-scheme= kernel command line switch, which may be used to
6252 pick a specific version of the naming scheme. This helps stabilizing
6253 interface names even as systemd/udev are updated and the naming logic
6256 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
6257 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
6258 initialize one to all 0xFF.
6260 * After loading the SELinux policy systemd will now recursively relabel
6261 all files and directories listed in
6262 /run/systemd/relabel-extra.d/*.relabel (which should be simple
6263 newline separated lists of paths) in addition to the ones it already
6264 implicitly relabels in /run, /dev and /sys. After the relabelling is
6265 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
6266 removed. This is useful to permit initrds (i.e. code running before
6267 the SELinux policy is in effect) to generate files in the host
6268 filesystem safely and ensure that the correct label is applied during
6269 the transition to the host OS.
6271 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
6272 mknod() handling in user namespaces. Previously mknod() would always
6273 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
6274 but device nodes generated that way cannot be opened, and attempts to
6275 open them result in EPERM. This breaks the "graceful fallback" logic
6276 in systemd's PrivateDevices= sand-boxing option. This option is
6277 implemented defensively, so that when systemd detects it runs in a
6278 restricted environment (such as a user namespace, or an environment
6279 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
6280 where device nodes cannot be created the effect of PrivateDevices= is
6281 bypassed (following the logic that 2nd-level sand-boxing is not
6282 essential if the system systemd runs in is itself already sand-boxed
6283 as a whole). This logic breaks with 4.18 in container managers where
6284 user namespacing is used: suddenly PrivateDevices= succeeds setting
6285 up a private /dev/ file system containing devices nodes — but when
6286 these are opened they don't work.
6288 At this point it is recommended that container managers utilizing
6289 user namespaces that intend to run systemd in the payload explicitly
6290 block mknod() with seccomp or similar, so that the graceful fallback
6293 We are very sorry for the breakage and the requirement to change
6294 container configurations for newer kernels. It's purely caused by an
6295 incompatible kernel change. The relevant kernel developers have been
6296 notified about this userspace breakage quickly, but they chose to
6299 * PermissionsStartOnly= setting is deprecated (but is still supported
6300 for backwards compatibility). The same functionality is provided by
6301 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
6304 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
6305 pam_systemd anymore.
6307 * The naming scheme for network devices was changed to always rename
6308 devices, even if they were already renamed by userspace. The "kernel"
6309 policy was changed to only apply as a fallback, if no other naming
6312 * The requirements to build systemd is bumped to meson-0.46 and
6315 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
6316 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
6317 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
6318 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
6319 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
6320 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
6321 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
6322 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
6323 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
6324 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
6325 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
6326 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
6327 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
6328 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
6329 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
6330 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
6331 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
6332 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
6333 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
6334 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
6335 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
6336 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
6337 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
6338 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
6339 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
6340 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
6341 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
6342 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
6343 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
6344 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
6345 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
6346 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
6347 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
6348 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
6349 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
6350 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
6351 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
6352 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
6353 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
6354 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
6355 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
6356 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
6357 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
6358 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
6359 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
6361 — Warsaw, 2018-12-21
6365 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
6366 builtin will name network interfaces differently than in previous
6367 versions for virtual network interfaces created with SR-IOV and NPAR
6368 and for devices where the PCI network controller device does not have
6369 a slot number associated.
6371 SR-IOV virtual devices are now named based on the name of the parent
6372 interface, with a suffix of "v<N>", where <N> is the virtual device
6373 number. Previously those virtual devices were named as if completely
6376 The ninth and later NPAR virtual devices will be named following the
6377 scheme used for the first eight NPAR partitions. Previously those
6378 devices were not renamed and the kernel default (eth<n>) was used.
6380 "net_id" will also generate names for PCI devices where the PCI
6381 network controller device does not have an associated slot number
6382 itself, but one of its parents does. Previously those devices were
6383 not renamed and the kernel default (eth<n>) was used.
6385 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
6386 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
6387 the unit. So, it is expected that the default behavior of
6388 systemd-logind is not changed. However, if distribution packagers or
6389 administrators disabled or modified IPAddressDeny= setting by a
6390 drop-in config file, then it may be necessary to update the file to
6391 re-enable AF_INET and AF_INET6 to support network user name services,
6394 * When the RestrictNamespaces= unit property is specified multiple
6395 times, then the specified types are merged now. Previously, only the
6396 last assignment was used. So, if distribution packagers or
6397 administrators modified the setting by a drop-in config file, then it
6398 may be necessary to update the file.
6400 * When OnFailure= is used in combination with Restart= on a service
6401 unit, then the specified units will no longer be triggered on
6402 failures that result in restarting. Previously, the specified units
6403 would be activated each time the unit failed, even when the unit was
6404 going to be restarted automatically. This behaviour contradicted the
6405 documentation. With this release the code is adjusted to match the
6408 * systemd-tmpfiles will now print a notice whenever it encounters
6409 tmpfiles.d/ lines referencing the /var/run/ directory. It will
6410 recommend reworking them to use the /run/ directory instead (for
6411 which /var/run/ is simply a symlinked compatibility alias). This way
6412 systemd-tmpfiles can properly detect line conflicts and merge lines
6413 referencing the same file by two paths, without having to access
6416 * systemctl disable/unmask/preset/preset-all cannot be used with
6417 --runtime. Previously this was allowed, but resulted in unintuitive
6418 behaviour that wasn't useful. systemctl disable/unmask will now undo
6419 both runtime and persistent enablement/masking, i.e. it will remove
6420 any relevant symlinks both in /run and /etc.
6422 * Note that all long-running system services shipped with systemd will
6423 now default to a system call allow list (rather than a deny list, as
6424 before). In particular, systemd-udevd will now enforce one too. For
6425 most cases this should be safe, however downstream distributions
6426 which disabled sandboxing of systemd-udevd (specifically the
6427 MountFlags= setting), might want to disable this security feature
6428 too, as the default allow-listing will prohibit all mount, swap,
6429 reboot and clock changing operations from udev rules.
6431 * sd-boot acquired new loader configuration settings to optionally turn
6432 off Windows and MacOS boot partition discovery as well as
6433 reboot-into-firmware menu items. It is also able to pick a better
6434 screen resolution for HiDPI systems, and now provides loader
6435 configuration settings to change the resolution explicitly.
6437 * systemd-resolved now supports DNS-over-TLS. It's still
6438 turned off by default, use DNSOverTLS=opportunistic to turn it on in
6439 resolved.conf. We intend to make this the default as soon as couple
6440 of additional techniques for optimizing the initial latency caused by
6441 establishing a TLS/TCP connection are implemented.
6443 * systemd-resolved.service and systemd-networkd.service now set
6444 DynamicUser=yes. The users systemd-resolve and systemd-network are
6445 not created by systemd-sysusers anymore.
6447 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
6448 that embed a network facing module into any process using getpwuid()
6449 or related call: the dynamic allocation of the user ID for
6450 systemd-resolved.service means the service manager has to check NSS
6451 if the user name is already taken when forking off the service. Since
6452 the user in the common case won't be defined in /etc/passwd the
6453 lookup is likely to trigger nss-ldap which in turn might use NSS to
6454 ask systemd-resolved for hostname lookups. This will hence result in
6455 a deadlock: a user name lookup in order to start
6456 systemd-resolved.service will result in a hostname lookup for which
6457 systemd-resolved.service needs to be started already. There are
6458 multiple ways to work around this problem: pre-allocate the
6459 "systemd-resolve" user on such systems, so that nss-ldap won't be
6460 triggered; or use a different NSS package that doesn't do networking
6461 in-process but provides a local asynchronous name cache; or configure
6462 the NSS package to avoid lookups for UIDs in the range `pkg-config
6463 systemd --variable=dynamicuidmin` … `pkg-config systemd
6464 --variable=dynamicuidmax`, so that it does not consider itself
6465 authoritative for the same UID range systemd allocates dynamic users
6468 * The systemd-resolve tool has been renamed to resolvectl (it also
6469 remains available under the old name, for compatibility), and its
6470 interface is now verb-based, similar in style to the other <xyz>ctl
6471 tools, such as systemctl or loginctl.
6473 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
6474 compatibility. It may be symlinked under the 'resolvconf' name, in
6475 which case it will take arguments and input compatible with the
6476 Debian and FreeBSD resolvconf tool.
6478 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
6479 where the system initially suspends, and after a timeout resumes and
6482 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
6483 set the client will only send a DUID as client identifier.
6485 * The nss-systemd glibc NSS module will now enumerate dynamic users and
6486 groups in effect. Previously, it could resolve UIDs/GIDs to user
6487 names/groups and vice versa, but did not support enumeration.
6489 * journald's Compress= configuration setting now optionally accepts a
6490 byte threshold value. All journal objects larger than this threshold
6491 will be compressed, smaller ones will not. Previously this threshold
6492 was not configurable and set to 512.
6494 * A new system.conf setting NoNewPrivileges= is now available which may
6495 be used to turn off acquisition of new privileges system-wide
6496 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
6497 for all its children). Note that turning this option on means setuid
6498 binaries and file system capabilities lose their special powers.
6499 While turning on this option is a big step towards a more secure
6500 system, doing so is likely to break numerous pre-existing UNIX tools,
6501 in particular su and sudo.
6503 * A new service systemd-time-sync-wait.service has been added. If
6504 enabled it will delay the time-sync.target unit at boot until time
6505 synchronization has been received from the network. This
6506 functionality is useful on systems lacking a local RTC or where it is
6507 acceptable that the boot process shall be delayed by external network
6510 * When hibernating, systemd will now inform the kernel of the image
6511 write offset, on kernels new enough to support this. This means swap
6512 files should work for hibernation now.
6514 * When loading unit files, systemd will now look for drop-in unit files
6515 extensions in additional places. Previously, for a unit file name
6516 "foo-bar-baz.service" it would look for dropin files in
6517 "foo-bar-baz.service.d/*.conf". Now, it will also look in
6518 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
6519 service name truncated after all inner dashes. This scheme allows
6520 writing drop-ins easily that apply to a whole set of unit files at
6521 once. It's particularly useful for mount and slice units (as their
6522 naming is prefix based), but is also useful for service and other
6523 units, for packages that install multiple unit files at once,
6524 following a strict naming regime of beginning the unit file name with
6525 the package's name. Two new specifiers are now supported in unit
6526 files to match this: %j and %J are replaced by the part of the unit
6527 name following the last dash.
6529 * Unit files and other configuration files that support specifier
6530 expansion now understand another three new specifiers: %T and %V will
6531 resolve to /tmp and /var/tmp respectively, or whatever temporary
6532 directory has been set for the calling user. %E will expand to either
6533 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
6535 * The ExecStart= lines of unit files are no longer required to
6536 reference absolute paths. If non-absolute paths are specified the
6537 specified binary name is searched within the service manager's
6538 built-in $PATH, which may be queried with 'systemd-path
6539 search-binaries-default'. It's generally recommended to continue to
6540 use absolute paths for all binaries specified in unit files.
6542 * Units gained a new load state "bad-setting", which is used when a
6543 unit file was loaded, but contained fatal errors which prevent it
6544 from being started (for example, a service unit has been defined
6545 lacking both ExecStart= and ExecStop= lines).
6547 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
6548 support alternative debuggers, for example lldb. The old name
6549 continues to be available however, for compatibility reasons. Use the
6550 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
6551 to pick an alternative debugger instead of the default gdb.
6553 * systemctl and the other tools will now output escape sequences that
6554 generate proper clickable hyperlinks in various terminal emulators
6555 where useful (for example, in the "systemctl status" output you can
6556 now click on the unit file name to quickly open it in the
6557 editor/viewer of your choice). Note that not all terminal emulators
6558 support this functionality yet, but many do. Unfortunately, the
6559 "less" pager doesn't support this yet, hence this functionality is
6560 currently automatically turned off when a pager is started (which
6561 happens quite often due to auto-paging). We hope to remove this
6562 limitation as soon as "less" learns these escape sequences. This new
6563 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
6564 environment variable. For details on these escape sequences see:
6565 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
6567 * networkd's .network files now support a new IPv6MTUBytes= option for
6568 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
6569 option in the [Route] section to configure the MTU to use for
6570 specific routes. It also gained support for configuration of the DHCP
6571 "UserClass" option through the new UserClass= setting. It gained
6572 three new options in the new [CAN] section for configuring CAN
6573 networks. The MULTICAST and ALLMULTI interface flags may now be
6574 controlled explicitly with the new Multicast= and AllMulticast=
6577 * networkd will now automatically make use of the kernel's route
6578 expiration feature, if it is available.
6580 * udevd's .link files now support setting the number of receive and
6581 transmit channels, using the RxChannels=, TxChannels=,
6582 OtherChannels=, CombinedChannels= settings.
6584 * Support for UDPSegmentationOffload= has been removed, given its
6585 limited support in hardware, and waning software support.
6587 * networkd's .netdev files now support creating "netdevsim" interfaces.
6589 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
6590 to query the unit belonging to a specific kernel control group.
6592 * systemd-analyze gained a new verb "cat-config", which may be used to
6593 dump the contents of any configuration file, with all its matching
6594 drop-in files added in, and honouring the usual search and masking
6595 logic applied to systemd configuration files. For example use
6596 "systemd-analyze cat-config systemd/system.conf" to get the complete
6597 system configuration file of systemd how it would be loaded by PID 1
6598 itself. Similar to this, various tools such as systemd-tmpfiles or
6599 systemd-sysusers, gained a new option "--cat-config", which does the
6600 corresponding operation for their own configuration settings. For
6601 example, "systemd-tmpfiles --cat-config" will now output the full
6602 list of tmpfiles.d/ lines in place.
6604 * timedatectl gained three new verbs: "show" shows bus properties of
6605 systemd-timedated, "timesync-status" shows the current NTP
6606 synchronization state of systemd-timesyncd, and "show-timesync"
6607 shows bus properties of systemd-timesyncd.
6609 * systemd-timesyncd gained a bus interface on which it exposes details
6612 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
6613 understood by systemd-timedated. It takes a colon-separated list of
6614 unit names of NTP client services. The list is used by
6615 "timedatectl set-ntp".
6617 * systemd-nspawn gained a new --rlimit= switch for setting initial
6618 resource limits for the container payload. There's a new switch
6619 --hostname= to explicitly override the container's hostname. A new
6620 --no-new-privileges= switch may be used to control the
6621 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
6622 --oom-score-adjust= switch controls the OOM scoring adjustment value
6623 for the payload. The new --cpu-affinity= switch controls the CPU
6624 affinity of the container payload. The new --resolv-conf= switch
6625 allows more detailed control of /etc/resolv.conf handling of the
6626 container. Similarly, the new --timezone= switch allows more detailed
6627 control of /etc/localtime handling of the container.
6629 * systemd-detect-virt gained a new --list switch, which will print a
6630 list of all currently known VM and container environments.
6632 * Support for "Portable Services" has been added, see
6633 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
6634 experimental, but this is expected to change soon. Reflecting this
6635 experimental state, the "portablectl" binary is not installed into
6636 /usr/bin yet. The binary has to be called with the full path
6637 /usr/lib/systemd/portablectl instead.
6639 * journalctl's and systemctl's -o switch now knows a new log output
6640 mode "with-unit". The output it generates is very similar to the
6641 regular "short" mode, but displays the unit name instead of the
6642 syslog tag for each log line. Also, the date is shown with timezone
6643 information. This mode is probably more useful than the classic
6644 "short" output mode for most purposes, except where pixel-perfect
6645 compatibility with classic /var/log/messages formatting is required.
6647 * A new --dump-bus-properties switch has been added to the systemd
6648 binary, which may be used to dump all supported D-Bus properties.
6649 (Options which are still supported, but are deprecated, are *not*
6652 * sd-bus gained a set of new calls:
6653 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
6654 enable/disable the "floating" state of a bus slot object,
6655 i.e. whether the slot object pins the bus it is allocated for into
6656 memory or if the bus slot object gets disconnected when the bus goes
6657 away. sd_bus_open_with_description(),
6658 sd_bus_open_user_with_description(),
6659 sd_bus_open_system_with_description() may be used to allocate bus
6660 objects and set their description string already during allocation.
6662 * sd-event gained support for watching inotify events from the event
6663 loop, in an efficient way, sharing inotify handles between multiple
6664 users. For this a new function sd_event_add_inotify() has been added.
6666 * sd-event and sd-bus gained support for calling special user-supplied
6667 destructor functions for userdata pointers associated with
6668 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
6669 functions sd_bus_slot_set_destroy_callback,
6670 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
6671 sd_bus_track_get_destroy_callback,
6672 sd_event_source_set_destroy_callback,
6673 sd_event_source_get_destroy_callback have been added.
6675 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
6677 * PID 1 will now automatically reschedule .timer units whenever the
6678 local timezone changes. (They previously got rescheduled
6679 automatically when the system clock changed.)
6681 * New documentation has been added to document cgroups delegation,
6682 portable services and the various code quality tools we have set up:
6684 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
6685 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
6686 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
6688 * The Boot Loader Specification has been added to the source tree.
6690 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
6692 While moving it into our source tree we have updated it and further
6693 changes are now accepted through the usual github PR workflow.
6695 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
6696 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
6697 earlier PAM modules. The data in these fields is used to initialize
6698 the session scope's resource properties. Thus external PAM modules
6699 may now configure per-session limits, for example sourced from
6700 external user databases.
6702 * socket units with Accept=yes will now maintain a "refused" counter in
6703 addition to the existing "accepted" counter, counting connections
6704 refused due to the enforced limits.
6706 * The "systemd-path search-binaries-default" command may now be use to
6707 query the default, built-in $PATH PID 1 will pass to the services it
6710 * A new unit file setting PrivateMounts= has been added. It's a boolean
6711 option. If enabled the unit's processes are invoked in their own file
6712 system namespace. Note that this behaviour is also implied if any
6713 other file system namespacing options (such as PrivateTmp=,
6714 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
6715 primarily useful for services that do not use any of the other file
6716 system namespacing options. One such service is systemd-udevd.service
6717 where this is now used by default.
6719 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
6720 when the system is booted in UEFI "secure mode".
6722 * A new unit "system-update-pre.target" is added, which defines an
6723 optional synchronization point for offline system updates, as
6724 implemented by the pre-existing "system-update.target" unit. It
6725 allows ordering services before the service that executes the actual
6726 update process in a generic way.
6728 * Systemd now emits warnings whenever .include syntax is used.
6730 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
6731 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
6732 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
6733 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
6734 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
6735 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
6736 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
6737 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
6738 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
6739 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
6740 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
6741 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
6742 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
6743 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
6744 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
6745 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
6746 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
6747 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
6748 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
6749 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
6750 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
6751 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
6752 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
6753 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
6754 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
6755 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
6756 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
6757 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
6758 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6760 — Berlin, 2018-06-22
6764 * The MemoryAccounting= unit property now defaults to on. After
6765 discussions with the upstream control group maintainers we learnt
6766 that the negative impact of cgroup memory accounting on current
6767 kernels is finally relatively minimal, so that it should be safe to
6768 enable this by default without affecting system performance. Besides
6769 memory accounting only task accounting is turned on by default, all
6770 other forms of resource accounting (CPU, IO, IP) remain off for now,
6771 because it's not clear yet that their impact is small enough to move
6772 from opt-in to opt-out. We recommend downstreams to leave memory
6773 accounting on by default if kernel 4.14 or higher is primarily
6774 used. On very resource constrained systems or when support for old
6775 kernels is a necessity, -Dmemory-accounting-default=false can be used
6776 to revert this change.
6778 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
6779 %udev_rules_update) and the journal catalog (%journal_catalog_update)
6780 from the upgrade scriptlets of individual packages now do nothing.
6781 Transfiletriggers have been added which will perform those updates
6782 once at the end of the transaction.
6784 Similar transfiletriggers have been added to execute any sysctl.d
6785 and binfmt.d rules. Thus, it should be unnecessary to provide any
6786 scriptlets to execute this configuration from package installation
6789 * systemd-sysusers gained a mode where the configuration to execute is
6790 specified on the command line, but this configuration is not executed
6791 directly, but instead it is merged with the configuration on disk,
6792 and the result is executed. This is useful for package installation
6793 scripts which want to create the user before installing any files on
6794 disk (in case some of those files are owned by that user), while
6795 still allowing local admin overrides.
6797 This functionality is exposed to rpm scriptlets through a new
6798 %sysusers_create_package macro. Old %sysusers_create and
6799 %sysusers_create_inline macros are deprecated.
6801 A transfiletrigger for sysusers.d configuration is now installed,
6802 which means that it should be unnecessary to call systemd-sysusers from
6803 package installation scripts, unless the package installs any files
6804 owned by those newly-created users, in which case
6805 %sysusers_create_package should be used.
6807 * Analogous change has been done for systemd-tmpfiles: it gained a mode
6808 where the command-line configuration is merged with the configuration
6809 on disk. This is exposed as the new %tmpfiles_create_package macro,
6810 and %tmpfiles_create is deprecated. A transfiletrigger is installed
6811 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
6812 from package installation scripts.
6814 * sysusers.d configuration for a user may now also specify the group
6815 number, in addition to the user number ("u username 123:456"), or
6816 without the user number ("u username -:456").
6818 * Configution items for systemd-sysusers can now be specified as
6819 positional arguments when the new --inline switch is used.
6821 * The login shell of users created through sysusers.d may now be
6822 specified (previously, it was always /bin/sh for root and
6823 /sbin/nologin for other users).
6825 * systemd-analyze gained a new --global switch to look at global user
6826 configuration. It also gained a unit-paths verb to list the unit load
6827 paths that are compiled into systemd (which can be used with
6828 --systemd, --user, or --global).
6830 * udevadm trigger gained a new --settle/-w option to wait for any
6831 triggered events to finish (but just those, and not any other events
6832 which are triggered meanwhile).
6834 * The action that systemd-logind takes when the lid is closed and the
6835 machine is connected to external power can now be configured using
6836 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
6837 was determined by HandleLidSwitch=, and, for backwards compatibility,
6838 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
6840 * journalctl will periodically call sd_journal_process() to make it
6841 resilient against inotify queue overruns when journal files are
6842 rotated very quickly.
6844 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
6845 sd_bus_get_n_queued_write — may be used to check the number of
6846 pending bus messages.
6848 * systemd gained a new
6849 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
6850 which can be used to migrate foreign processes to scope and service
6851 units. The primary user for this new API is systemd itself: the
6852 systemd --user instance uses this call of the systemd --system
6853 instance to migrate processes if it itself gets the request to
6854 migrate processes and the kernel refuses this due to access
6855 restrictions. Thanks to this "systemd-run --scope --user …" works
6856 again in pure cgroup v2 environments when invoked from the user
6859 * A new TemporaryFileSystem= setting can be used to mask out part of
6860 the real file system tree with tmpfs mounts. This may be combined
6861 with BindPaths= and BindReadOnlyPaths= to hide files or directories
6862 not relevant to the unit, while still allowing some paths lower in
6863 the tree to be accessed.
6865 ProtectHome=tmpfs may now be used to hide user home and runtime
6866 directories from units, in a way that is mostly equivalent to
6867 "TemporaryFileSystem=/home /run/user /root".
6869 * Non-service units are now started with KeyringMode=shared by default.
6870 This means that mount and swapon and other mount tools have access
6871 to keys in the main keyring.
6873 * /sys/fs/bpf is now mounted automatically.
6875 * QNX virtualization is now detected by systemd-detect-virt and may
6876 be used in ConditionVirtualization=.
6878 * IPAccounting= may now be enabled also for slice units.
6880 * A new -Dsplit-bin= build configuration switch may be used to specify
6881 whether bin and sbin directories are merged, or if they should be
6882 included separately in $PATH and various listings of executable
6883 directories. The build configuration scripts will try to autodetect
6884 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
6885 system, but distributions are encouraged to configure this
6888 * A new -Dok-color= build configuration switch may be used to change
6889 the colour of "OK" status messages.
6891 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
6892 PrivateNetwork=yes was buggy in previous versions of systemd. This
6893 means that after the upgrade and daemon-reexec, any such units must
6896 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
6897 will not exclude read-only files owned by root from cleanup.
6899 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
6900 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
6901 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
6902 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
6903 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
6904 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
6905 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
6906 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
6907 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
6908 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
6909 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
6910 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
6911 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
6912 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
6913 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
6914 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
6916 — Warsaw, 2018-03-05
6920 * Some keyboards come with a zoom see-saw or rocker which until now got
6921 mapped to the Linux "zoomin/out" keys in hwdb. However, these
6922 keycodes are not recognized by any major desktop. They now produce
6923 Up/Down key events so that they can be used for scrolling.
6925 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
6926 slightly: previously, if an argument was specified for lines of this
6927 type (i.e. the right-most column was set) this string was appended to
6928 existing files each time systemd-tmpfiles was run. This behaviour was
6929 different from what the documentation said, and not particularly
6930 useful, as repeated systemd-tmpfiles invocations would not be
6931 idempotent and grow such files without bounds. With this release
6932 behaviour has been altered to match what the documentation says:
6933 lines of this type only have an effect if the indicated files don't
6934 exist yet, and only then the argument string is written to the file.
6936 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
6937 systemd-tmpfiles behaviour: previously, read-only files owned by root
6938 were always excluded from the file "aging" algorithm (i.e. the
6939 automatic clean-up of directories like /tmp based on
6940 atime/mtime/ctime). We intend to drop this restriction, and age files
6941 by default even when owned by root and read-only. This behaviour was
6942 inherited from older tools, but there have been requests to remove
6943 it, and it's not obvious why this restriction was made in the first
6944 place. Please speak up now, if you are aware of software that requires
6945 this behaviour, otherwise we'll remove the restriction in v238.
6947 * A new environment variable $SYSTEMD_OFFLINE is now understood by
6948 systemctl. It takes a boolean argument. If on, systemctl assumes it
6949 operates on an "offline" OS tree, and will not attempt to talk to the
6950 service manager. Previously, this mode was implicitly enabled if a
6951 chroot() environment was detected, and this new environment variable
6952 now provides explicit control.
6954 * .path and .socket units may now be created transiently, too.
6955 Previously only service, mount, automount and timer units were
6956 supported as transient units. The systemd-run tool has been updated
6957 to expose this new functionality, you may hence use it now to bind
6958 arbitrary commands to path or socket activation on-the-fly from the
6959 command line. Moreover, almost all properties are now exposed for the
6960 unit types that already supported transient operation.
6962 * The systemd-mount command gained support for a new --owner= parameter
6963 which takes a user name, which is then resolved and included in uid=
6964 and gid= mount options string of the file system to mount.
6966 * A new unit condition ConditionControlGroupController= has been added
6967 that checks whether a specific cgroup controller is available.
6969 * Unit files, udev's .link files, and systemd-networkd's .netdev and
6970 .network files all gained support for a new condition
6971 ConditionKernelVersion= for checking against specific kernel
6974 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
6975 support for configuring device flags in the Flags= setting. In the
6976 same files, the [Tunnel] section gained support for configuring
6977 AllowLocalRemote=. The [Route] section in .network files gained
6978 support for configuring InitialCongestionWindow=,
6979 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
6980 understands RapidCommit=.
6982 * systemd-networkd's DHCPv6 support gained support for Prefix
6985 * sd-bus gained support for a new "watch-bind" feature. When this
6986 feature is enabled, an sd_bus connection may be set up to connect to
6987 an AF_UNIX socket in the file system as soon as it is created. This
6988 functionality is useful for writing early-boot services that
6989 automatically connect to the system bus as soon as it is started,
6990 without ugly time-based polling. systemd-networkd and
6991 systemd-resolved have been updated to make use of this
6992 functionality. busctl exposes this functionality in a new
6993 --watch-bind= command line switch.
6995 * sd-bus will now optionally synthesize a local "Connected" signal as
6996 soon as a D-Bus connection is set up fully. This message mirrors the
6997 already existing "Disconnected" signal which is synthesized when the
6998 connection is terminated. This signal is generally useful but
6999 particularly handy in combination with the "watch-bind" feature
7000 described above. Synthesizing of this message has to be requested
7001 explicitly through the new API call sd_bus_set_connected_signal(). In
7002 addition a new call sd_bus_is_ready() has been added that checks
7003 whether a connection is fully set up (i.e. between the "Connected" and
7004 "Disconnected" signals).
7006 * sd-bus gained two new calls sd_bus_request_name_async() and
7007 sd_bus_release_name_async() for asynchronously registering bus
7008 names. Similar, there is now sd_bus_add_match_async() for installing
7009 a signal match asynchronously. All of systemd's own services have
7010 been updated to make use of these calls. Doing these operations
7011 asynchronously has two benefits: it reduces the risk of deadlocks in
7012 case of cyclic dependencies between bus services, and it speeds up
7013 service initialization since synchronization points for bus
7014 round-trips are removed.
7016 * sd-bus gained two new calls sd_bus_match_signal() and
7017 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
7018 and sd_bus_add_match_async() but instead of taking a D-Bus match
7019 string take match fields as normal function parameters.
7021 * sd-bus gained two new calls sd_bus_set_sender() and
7022 sd_bus_message_set_sender() for setting the sender name of outgoing
7023 messages (either for all outgoing messages or for just one specific
7024 one). These calls are only useful in direct connections as on
7025 brokered connections the broker fills in the sender anyway,
7026 overwriting whatever the client filled in.
7028 * sd-event gained a new pseudo-handle that may be specified on all API
7029 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
7030 used this refers to the default event loop object of the calling
7031 thread. Note however that this does not implicitly allocate one —
7032 which has to be done prior by using sd_event_default(). Similarly
7033 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
7034 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
7035 to the default bus of the specified type of the calling thread. Here
7036 too this does not implicitly allocate bus connection objects, this
7037 has to be done prior with sd_bus_default() and friends.
7039 * sd-event gained a new call pair
7040 sd_event_source_{get|set}_io_fd_own(). This may be used to request
7041 automatic closure of the file descriptor an IO event source watches
7042 when the event source is destroyed.
7044 * systemd-networkd gained support for natively configuring WireGuard
7047 * In previous versions systemd synthesized user records both for the
7048 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
7049 internally. In order to simplify distribution-wide renames of the
7050 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
7051 new transitional flag file has been added: if
7052 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
7053 user and group record within the systemd codebase is disabled.
7055 * systemd-notify gained a new --uid= option for selecting the source
7056 user/UID to use for notification messages sent to the service
7059 * journalctl gained a new --grep= option to list only entries in which
7060 the message matches a certain pattern. By default matching is case
7061 insensitive if the pattern is lowercase, and case sensitive
7062 otherwise. Option --case-sensitive=yes|no can be used to override
7063 this an specify case sensitivity or case insensitivity.
7065 * There's now a "systemd-analyze service-watchdogs" command for printing
7066 the current state of the service runtime watchdog, and optionally
7067 enabling or disabling the per-service watchdogs system-wide if given a
7068 boolean argument (i.e. the concept you configure in WatchdogSec=), for
7069 debugging purposes. There's also a kernel command line option
7070 systemd.service_watchdogs= for controlling the same.
7072 * Two new "log-level" and "log-target" options for systemd-analyze were
7073 added that merge the now deprecated get-log-level, set-log-level and
7074 get-log-target, set-log-target pairs. The deprecated options are still
7075 understood for backwards compatibility. The two new options print the
7076 current value when no arguments are given, and set them when a
7077 level/target is given as an argument.
7079 * sysusers.d's "u" lines now optionally accept both a UID and a GID
7080 specification, separated by a ":" character, in order to create users
7081 where UID and GID do not match.
7083 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
7084 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
7085 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
7086 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
7087 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
7088 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
7089 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
7090 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
7091 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
7092 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
7093 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
7094 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
7095 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
7096 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
7097 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
7098 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
7099 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
7100 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
7101 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
7102 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
7109 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
7110 in v235 has been extended to also set the dummy.ko module option
7111 numdummies=0, preventing the kernel from automatically creating
7112 dummy0. All dummy interfaces must now be explicitly created.
7114 * Unknown '%' specifiers in configuration files are now rejected. This
7115 applies to units and tmpfiles.d configuration. Any percent characters
7116 that are followed by a letter or digit that are not supposed to be
7117 interpreted as the beginning of a specifier should be escaped by
7118 doubling ("%%"). (So "size=5%" is still accepted, as well as
7119 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
7120 valid specifiers today.)
7122 * systemd-resolved now maintains a new dynamic
7123 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
7124 recommended to make /etc/resolv.conf a symlink to it. This file
7125 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
7126 includes dynamically acquired search domains, achieving more correct
7127 DNS resolution by software that bypasses local DNS APIs such as NSS.
7129 * The "uaccess" udev tag has been dropped from /dev/kvm and
7130 /dev/dri/renderD*. These devices now have the 0666 permissions by
7131 default (but this may be changed at build-time). /dev/dri/renderD*
7132 will now be owned by the "render" group along with /dev/kfd.
7134 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
7135 systemd-journal-gatewayd.service and
7136 systemd-journal-upload.service. This means "nss-systemd" must be
7137 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
7138 services are resolved properly.
7140 * In /etc/fstab two new mount options are now understood:
7141 x-systemd.makefs and x-systemd.growfs. The former has the effect that
7142 the configured file system is formatted before it is mounted, the
7143 latter that the file system is resized to the full block device size
7144 after it is mounted (i.e. if the file system is smaller than the
7145 partition it resides on, it's grown). This is similar to the fsck
7146 logic in /etc/fstab, and pulls in systemd-makefs@.service and
7147 systemd-growfs@.service as necessary, similar to
7148 systemd-fsck@.service. Resizing is currently only supported on ext4
7151 * In systemd-networkd, the IPv6 RA logic now optionally may announce
7152 DNS server and domain information.
7154 * Support for the LUKS2 on-disk format for encrypted partitions has
7155 been added. This requires libcryptsetup2 during compilation and
7158 * The systemd --user instance will now signal "readiness" when its
7159 basic.target unit has been reached, instead of when the run queue ran
7160 empty for the first time.
7162 * Tmpfiles.d with user configuration are now also supported.
7163 systemd-tmpfiles gained a new --user switch, and snippets placed in
7164 ~/.config/user-tmpfiles.d/ and corresponding directories will be
7165 executed by systemd-tmpfiles --user running in the new
7166 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
7167 running in the user session.
7169 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
7170 %S resolves to the top-level state directory (/var/lib for the system
7171 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
7172 top-level cache directory (/var/cache for the system instance,
7173 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
7174 logs directory (/var/log for the system instance,
7175 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
7176 existing %t specifier, that resolves to the top-level runtime
7177 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
7180 * journalctl learnt a new parameter --output-fields= for limiting the
7181 set of journal fields to output in verbose and JSON output modes.
7183 * systemd-timesyncd's configuration file gained a new option
7184 RootDistanceMaxSec= for setting the maximum root distance of servers
7185 it'll use, as well as the new options PollIntervalMinSec= and
7186 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
7188 * bootctl gained a new command "list" for listing all available boot
7189 menu items on systems that follow the boot loader specification.
7191 * systemctl gained a new --dry-run switch that shows what would be done
7192 instead of doing it, and is currently supported by the shutdown and
7195 * ConditionSecurity= can now detect the TOMOYO security module.
7197 * Unit file [Install] sections are now also respected in unit drop-in
7198 files. This is intended to be used by drop-ins under /usr/lib/.
7200 * systemd-firstboot may now also set the initial keyboard mapping.
7202 * Udev "changed" events for devices which are exposed as systemd
7203 .device units are now propagated to units specified in
7204 ReloadPropagatedFrom= as reload requests.
7206 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
7207 unit template name (i.e. a name in the form of 'foobar@.service',
7208 without the instance component between the '@' and - the '.'), then
7209 the escaped sysfs path of the device is automatically used as the
7212 * SystemCallFilter= in unit files has been extended so that an "errno"
7213 can be specified individually for each system call. Example:
7214 SystemCallFilter=~uname:EILSEQ.
7216 * The cgroup delegation logic has been substantially updated. Delegate=
7217 now optionally takes a list of controllers (instead of a boolean, as
7218 before), which lists the controllers to delegate at least.
7220 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
7222 * A new LogLevelMax= setting configures the maximum log level any
7223 process of the service may log at (i.e. anything with a lesser
7224 priority than what is specified is automatically dropped). A new
7225 LogExtraFields= setting allows configuration of additional journal
7226 fields to attach to all log records generated by any of the unit's
7229 * New StandardInputData= and StandardInputText= settings along with the
7230 new option StandardInput=data may be used to configure textual or
7231 binary data that shall be passed to the executed service process via
7232 standard input, encoded in-line in the unit file.
7234 * StandardInput=, StandardOutput= and StandardError= may now be used to
7235 connect stdin/stdout/stderr of executed processes directly with a
7236 file or AF_UNIX socket in the file system, using the new "file:" option.
7238 * A new unit file option CollectMode= has been added, that allows
7239 tweaking the garbage collection logic for units. It may be used to
7240 tell systemd to garbage collect units that have failed automatically
7241 (normally it only GCs units that exited successfully). systemd-run
7242 and systemd-mount expose this new functionality with a new -G option.
7244 * "machinectl bind" may now be used to bind mount non-directories
7245 (i.e. regularfiles, devices, fifos, sockets).
7247 * systemd-analyze gained a new verb "calendar" for validating and
7248 testing calendar time specifications to use for OnCalendar= in timer
7249 units. Besides validating the expression it will calculate the next
7250 time the specified expression would elapse.
7252 * In addition to the pre-existing FailureAction= unit file setting
7253 there's now SuccessAction=, for configuring a shutdown action to
7254 execute when a unit completes successfully. This is useful in
7255 particular inside containers that shall terminate after some workload
7256 has been completed. Also, both options are now supported for all unit
7257 types, not just services.
7259 * networkds's IP rule support gained two new options
7260 IncomingInterface= and OutgoingInterface= for configuring the incoming
7261 and outgoing interfaces of configured rules. systemd-networkd also
7262 gained support for "vxcan" network devices.
7264 * networkd gained a new setting RequiredForOnline=, taking a
7265 boolean. If set, systemd-wait-online will take it into consideration
7266 when determining that the system is up, otherwise it will ignore the
7267 interface for this purpose.
7269 * The sd_notify() protocol gained support for a new operation: with
7270 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
7271 store again, ahead of POLLHUP or POLLERR when they are removed
7274 * A new document doc/UIDS-GIDS.md has been added to the source tree,
7275 that documents the UID/GID range and assignment assumptions and
7276 requirements of systemd.
7278 * The watchdog device PID 1 will ping may now be configured through the
7279 WatchdogDevice= configuration file setting, or by setting the
7280 systemd.watchdog_service= kernel commandline option.
7282 * systemd-resolved's gained support for registering DNS-SD services on
7283 the local network using MulticastDNS. Services may either be
7284 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
7285 the same dir below /run, /usr/lib), or through its D-Bus API.
7287 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
7288 extend the effective start, runtime, and stop time. The service must
7289 continue to send EXTEND_TIMEOUT_USEC within the period specified to
7290 prevent the service manager from making the service as timedout.
7292 * systemd-resolved's DNSSEC support gained support for RFC 8080
7293 (Ed25519 keys and signatures).
7295 * The systemd-resolve command line tool gained a new set of options
7296 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
7297 --set-nta= and --revert to configure per-interface DNS configuration
7298 dynamically during runtime. It's useful for pushing DNS information
7299 into systemd-resolved from DNS hook scripts that various interface
7300 managing software supports (such as pppd).
7302 * systemd-nspawn gained a new --network-namespace-path= command line
7303 option, which may be used to make a container join an existing
7304 network namespace, by specifying a path to a "netns" file.
7306 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
7307 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
7308 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
7309 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
7310 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
7311 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
7312 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
7313 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
7314 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
7315 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
7316 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
7317 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
7318 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
7319 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
7320 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
7321 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
7322 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
7323 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
7324 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
7325 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
7326 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
7327 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
7328 Jędrzejewski-Szmek, Zeal Jagannatha
7330 — Berlin, 2017-12-14
7334 * INCOMPATIBILITY: systemd-logind.service and other long-running
7335 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
7336 communication with the outside. This generally improves security of
7337 the system, and is in almost all cases a safe and good choice, as
7338 these services do not and should not provide any network-facing
7339 functionality. However, systemd-logind uses the glibc NSS API to
7340 query the user database. This creates problems on systems where NSS
7341 is set up to directly consult network services for user database
7342 lookups. In particular, this creates incompatibilities with the
7343 "nss-nis" module, which attempts to directly contact the NIS/YP
7344 network servers it is configured for, and will now consistently
7345 fail. In such cases, it is possible to turn off IP sandboxing for
7346 systemd-logind.service (set IPAddressDeny= in its [Service] section
7347 to the empty string, via a .d/ unit file drop-in). Downstream
7348 distributions might want to update their nss-nis packaging to include
7349 such a drop-in snippet, accordingly, to hide this incompatibility
7350 from the user. Another option is to make use of glibc's nscd service
7351 to proxy such network requests through a privilege-separated, minimal
7352 local caching daemon, or to switch to more modern technologies such
7353 sssd, whose NSS hook-ups generally do not involve direct network
7354 access. In general, we think it's definitely time to question the
7355 implementation choices of nss-nis, i.e. whether it's a good idea
7356 today to embed a network-facing loadable module into all local
7357 processes that need to query the user database, including the most
7358 trivial and benign ones, such as "ls". For more details about
7359 IPAddressDeny= see below.
7361 * A new modprobe.d drop-in is now shipped by default that sets the
7362 bonding module option max_bonds=0. This overrides the kernel default,
7363 to avoid conflicts and ambiguity as to whether or not bond0 should be
7364 managed by systemd-networkd or not. This resolves multiple issues
7365 with bond0 properties not being applied, when bond0 is configured
7366 with systemd-networkd. Distributors may choose to not package this,
7367 however in that case users will be prevented from correctly managing
7368 bond0 interface using systemd-networkd.
7370 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
7371 which print the logging level and target of the system manager. They
7372 complement the existing "set-log-level" and "set-log-target" verbs
7373 used to change those values.
7375 * journald.conf gained a new boolean setting ReadKMsg= which defaults
7376 to on. If turned off kernel log messages will not be read by
7377 systemd-journald or included in the logs. It also gained a new
7378 setting LineMax= for configuring the maximum line length in
7379 STDOUT/STDERR log streams. The new default for this value is 48K, up
7380 from the previous hardcoded 2048.
7382 * A new unit setting RuntimeDirectoryPreserve= has been added, which
7383 allows more detailed control of what to do with a runtime directory
7384 configured with RuntimeDirectory= (i.e. a directory below /run or
7385 $XDG_RUNTIME_DIR) after a unit is stopped.
7387 * The RuntimeDirectory= setting for units gained support for creating
7388 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
7389 one top-level directory.
7391 * Units gained new options StateDirectory=, CacheDirectory=,
7392 LogsDirectory= and ConfigurationDirectory= which are closely related
7393 to RuntimeDirectory= but manage per-service directories below
7394 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
7395 possible to write unit files which when activated automatically gain
7396 properly owned service specific directories in these locations, thus
7397 making unit files self-contained and increasing compatibility with
7398 stateless systems and factory reset where /etc or /var are
7399 unpopulated at boot. Matching these new settings there's also
7400 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
7401 ConfigurationDirectoryMode= for configuring the access mode of these
7402 directories. These settings are particularly useful in combination
7403 with DynamicUser=yes as they provide secure, properly-owned,
7404 writable, and stateful locations for storage, excluded from the
7405 sandbox that such services live in otherwise.
7407 * Automake support has been removed from this release. systemd is now
7410 * systemd-journald will now aggressively cache client metadata during
7411 runtime, speeding up log write performance under pressure. This comes
7412 at a small price though: as much of the metadata is read
7413 asynchronously from /proc/ (and isn't implicitly attached to log
7414 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
7415 metadata stored alongside a log entry might be slightly
7416 out-of-date. Previously it could only be slightly newer than the log
7417 message. The time window is small however, and given that the kernel
7418 is unlikely to be improved anytime soon in this regard, this appears
7421 * nss-myhostname/systemd-resolved will now by default synthesize an
7422 A/AAAA resource record for the "_gateway" hostname, pointing to the
7423 current default IP gateway. Previously it did that for the "gateway"
7424 name, hampering adoption, as some distributions wanted to leave that
7425 hostname open for local use. The old behaviour may still be
7426 requested at build time.
7428 * systemd-networkd's [Address] section in .network files gained a new
7429 Scope= setting for configuring the IP address scope. The [Network]
7430 section gained a new boolean setting ConfigureWithoutCarrier= that
7431 tells systemd-networkd to ignore link sensing when configuring the
7432 device. The [DHCP] section gained a new Anonymize= boolean option for
7433 turning on a number of options suggested in RFC 7844. A new
7434 [RoutingPolicyRule] section has been added for configuring the IP
7435 routing policy. The [Route] section has gained support for a new
7436 Type= setting which permits configuring
7437 blackhole/unreachable/prohibit routes.
7439 * The [VRF] section in .netdev files gained a new Table= setting for
7440 configuring the routing table to use. The [Tunnel] section gained a
7441 new Independent= boolean field for configuring tunnels independent of
7442 an underlying network interface. The [Bridge] section gained a new
7443 GroupForwardMask= option for configuration of propagation of link
7444 local frames between bridge ports.
7446 * The WakeOnLan= setting in .link files gained support for a number of
7447 new modes. A new TCP6SegmentationOffload= setting has been added for
7448 configuring TCP/IPv6 hardware segmentation offload.
7450 * The IPv6 RA sender implementation may now optionally send out RDNSS
7451 and RDNSSL records to supply DNS configuration to peers.
7453 * systemd-nspawn gained support for a new --system-call-filter= command
7454 line option for adding and removing entries in the default system
7455 call filter it applies. Moreover systemd-nspawn has been changed to
7456 implement a system call allow list instead of a deny list.
7458 * systemd-run gained support for a new --pipe command line option. If
7459 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
7460 are directly passed on to the activated transient service
7461 executable. This allows invoking arbitrary processes as systemd
7462 services (for example to take benefit of dependency management,
7463 accounting management, resource management or log management that is
7464 done automatically for services) — while still allowing them to be
7465 integrated in a classic UNIX shell pipeline.
7467 * When a service sends RELOAD=1 via sd_notify() and reload propagation
7468 using ReloadPropagationTo= is configured, a reload is now propagated
7469 to configured units. (Previously this was only done on explicitly
7470 requested reloads, using "systemctl reload" or an equivalent
7473 * For each service unit a restart counter is now kept: it is increased
7474 each time the service is restarted due to Restart=, and may be
7475 queried using "systemctl show -p NRestarts …".
7477 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
7478 @signal and @timer have been added, for usage with SystemCallFilter=
7479 in unit files and the new --system-call-filter= command line option
7480 of systemd-nspawn (see above).
7482 * ExecStart= lines in unit files gained two new modifiers: when a
7483 command line is prefixed with "!" the command will be executed as
7484 configured, except for the credentials applied by
7485 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
7486 "+", but does still apply namespacing options unlike "+". There's
7487 also "!!" now, which is mostly identical, but becomes a NOP on
7488 systems that support ambient capabilities. This is useful to write
7489 unit files that work with ambient capabilities where possible but
7490 automatically fall back to traditional privilege dropping mechanisms
7491 on systems where this is not supported.
7493 * ListenNetlink= settings in socket units now support RDMA netlink
7496 * A new unit file setting LockPersonality= has been added which permits
7497 locking down the chosen execution domain ("personality") of a service
7500 * A new special target "getty-pre.target" has been added, which is
7501 ordered before all text logins, and may be used to order services
7502 before textual logins acquire access to the console.
7504 * systemd will now attempt to load the virtio-rng.ko kernel module very
7505 early on if a VM environment supporting this is detected. This should
7506 improve entropy during early boot in virtualized environments.
7508 * A _netdev option is now supported in /etc/crypttab that operates in a
7509 similar way as the same option in /etc/fstab: it permits configuring
7510 encrypted devices that need to be ordered after the network is up.
7511 Following this logic, two new special targets
7512 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
7513 added that are to cryptsetup.target what remote-fs.target and
7514 remote-fs-pre.target are to local-fs.target.
7516 * Service units gained a new UnsetEnvironment= setting which permits
7517 unsetting specific environment variables for services that are
7518 normally passed to it (for example in order to mask out locale
7519 settings for specific services that can't deal with it).
7521 * Units acquired a new boolean option IPAccounting=. When turned on, IP
7522 traffic accounting (packet count as well as byte count) is done for
7523 the service, and shown as part of "systemctl status" or "systemd-run
7526 * Service units acquired two new options IPAddressAllow= and
7527 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
7528 for configuring a simple IP access control list for all sockets of
7529 the unit. These options are available also on .slice and .socket
7530 units, permitting flexible access list configuration for individual
7531 services as well as groups of services (as defined by a slice unit),
7532 including system-wide. Note that IP ACLs configured this way are
7533 enforced on every single IPv4 and IPv6 socket created by any process
7534 of the service unit, and apply to ingress as well as egress traffic.
7536 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
7537 structured log message is generated each time the unit is stopped,
7538 containing information about the consumed resources of this
7541 * A new setting KeyringMode= has been added to unit files, which may be
7542 used to control how the kernel keyring is set up for executed
7545 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
7546 "systemctl kexec" and "systemctl exit" are now always asynchronous in
7547 behaviour (that is: these commands return immediately after the
7548 operation was enqueued instead of waiting for the operation to
7549 complete). Previously, "systemctl poweroff" and "systemctl reboot"
7550 were asynchronous on systems using systemd-logind (i.e. almost
7551 always, and like they were on sysvinit), and the other three commands
7552 were unconditionally synchronous. With this release this is cleaned
7553 up, and callers will see the same asynchronous behaviour on all
7554 systems for all five operations.
7556 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
7559 * .timer units now accept calendar specifications in other timezones
7560 than UTC or the local timezone.
7562 * The tmpfiles snippet var.conf has been changed to create
7563 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
7564 the "utmp" group already, and it appears to be generally understood
7565 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
7566 databases. Previously this was implemented correctly for all these
7567 databases excepts btmp, which has been opened up like this now
7568 too. Note that while the other databases are world-readable
7569 (i.e. 0644), btmp is not and remains more restrictive.
7571 * The systemd-resolve tool gained a new --reset-server-features
7572 switch. When invoked like this systemd-resolved will forget
7573 everything it learnt about the features supported by the configured
7574 upstream DNS servers, and restarts the feature probing logic on the
7575 next resolver look-up for them at the highest feature level
7578 * The status dump systemd-resolved sends to the logs upon receiving
7579 SIGUSR1 now also includes information about all DNS servers it is
7580 configured to use, and the features levels it probed for them.
7582 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
7583 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
7584 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
7585 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
7586 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
7587 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
7588 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
7589 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
7590 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
7591 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
7592 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
7593 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
7594 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
7595 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
7596 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
7597 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
7598 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
7599 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
7600 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
7601 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
7603 — Berlin, 2017-10-06
7607 * Meson is now supported as build system in addition to Automake. It is
7608 our plan to remove Automake in one of our next releases, so that
7609 Meson becomes our exclusive build system. Hence, please start using
7610 the Meson build system in your downstream packaging. There's plenty
7611 of documentation around how to use Meson, the extremely brief
7614 ./autogen.sh && ./configure && make && sudo make install
7618 meson build && ninja -C build && sudo ninja -C build install
7620 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
7621 which permits configuring a timeout on the time a job is
7622 running. This is particularly useful for setting timeouts on jobs for
7625 * Unit files gained two new options ConditionUser= and ConditionGroup=
7626 for conditionalizing units based on the identity of the user/group
7627 running a systemd user instance.
7629 * systemd-networkd now understands a new FlowLabel= setting in the
7630 [VXLAN] section of .network files, as well as a Priority= in
7631 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
7632 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
7633 gained support for configuration of GENEVE links, and IPv6 address
7634 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
7636 * .link files now understand a new Port= setting.
7638 * systemd-networkd's DHCP support gained support for DHCP option 119
7639 (domain search list).
7641 * systemd-networkd gained support for serving IPv6 address ranges using
7642 the Router Advertisement protocol. The new .network configuration
7643 section [IPv6Prefix] may be used to configure the ranges to
7644 serve. This is implemented based on a new, minimal, native server
7645 implementation of RA.
7647 * journalctl's --output= switch gained support for a new parameter
7648 "short-iso-precise" for a mode where timestamps are shown as precise
7651 * systemd-udevd's "net_id" builtin may now generate stable network
7652 interface names from IBM PowerVM VIO devices as well as ACPI platform
7655 * MulticastDNS support in systemd-resolved may now be explicitly
7656 enabled/disabled using the new MulticastDNS= configuration file
7659 * systemd-resolved may now optionally use libidn2 instead of the libidn
7660 for processing internationalized domain names. Support for libidn2
7661 should be considered experimental and should not be enabled by
7664 * "machinectl pull-tar" and related call may now do verification of
7665 downloaded images using SUSE-style .sha256 checksum files in addition
7666 to the already existing support for validating using Ubuntu-style
7669 * sd-bus gained support for a new sd_bus_message_appendv() call which
7670 is va_list equivalent of sd_bus_message_append().
7672 * sd-boot gained support for validating images using SHIM/MOK.
7674 * The SMACK code learnt support for "onlycap".
7676 * systemd-mount --umount is now much smarter in figuring out how to
7677 properly unmount a device given its mount or device path.
7679 * The code to call libnss_dns as a fallback from libnss_resolve when
7680 the communication with systemd-resolved fails was removed. This
7681 fallback was redundant and interfered with the [!UNAVAIL=return]
7682 suffix. See nss-resolve(8) for the recommended configuration.
7684 * systemd-logind may now be restarted without losing state. It stores
7685 the file descriptors for devices it manages in the system manager
7686 using the FDSTORE= mechanism. Please note that further changes in
7687 other components may be required to make use of this (for example
7688 Xorg has code to listen for stops of systemd-logind and terminate
7689 itself when logind is stopped or restarted, in order to avoid using
7690 stale file descriptors for graphical devices, which is now
7691 counterproductive and must be reverted in order for restarts of
7692 systemd-logind to be safe. See
7693 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
7695 * All kernel-install plugins are called with the environment variable
7696 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
7697 /etc/machine-id. If the machine ID could not be determined,
7698 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
7699 anything in the entry directory (passed as the second argument) if
7700 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
7701 temporary directory is passed as the entry directory and removed
7702 after all the plugins exit.
7704 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
7705 will now use its value as the machine ID instead of the machine ID
7706 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
7707 /etc/machine-info and no machine ID is set in /etc/machine-id,
7708 kernel-install will try to store the current machine ID there as
7709 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
7710 will generate a new UUID, store it in /etc/machine-info as
7711 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
7713 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
7714 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
7715 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
7716 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
7717 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
7718 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
7719 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
7720 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
7721 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
7722 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
7723 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
7724 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
7725 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
7726 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
7727 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
7728 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
7729 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
7730 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
7731 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
7732 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
7733 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
7734 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
7735 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
7736 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
7737 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
7738 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
7739 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
7740 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
7741 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
7744 — Berlin, 2017-07-12
7748 * The "hybrid" control group mode has been modified to improve
7749 compatibility with "legacy" cgroups-v1 setups. Specifically, the
7750 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
7751 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
7752 cgroups-v1 hierarchy), the only externally visible change being that
7753 the cgroups-v2 hierarchy is also mounted, to
7754 /sys/fs/cgroup/unified. This should provide a large degree of
7755 compatibility with "legacy" cgroups-v1, while taking benefit of the
7756 better management capabilities of cgroups-v2.
7758 * The default control group setup mode may be selected both a boot-time
7759 via a set of kernel command line parameters (specifically:
7760 systemd.unified_cgroup_hierarchy= and
7761 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
7762 default selected on the configure command line
7763 (--with-default-hierarchy=). The upstream default is "hybrid"
7764 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
7765 this will change in a future systemd version to be "unified" (pure
7766 cgroups-v2 mode). The third option for the compile time option is
7767 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
7768 distributions to default to "hybrid" mode for release distributions,
7769 starting with v233. We recommend "unified" for development
7770 distributions (specifically: distributions such as Fedora's rawhide)
7771 as that's where things are headed in the long run. Use "legacy" for
7772 greatest stability and compatibility only.
7774 * Note one current limitation of "unified" and "hybrid" control group
7775 setup modes: the kernel currently does not permit the systemd --user
7776 instance (i.e. unprivileged code) to migrate processes between two
7777 disconnected cgroup subtrees, even if both are managed and owned by
7778 the user. This effectively means "systemd-run --user --scope" doesn't
7779 work when invoked from outside of any "systemd --user" service or
7780 scope. Specifically, it is not supported from session scopes. We are
7781 working on fixing this in a future systemd version. (See #3388 for
7782 further details about this.)
7784 * DBus policy files are now installed into /usr rather than /etc. Make
7785 sure your system has dbus >= 1.9.18 running before upgrading to this
7786 version, or override the install path with --with-dbuspolicydir= .
7788 * All python scripts shipped with systemd (specifically: the various
7789 tests written in Python) now require Python 3.
7791 * systemd unit tests can now run standalone (without the source or
7792 build directories), and can be installed into /usr/lib/systemd/tests/
7793 with 'make install-tests'.
7795 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
7796 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
7799 * Support for the %c, %r, %R specifiers in unit files has been
7800 removed. Specifiers are not supposed to be dependent on configuration
7801 in the unit file itself (so that they resolve the same regardless
7802 where used in the unit files), but these specifiers were influenced
7803 by the Slice= option.
7805 * The shell invoked by debug-shell.service now defaults to /bin/sh in
7806 all cases. If distributions want to use a different shell for this
7807 purpose (for example Fedora's /sbin/sushell) they need to specify
7808 this explicitly at configure time using --with-debug-shell=.
7810 * The confirmation spawn prompt has been reworked to offer the
7813 (c)ontinue, proceed without asking anymore
7814 (D)ump, show the state of the unit
7815 (f)ail, don't execute the command and pretend it failed
7817 (i)nfo, show a short summary of the unit
7818 (j)obs, show jobs that are in progress
7819 (s)kip, don't execute the command and pretend it succeeded
7820 (y)es, execute the command
7822 The 'n' choice for the confirmation spawn prompt has been removed,
7823 because its meaning was confusing.
7825 The prompt may now also be redirected to an alternative console by
7826 specifying the console as parameter to systemd.confirm_spawn=.
7828 * Services of Type=notify require a READY=1 notification to be sent
7829 during startup. If no such message is sent, the service now fails,
7830 even if the main process exited with a successful exit code.
7832 * Services that fail to start up correctly now always have their
7833 ExecStopPost= commands executed. Previously, they'd enter "failed"
7834 state directly, without executing these commands.
7836 * The option MulticastDNS= of network configuration files has acquired
7837 an actual implementation. With MulticastDNS=yes a host can resolve
7838 names of remote hosts and reply to mDNS A and AAAA requests.
7840 * When units are about to be started an additional check is now done to
7841 ensure that all dependencies of type BindsTo= (when used in
7842 combination with After=) have been started.
7844 * systemd-analyze gained a new verb "syscall-filter" which shows which
7845 system call groups are defined for the SystemCallFilter= unit file
7846 setting, and which system calls they contain.
7848 * A new system call filter group "@filesystem" has been added,
7849 consisting of various file system related system calls. Group
7850 "@reboot" has been added, covering reboot, kexec and shutdown related
7851 calls. Finally, group "@swap" has been added covering swap
7852 configuration related calls.
7854 * A new unit file option RestrictNamespaces= has been added that may be
7855 used to restrict access to the various process namespace types the
7856 Linux kernel provides. Specifically, it may be used to take away the
7857 right for a service unit to create additional file system, network,
7858 user, and other namespaces. This sandboxing option is particularly
7859 relevant due to the high amount of recently discovered namespacing
7860 related vulnerabilities in the kernel.
7862 * systemd-udev's .link files gained support for a new AutoNegotiation=
7863 setting for configuring Ethernet auto-negotiation.
7865 * systemd-networkd's .network files gained support for a new
7866 ListenPort= setting in the [DHCP] section to explicitly configure the
7867 UDP client port the DHCP client shall listen on.
7869 * .network files gained a new Unmanaged= boolean setting for explicitly
7870 excluding one or more interfaces from management by systemd-networkd.
7872 * The systemd-networkd ProxyARP= option has been renamed to
7873 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
7874 renamed to ReduceARPProxy=. The old names continue to be available
7877 * systemd-networkd gained support for configuring IPv6 Proxy NDP
7878 addresses via the new IPv6ProxyNDPAddress= .network file setting.
7880 * systemd-networkd's bonding device support gained support for two new
7881 configuration options ActiveSlave= and PrimarySlave=.
7883 * The various options in the [Match] section of .network files gained
7884 support for negative matching.
7886 * New systemd-specific mount options are now understood in /etc/fstab:
7888 x-systemd.mount-timeout= may be used to configure the maximum
7889 permitted runtime of the mount command.
7891 x-systemd.device-bound may be set to bind a mount point to its
7892 backing device unit, in order to automatically remove a mount point
7893 if its backing device is unplugged. This option may also be
7894 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
7895 on the block device, which is now automatically set for all CDROM
7896 drives, so that mounted CDs are automatically unmounted when they are
7897 removed from the drive.
7899 x-systemd.after= and x-systemd.before= may be used to explicitly
7900 order a mount after or before another unit or mount point.
7902 * Enqueued start jobs for device units are now automatically garbage
7903 collected if there are no jobs waiting for them anymore.
7905 * systemctl list-jobs gained two new switches: with --after, for every
7906 queued job the jobs it's waiting for are shown; with --before the
7907 jobs which it's blocking are shown.
7909 * systemd-nspawn gained support for ephemeral boots from disk images
7910 (or in other words: --ephemeral and --image= may now be
7911 combined). Moreover, ephemeral boots are now supported for normal
7912 directories, even if the backing file system is not btrfs. Of course,
7913 if the file system does not support file system snapshots or
7914 reflinks, the initial copy operation will be relatively expensive, but
7915 this should still be suitable for many use cases.
7917 * Calendar time specifications in .timer units now support
7918 specifications relative to the end of a month by using "~" instead of
7919 "-" as separator between month and day. For example, "*-02~03" means
7920 "the third last day in February". In addition a new syntax for
7921 repeated events has been added using the "/" character. For example,
7922 "9..17/2:00" means "every two hours from 9am to 5pm".
7924 * systemd-socket-proxyd gained a new parameter --connections-max= for
7925 configuring the maximum number of concurrent connections.
7927 * sd-id128 gained a new API for generating unique IDs for the host in a
7928 way that does not leak the machine ID. Specifically,
7929 sd_id128_get_machine_app_specific() derives an ID based on the
7930 machine ID in a well-defined, non-reversible, stable way. This is
7931 useful whenever an identifier for the host is needed but where the
7932 identifier shall not be useful to identify the system beyond the
7933 scope of the application itself. (Internally this uses HMAC-SHA256 as
7934 keyed hash function using the machine ID as input.)
7936 * NotifyAccess= gained a new supported value "exec". When set
7937 notifications are accepted from all processes systemd itself invoked,
7938 including all control processes.
7940 * .nspawn files gained support for defining overlay mounts using the
7941 Overlay= and OverlayReadOnly= options. Previously this functionality
7942 was only available on the systemd-nspawn command line.
7944 * systemd-nspawn's --bind= and --overlay= options gained support for
7945 bind/overlay mounts whose source lies within the container tree by
7946 prefixing the source path with "+".
7948 * systemd-nspawn's --bind= and --overlay= options gained support for
7949 automatically allocating a temporary source directory in /var/tmp
7950 that is removed when the container dies. Specifically, if the source
7951 directory is specified as empty string this mechanism is selected. An
7952 example usage is --overlay=+/var::/var, which creates an overlay
7953 mount based on the original /var contained in the image, overlaid
7954 with a temporary directory in the host's /var/tmp. This way changes
7955 to /var are automatically flushed when the container shuts down.
7957 * systemd-nspawn --image= option does now permit raw file system block
7958 devices (in addition to images containing partition tables, as
7961 * The disk image dissection logic in systemd-nspawn gained support for
7962 automatically setting up LUKS encrypted as well as Verity protected
7963 partitions. When a container is booted from an encrypted image the
7964 passphrase is queried at start-up time. When a container with Verity
7965 data is started, the root hash is search in a ".roothash" file
7966 accompanying the disk image (alternatively, pass the root hash via
7967 the new --root-hash= command line option).
7969 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
7970 be used to dissect disk images the same way as systemd-nspawn does
7971 it, following the Bootable Partition Specification. It may even be
7972 used to mount disk images with complex partition setups (including
7973 LUKS and Verity partitions) to a local host directory, in order to
7974 inspect them. This tool is not considered public API (yet), and is
7975 thus not installed into /usr/bin. Please do not rely on its
7976 existence, since it might go away or be changed in later systemd
7979 * A new generator "systemd-verity-generator" has been added, similar in
7980 style to "systemd-cryptsetup-generator", permitting automatic setup of
7981 Verity root partitions when systemd boots up. In order to make use of
7982 this your partition setup should follow the Discoverable Partitions
7983 Specification, and the GPT partition ID of the root file system
7984 partition should be identical to the upper 128bit of the Verity root
7985 hash. The GPT partition ID of the Verity partition protecting it
7986 should be the lower 128bit of the Verity root hash. If the partition
7987 image follows this model it is sufficient to specify a single
7988 "roothash=" kernel command line argument to both configure which root
7989 image and verity partition to use as well as the root hash for
7990 it. Note that systemd-nspawn's Verity support follows the same
7991 semantics, meaning that disk images with proper Verity data in place
7992 may be booted in containers with systemd-nspawn as well as on
7993 physical systems via the verity generator. Also note that the "mkosi"
7994 tool available at https://github.com/systemd/mkosi has been updated
7995 to generate Verity protected disk images following this scheme. In
7996 fact, it has been updated to generate disk images that optionally
7997 implement a complete UEFI SecureBoot trust chain, involving a signed
7998 kernel and initrd image that incorporates such a root hash as well as
7999 a Verity-enabled root partition.
8001 * The hardware database (hwdb) udev supports has been updated to carry
8002 accelerometer quirks.
8004 * All system services are now run with a fresh kernel keyring set up
8005 for them. The invocation ID is stored by default in it, thus
8006 providing a safe, non-overridable way to determine the invocation
8009 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
8010 options for bind mounting arbitrary paths in a service-specific
8011 way. When these options are used, arbitrary host or service files and
8012 directories may be mounted to arbitrary locations in the service's
8015 * Documentation has been added that lists all of systemd's low-level
8016 environment variables:
8018 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
8020 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
8021 whether a specific socket file descriptor matches a specified socket
8024 * systemd-firstboot has been updated to check for the
8025 systemd.firstboot= kernel command line option. It accepts a boolean
8026 and when set to false the first boot questions are skipped.
8028 * systemd-fstab-generator has been updated to check for the
8029 systemd.volatile= kernel command line option, which either takes an
8030 optional boolean parameter or the special value "state". If used the
8031 system may be booted in a "volatile" boot mode. Specifically,
8032 "systemd.volatile" is used, the root directory will be mounted as
8033 tmpfs, and only /usr is mounted from the actual root file system. If
8034 "systemd.volatile=state" is used, the root directory will be mounted
8035 as usual, but /var is mounted as tmpfs. This concept provides similar
8036 functionality as systemd-nspawn's --volatile= option, but provides it
8037 on physical boots. Use this option for implementing stateless
8038 systems, or testing systems with all state and/or configuration reset
8039 to the defaults. (Note though that many distributions are not
8040 prepared to boot up without a populated /etc or /var, though.)
8042 * systemd-gpt-auto-generator gained support for LUKS encrypted root
8043 partitions. Previously it only supported LUKS encrypted partitions
8044 for all other uses, except for the root partition itself.
8046 * Socket units gained support for listening on AF_VSOCK sockets for
8047 communication in virtualized QEMU environments.
8049 * The "configure" script gained a new option --with-fallback-hostname=
8050 for specifying the fallback hostname to use if none is configured in
8051 /etc/hostname. For example, by specifying
8052 --with-fallback-hostname=fedora it is possible to default to a
8053 hostname of "fedora" on pristine installations.
8055 * systemd-cgls gained support for a new --unit= switch for listing only
8056 the control groups of a specific unit. Similar --user-unit= has been
8057 added for listing only the control groups of a specific user unit.
8059 * systemd-mount gained a new --umount switch for unmounting a mount or
8060 automount point (and all mount/automount points below it).
8062 * systemd will now refuse full configuration reloads (via systemctl
8063 daemon-reload and related calls) unless at least 16MiB of free space
8064 are available in /run. This is a safety precaution in order to ensure
8065 that generators can safely operate after the reload completed.
8067 * A new unit file option RootImage= has been added, which has a similar
8068 effect as RootDirectory= but mounts the service's root directory from
8069 a disk image instead of plain directory. This logic reuses the same
8070 image dissection and mount logic that systemd-nspawn already uses,
8071 and hence supports any disk images systemd-nspawn supports, including
8072 those following the Discoverable Partition Specification, as well as
8073 Verity enabled images. This option enables systemd to run system
8074 services directly off disk images acting as resource bundles,
8075 possibly even including full integrity data.
8077 * A new MountAPIVFS= unit file option has been added, taking a boolean
8078 argument. If enabled /proc, /sys and /dev (collectively called the
8079 "API VFS") will be mounted for the service. This is only relevant if
8080 RootDirectory= or RootImage= is used for the service, as these mounts
8081 are of course in place in the host mount namespace anyway.
8083 * systemd-nspawn gained support for a new --pivot-root= switch. If
8084 specified the root directory within the container image is pivoted to
8085 the specified mount point, while the original root disk is moved to a
8086 different place. This option enables booting of ostree images
8087 directly with systemd-nspawn.
8089 * The systemd build scripts will no longer complain if the NTP server
8090 addresses are not changed from the defaults. Google now supports
8091 these NTP servers officially. We still recommend downstreams to
8092 properly register an NTP pool with the NTP pool project though.
8094 * coredumpctl gained a new "--reverse" option for printing the list
8095 of coredumps in reverse order.
8097 * coredumpctl will now show additional information about truncated and
8098 inaccessible coredumps, as well as coredumps that are still being
8099 processed. It also gained a new --quiet switch for suppressing
8100 additional informational message in its output.
8102 * coredumpctl gained support for only showing coredumps newer and/or
8103 older than specific timestamps, using the new --since= and --until=
8104 options, reminiscent of journalctl's options by the same name.
8106 * The systemd-coredump logic has been improved so that it may be reused
8107 to collect backtraces in non-compiled languages, for example in
8108 scripting languages such as Python.
8110 * machinectl will now show the UID shift of local containers, if user
8111 namespacing is enabled for them.
8113 * systemd will now optionally run "environment generator" binaries at
8114 configuration load time. They may be used to add environment
8115 variables to the environment block passed to services invoked. One
8116 user environment generator is shipped by default that sets up
8117 environment variables based on files dropped into /etc/environment.d
8118 and ~/.config/environment.d/.
8120 * systemd-resolved now includes the new, recently published 2017 DNSSEC
8123 * hostnamed has been updated to report a new chassis type of
8124 "convertible" to cover "foldable" laptops that can both act as a
8125 tablet and as a laptop, such as various Lenovo Yoga devices.
8127 Contributions from: Adrián López, Alexander Galanin, Alexander
8128 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
8129 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
8130 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
8131 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
8132 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
8133 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
8134 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
8135 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
8136 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
8137 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
8138 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
8139 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
8140 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
8141 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
8142 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
8143 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
8144 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
8145 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
8146 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
8147 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
8148 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
8149 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
8150 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
8151 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
8152 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
8153 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
8154 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
8157 — Berlin, 2017-03-01
8161 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
8162 RestrictAddressFamilies= enabled. These sandboxing options should
8163 generally be compatible with the various external udev call-out
8164 binaries we are aware of, however there may be exceptions, in
8165 particular when exotic languages for these call-outs are used. In
8166 this case, consider turning off these settings locally.
8168 * The new RemoveIPC= option can be used to remove IPC objects owned by
8169 the user or group of a service when that service exits.
8171 * The new ProtectKernelModules= option can be used to disable explicit
8172 load and unload operations of kernel modules by a service. In
8173 addition access to /usr/lib/modules is removed if this option is set.
8175 * ProtectSystem= option gained a new value "strict", which causes the
8176 whole file system tree with the exception of /dev, /proc, and /sys,
8177 to be remounted read-only for a service.
8179 * The new ProtectKernelTunables= option can be used to disable
8180 modification of configuration files in /sys and /proc by a service.
8181 Various directories and files are remounted read-only, so access is
8182 restricted even if the file permissions would allow it.
8184 * The new ProtectControlGroups= option can be used to disable write
8185 access by a service to /sys/fs/cgroup.
8187 * Various systemd services have been hardened with
8188 ProtectKernelTunables=yes, ProtectControlGroups=yes,
8189 RestrictAddressFamilies=.
8191 * Support for dynamically creating users for the lifetime of a service
8192 has been added. If DynamicUser=yes is specified, user and group IDs
8193 will be allocated from the range 61184…65519 for the lifetime of the
8194 service. They can be resolved using the new nss-systemd.so NSS
8195 module. The module must be enabled in /etc/nsswitch.conf. Services
8196 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
8197 any resources allocated by the service will be cleaned up when the
8198 service exits. They also have ProtectHome=read-only and
8199 ProtectSystem=strict enabled, so they are not able to make any
8200 permanent modifications to the system.
8202 * The nss-systemd module also always resolves root and nobody, making
8203 it possible to have no /etc/passwd or /etc/group files in minimal
8204 container or chroot environments.
8206 * Services may be started with their own user namespace using the new
8207 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
8208 under which the service is running are mapped. All other users are
8211 * Support for the cgroup namespace has been added to systemd-nspawn. If
8212 supported by kernel, the container system started by systemd-nspawn
8213 will have its own view of the cgroup hierarchy. This new behaviour
8214 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
8216 * The new MemorySwapMax= option can be used to limit the maximum swap
8217 usage under the unified cgroup hierarchy.
8219 * Support for the CPU controller in the unified cgroup hierarchy has
8220 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
8221 options. This controller requires out-of-tree patches for the kernel
8222 and the support is provisional.
8224 * Mount and automount units may now be created transiently
8225 (i.e. dynamically at runtime via the bus API, instead of requiring
8226 unit files in the file system).
8228 * systemd-mount is a new tool which may mount file systems – much like
8229 mount(8), optionally pulling in additional dependencies through
8230 transient .mount and .automount units. For example, this tool
8231 automatically runs fsck on a backing block device before mounting,
8232 and allows the automount logic to be used dynamically from the
8233 command line for establishing mount points. This tool is particularly
8234 useful when dealing with removable media, as it will ensure fsck is
8235 run – if necessary – before the first access and that the file system
8236 is quickly unmounted after each access by utilizing the automount
8237 logic. This maximizes the chance that the file system on the
8238 removable media stays in a clean state, and if it isn't in a clean
8239 state is fixed automatically.
8241 * LazyUnmount=yes option for mount units has been added to expose the
8242 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
8245 * /efi will be used as the mount point of the EFI boot partition, if
8246 the directory is present, and the mount point was not configured
8247 through other means (e.g. fstab). If /efi directory does not exist,
8248 /boot will be used as before. This makes it easier to automatically
8249 mount the EFI partition on systems where /boot is used for something
8252 * When operating on GPT disk images for containers, systemd-nspawn will
8253 now mount the ESP to /boot or /efi according to the same rules as PID
8254 1 running on a host. This allows tools like "bootctl" to operate
8255 correctly within such containers, in order to make container images
8256 bootable on physical systems.
8258 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
8260 * Two new user session targets have been added to support running
8261 graphical sessions under the systemd --user instance:
8262 graphical-session.target and graphical-session-pre.target. See
8263 systemd.special(7) for a description of how those targets should be
8266 * The vconsole initialization code has been significantly reworked to
8267 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
8268 support unicode keymaps. Font and keymap configuration will now be
8269 copied to all allocated virtual consoles.
8271 * FreeBSD's bhyve virtualization is now detected.
8273 * Information recorded in the journal for core dumps now includes the
8274 contents of /proc/mountinfo and the command line of the process at
8275 the top of the process hierarchy (which is usually the init process
8278 * systemd-journal-gatewayd learned the --directory= option to serve
8279 files from the specified location.
8281 * journalctl --root=… can be used to peruse the journal in the
8282 /var/log/ directories inside of a container tree. This is similar to
8283 the existing --machine= option, but does not require the container to
8286 * The hardware database has been extended to support
8287 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
8290 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
8291 specify the click rate for mice which include a horizontal wheel with
8292 a click rate that is different than the one for the vertical wheel.
8294 * systemd-run gained a new --wait option that makes service execution
8295 synchronous. (Specifically, the command will not return until the
8296 specified service binary exited.)
8298 * systemctl gained a new --wait option that causes the start command to
8299 wait until the units being started have terminated again.
8301 * A new journal output mode "short-full" has been added which displays
8302 timestamps with abbreviated English day names and adds a timezone
8303 suffix. Those timestamps include more information than the default
8304 "short" output mode, and can be passed directly to journalctl's
8305 --since= and --until= options.
8307 * /etc/resolv.conf will be bind-mounted into containers started by
8308 systemd-nspawn, if possible, so any changes to resolv.conf contents
8309 are automatically propagated to the container.
8311 * The number of instances for socket-activated services originating
8312 from a single IP address can be limited with
8313 MaxConnectionsPerSource=, extending the existing setting of
8316 * systemd-networkd gained support for vcan ("Virtual CAN") interface
8319 * .netdev and .network configuration can now be extended through
8322 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
8323 Segmentation Offload, Generic Receive Offload, Large Receive Offload
8324 can be enabled and disabled using the new UDPSegmentationOffload=,
8325 TCPSegmentationOffload=, GenericSegmentationOffload=,
8326 GenericReceiveOffload=, LargeReceiveOffload= options in the
8327 [Link] section of .link files.
8329 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
8330 Port VLAN ID can be configured for bridge devices using the new STP=,
8331 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
8332 section of .netdev files.
8334 * The route table to which routes received over DHCP or RA should be
8335 added can be configured with the new RouteTable= option in the [DHCP]
8336 and [IPv6AcceptRA] sections of .network files.
8338 * The Address Resolution Protocol can be disabled on links managed by
8339 systemd-networkd using the ARP=no setting in the [Link] section of
8342 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
8343 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
8344 encode information about the result and exit codes of the current
8345 service runtime cycle.
8347 * systemd-sysctl will now configure kernel parameters in the order
8348 they occur in the configuration files. This matches what sysctl
8349 has been traditionally doing.
8351 * kernel-install "plugins" that are executed to perform various
8352 tasks after a new kernel is added and before an old one is removed
8353 can now return a special value to terminate the procedure and
8354 prevent any later plugins from running.
8356 * Journald's SplitMode=login setting has been deprecated. It has been
8357 removed from documentation, and its use is discouraged. In a future
8358 release it will be completely removed, and made equivalent to current
8359 default of SplitMode=uid.
8361 * Storage=both option setting in /etc/systemd/coredump.conf has been
8362 removed. With fast LZ4 compression storing the core dump twice is not
8365 * The --share-system systemd-nspawn option has been replaced with an
8366 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
8367 this functionality is discouraged. In addition the variables
8368 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
8369 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
8370 individual namespaces.
8372 * "machinectl list" now shows the IP address of running containers in
8373 the output, as well as OS release information.
8375 * "loginctl list" now shows the TTY of each session in the output.
8377 * sd-bus gained new API calls sd_bus_track_set_recursive(),
8378 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
8379 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
8380 tracking objects in a "recursive" mode, where a single client can be
8381 counted multiple times, if it takes multiple references.
8383 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
8384 sd_bus_get_exit_on_disconnect(). They may be used to make a
8385 process using sd-bus automatically exit if the bus connection is
8388 * Bus clients of the service manager may now "pin" loaded units into
8389 memory, by taking an explicit reference on them. This is useful to
8390 ensure the client can retrieve runtime data about the service even
8391 after the service completed execution. Taking such a reference is
8392 available only for privileged clients and should be helpful to watch
8393 running services in a race-free manner, and in particular collect
8394 information about exit statuses and results.
8396 * The nss-resolve module has been changed to strictly return UNAVAIL
8397 when communication via D-Bus with resolved failed, and NOTFOUND when
8398 a lookup completed but was negative. This means it is now possible to
8399 neatly configure fallbacks using nsswitch.conf result checking
8400 expressions. Taking benefit of this, the new recommended
8401 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
8403 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
8405 * A new setting CtrlAltDelBurstAction= has been added to
8406 /etc/systemd/system.conf which may be used to configure the precise
8407 behaviour if the user on the console presses Ctrl-Alt-Del more often
8408 than 7 times in 2s. Previously this would unconditionally result in
8409 an expedited, immediate reboot. With this new setting the precise
8410 operation may be configured in more detail, and also turned off
8413 * In .netdev files two new settings RemoteChecksumTx= and
8414 RemoteChecksumRx= are now understood that permit configuring the
8415 remote checksumming logic for VXLAN networks.
8417 * The service manager learnt a new "invocation ID" concept for invoked
8418 services. Each runtime cycle of a service will get a new invocation
8419 ID (a 128bit random UUID) assigned that identifies the current
8420 run of the service uniquely and globally. A new invocation ID
8421 is generated each time a service starts up. The journal will store
8422 the invocation ID of a service along with any logged messages, thus
8423 making the invocation ID useful for matching the online runtime of a
8424 service with the offline log data it generated in a safe way without
8425 relying on synchronized timestamps. In many ways this new service
8426 invocation ID concept is similar to the kernel's boot ID concept that
8427 uniquely and globally identifies the runtime of each boot. The
8428 invocation ID of a service is passed to the service itself via an
8429 environment variable ($INVOCATION_ID). A new bus call
8430 GetUnitByInvocationID() has been added that is similar to GetUnit()
8431 but instead of retrieving the bus path for a unit by its name
8432 retrieves it by its invocation ID. The returned path is valid only as
8433 long as the passed invocation ID is current.
8435 * systemd-resolved gained a new "DNSStubListener" setting in
8436 resolved.conf. It either takes a boolean value or the special values
8437 "udp" and "tcp", and configures whether to enable the stub DNS
8438 listener on 127.0.0.53:53.
8440 * IP addresses configured via networkd may now carry additional
8441 configuration settings supported by the kernel. New options include:
8442 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
8443 PrefixRoute=, AutoJoin=.
8445 * The PAM configuration fragment file for "user@.service" shipped with
8446 systemd (i.e. the --user instance of systemd) has been stripped to
8447 the minimum necessary to make the system boot. Previously, it
8448 contained Fedora-specific stanzas that did not apply to other
8449 distributions. It is expected that downstream distributions add
8450 additional configuration lines, matching their needs to this file,
8451 using it only as rough template of what systemd itself needs. Note
8452 that this reduced fragment does not even include an invocation of
8453 pam_limits which most distributions probably want to add, even though
8454 systemd itself does not need it. (There's also the new build time
8455 option --with-pamconfdir=no to disable installation of the PAM
8458 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
8459 capability is now also dropped from its set (in addition to
8460 CAP_SYS_MKNOD as before).
8462 * In service unit files it is now possible to connect a specific named
8463 file descriptor with stdin/stdout/stdout of an executed service. The
8464 name may be specified in matching .socket units using the
8465 FileDescriptorName= setting.
8467 * A number of journal settings may now be configured on the kernel
8468 command line. Specifically, the following options are now understood:
8469 systemd.journald.max_level_console=,
8470 systemd.journald.max_level_store=,
8471 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
8472 systemd.journald.max_level_wall=.
8474 * "systemctl is-enabled --full" will now show by which symlinks a unit
8475 file is enabled in the unit dependency tree.
8477 * Support for VeraCrypt encrypted partitions has been added to the
8478 "cryptsetup" logic and /etc/crypttab.
8480 * systemd-detect-virt gained support for a new --private-users switch
8481 that checks whether the invoking processes are running inside a user
8482 namespace. Similar, a new special value "private-users" for the
8483 existing ConditionVirtualization= setting has been added, permitting
8484 skipping of specific units in user namespace environments.
8486 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
8487 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
8488 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
8489 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
8490 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
8491 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
8492 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
8493 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
8494 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
8495 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
8496 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
8497 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
8498 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
8499 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
8500 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
8501 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
8502 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
8503 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
8504 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
8505 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
8506 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
8507 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
8508 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
8509 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
8510 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
8511 Jędrzejewski-Szmek, Zeal Jagannatha
8513 — Santa Fe, 2016-11-03
8517 * In service units the various ExecXYZ= settings have been extended
8518 with an additional special character as first argument of the
8519 assigned value: if the character '+' is used the specified command
8520 line it will be run with full privileges, regardless of User=,
8521 Group=, CapabilityBoundingSet= and similar options. The effect is
8522 similar to the existing PermissionsStartOnly= option, but allows
8523 configuration of this concept for each executed command line
8526 * Services may now alter the service watchdog timeout at runtime by
8527 sending a WATCHDOG_USEC= message via sd_notify().
8529 * MemoryLimit= and related unit settings now optionally take percentage
8530 specifications. The percentage is taken relative to the amount of
8531 physical memory in the system (or in case of containers, the assigned
8532 amount of memory). This allows scaling service resources neatly with
8533 the amount of RAM available on the system. Similarly, systemd-logind's
8534 RuntimeDirectorySize= option now also optionally takes percentage
8537 * In similar fashion TasksMax= takes percentage values now, too. The
8538 value is taken relative to the configured maximum number of processes
8539 on the system. The per-service task maximum has been changed to 15%
8540 using this functionality. (Effectively this is an increase of 512 →
8541 4915 for service units, given the kernel's default pid_max setting.)
8543 * Calendar time specifications in .timer units now understand a ".."
8544 syntax for time ranges. Example: "4..7:10" may now be used for
8545 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
8548 * The InaccessableDirectories=, ReadOnlyDirectories= and
8549 ReadWriteDirectories= unit file settings have been renamed to
8550 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
8551 applied to all kinds of file nodes, and not just directories, with
8552 the exception of symlinks. Specifically these settings may now be
8553 used on block and character device nodes, UNIX sockets and FIFOS as
8554 well as regular files. The old names of these settings remain
8555 available for compatibility.
8557 * systemd will now log about all service processes it kills forcibly
8558 (using SIGKILL) because they remained after the clean shutdown phase
8559 of the service completed. This should help identifying services that
8560 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
8561 systemd-logind's configuration a similar log message is generated for
8562 processes killed at the end of each session due to this setting.
8564 * systemd will now set the $JOURNAL_STREAM environment variable for all
8565 services whose stdout/stderr are connected to the Journal (which
8566 effectively means by default: all services). The variable contains
8567 the device and inode number of the file descriptor used for
8568 stdout/stderr. This may be used by invoked programs to detect whether
8569 their stdout/stderr is connected to the Journal, in which case they
8570 can switch over to direct Journal communication, thus being able to
8571 pass extended, structured metadata along with their log messages. As
8572 one example, this is now used by glib's logging primitives.
8574 * When using systemd's default tmp.mount unit for /tmp, the mount point
8575 will now be established with the "nosuid" and "nodev" options. This
8576 avoids privilege escalation attacks that put traps and exploits into
8577 /tmp. However, this might cause problems if you e. g. put container
8578 images or overlays into /tmp; if you need this, override tmp.mount's
8579 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
8582 * systemd now supports the "memory" cgroup controller also on
8585 * The systemd-cgtop tool now optionally takes a control group path as
8586 command line argument. If specified, the control group list shown is
8587 limited to subgroups of that group.
8589 * The SystemCallFilter= unit file setting gained support for
8590 pre-defined, named system call filter sets. For example
8591 SystemCallFilter=@clock is now an effective way to make all clock
8592 changing-related system calls unavailable to a service. A number of
8593 similar pre-defined groups are defined. Writing system call filters
8594 for system services is simplified substantially with this new
8595 concept. Accordingly, all of systemd's own, long-running services now
8596 enable system call filtering based on this, by default.
8598 * A new service setting MemoryDenyWriteExecute= has been added, taking
8599 a boolean value. If turned on, a service may no longer create memory
8600 mappings that are writable and executable at the same time. This
8601 enhances security for services where this is enabled as it becomes
8602 harder to dynamically write and then execute memory in exploited
8603 service processes. This option has been enabled for all of systemd's
8604 own long-running services.
8606 * A new RestrictRealtime= service setting has been added, taking a
8607 boolean argument. If set the service's processes may no longer
8608 acquire realtime scheduling. This improves security as realtime
8609 scheduling may otherwise be used to easily freeze the system.
8611 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
8612 value. This may be used for requesting that the system manager inside
8613 of the container reports start-up completion to nspawn which then
8614 propagates this notification further to the service manager
8615 supervising nspawn itself. A related option NotifyReady= in .nspawn
8616 files has been added too. This functionality allows ordering of the
8617 start-up of multiple containers using the usual systemd ordering
8620 * machinectl gained a new command "stop" that is an alias for
8623 * systemd-resolved gained support for contacting DNS servers on
8624 link-local IPv6 addresses.
8626 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
8627 its caches. A method call for requesting the same operation has been
8628 added to the bus API too, and is made available via "systemd-resolve
8631 * systemd-resolve gained a new --status switch. If passed a brief
8632 summary of the used DNS configuration with per-interface information
8635 * resolved.conf gained a new Cache= boolean option, defaulting to
8636 on. If turned off local DNS caching is disabled. This comes with a
8637 performance penalty in particular when DNSSEC is enabled. Note that
8638 resolved disables its internal caching implicitly anyway, when the
8639 configured DNS server is on a host-local IP address such as ::1 or
8640 127.0.0.1, thus automatically avoiding double local caching.
8642 * systemd-resolved now listens on the local IP address 127.0.0.53:53
8643 for DNS requests. This improves compatibility with local programs
8644 that do not use the libc NSS or systemd-resolved's bus APIs for name
8645 resolution. This minimal DNS service is only available to local
8646 programs and does not implement the full DNS protocol, but enough to
8647 cover local DNS clients. A new, static resolv.conf file, listing just
8648 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
8649 now recommended to make /etc/resolv.conf a symlink to this file in
8650 order to route all DNS lookups to systemd-resolved, regardless if
8651 done via NSS, the bus API or raw DNS packets. Note that this local
8652 DNS service is not as fully featured as the libc NSS or
8653 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
8654 used to deliver link-local address information (as this implies
8655 sending a local interface index along), LLMNR/mDNS support via this
8656 interface is severely restricted. It is thus strongly recommended for
8657 all applications to use the libc NSS API or native systemd-resolved
8660 * systemd-networkd's bridge support learned a new setting
8661 VLANFiltering= for controlling VLAN filtering. Moreover a new section
8662 in .network files has been added for configuring VLAN bridging in
8663 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
8665 * systemd-networkd's IPv6 Router Advertisement code now makes use of
8666 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
8667 now be acquired without relying on DHCPv6. Two new options
8668 UseDomains= and UseDNS= have been added to configure this behaviour.
8670 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
8671 renamed IPv6AcceptRA=, without altering its behaviour. The old
8672 setting name remains available for compatibility reasons.
8674 * The systemd-networkd VTI/VTI6 tunneling support gained new options
8675 Key=, InputKey= and OutputKey=.
8677 * systemd-networkd gained support for VRF ("Virtual Routing Function")
8678 interface configuration.
8680 * "systemctl edit" may now be used to create new unit files by
8681 specifying the --force switch.
8683 * sd-event gained a new function sd_event_get_iteration() for
8684 requesting the current iteration counter of the event loop. It starts
8685 at zero and is increased by one with each event loop iteration.
8687 * A new rpm macro %systemd_ordering is provided by the macros.systemd
8688 file. It can be used in lieu of %systemd_requires in packages which
8689 don't use any systemd functionality and are intended to be installed
8690 in minimal containers without systemd present. This macro provides
8691 ordering dependencies to ensure that if the package is installed in
8692 the same rpm transaction as systemd, systemd will be installed before
8693 the scriptlets for the package are executed, allowing unit presets
8696 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
8697 been added to simplify packaging of generators.
8699 * The os-release file gained VERSION_CODENAME field for the
8700 distribution nickname (e.g. VERSION_CODENAME=woody).
8702 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
8703 can be set to disable parsing of metadata and the creation
8704 of persistent symlinks for that device.
8706 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
8707 to make them available to logged-in users has been reverted.
8709 * Much of the common code of the various systemd components is now
8710 built into an internal shared library libsystemd-shared-231.so
8711 (incorporating the systemd version number in the name, to be updated
8712 with future releases) that the components link to. This should
8713 decrease systemd footprint both in memory during runtime and on
8714 disk. Note that the shared library is not for public use, and is
8715 neither API nor ABI stable, but is likely to change with every new
8716 released update. Packagers need to make sure that binaries
8717 linking to libsystemd-shared.so are updated in step with the
8720 * Configuration for "mkosi" is now part of the systemd
8721 repository. mkosi is a tool to easily build legacy-free OS images,
8722 and is available on github: https://github.com/systemd/mkosi. If
8723 "mkosi" is invoked in the build tree a new raw OS image is generated
8724 incorporating the systemd sources currently being worked on and a
8725 clean, fresh distribution installation. The generated OS image may be
8726 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
8727 UEFI PC. This functionality is particularly useful to easily test
8728 local changes made to systemd in a pristine, defined environment. See
8729 doc/HACKING for details.
8731 * configure learned the --with-support-url= option to specify the
8732 distribution's bugtracker.
8734 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
8735 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
8736 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
8737 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
8738 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
8739 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
8740 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
8741 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
8742 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
8743 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
8744 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
8745 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
8746 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
8747 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
8748 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
8749 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
8750 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
8751 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
8752 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
8754 — Berlin, 2016-07-25
8758 * DNSSEC is now turned on by default in systemd-resolved (in
8759 "allow-downgrade" mode), but may be turned off during compile time by
8760 passing "--with-default-dnssec=no" to "configure" (and of course,
8761 during runtime with DNSSEC= in resolved.conf). We recommend
8762 downstreams to leave this on at least during development cycles and
8763 report any issues with the DNSSEC logic upstream. We are very
8764 interested in collecting feedback about the DNSSEC validator and its
8765 limitations in the wild. Note however, that DNSSEC support is
8766 probably nothing downstreams should turn on in stable distros just
8767 yet, as it might create incompatibilities with a few DNS servers and
8768 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
8769 automatically whenever we detect such incompatible setups, but there
8770 might be systems we do not cover yet. Hence: please help us testing
8771 the DNSSEC code, leave this on where you can, report back, but then
8772 again don't consider turning this on in your stable, LTS or
8773 production release just yet. (Note that you have to enable
8774 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
8775 and its DNSSEC mode for hostname resolution from local
8778 * systemd-resolve conveniently resolves DANE records with the --tlsa
8779 option and OPENPGPKEY records with the --openpgp option. It also
8780 supports dumping raw DNS record data via the new --raw= switch.
8782 * systemd-logind will now by default terminate user processes that are
8783 part of the user session scope unit (session-XX.scope) when the user
8784 logs out. This behavior is controlled by the KillUserProcesses=
8785 setting in logind.conf, and the previous default of "no" is now
8786 changed to "yes". This means that user sessions will be properly
8787 cleaned up after, but additional steps are necessary to allow
8788 intentionally long-running processes to survive logout.
8790 While the user is logged in at least once, user@.service is running,
8791 and any service that should survive the end of any individual login
8792 session can be started at a user service or scope using systemd-run.
8793 systemd-run(1) man page has been extended with an example which shows
8794 how to run screen in a scope unit underneath user@.service. The same
8795 command works for tmux.
8797 After the user logs out of all sessions, user@.service will be
8798 terminated too, by default, unless the user has "lingering" enabled.
8799 To effectively allow users to run long-term tasks even if they are
8800 logged out, lingering must be enabled for them. See loginctl(1) for
8801 details. The default polkit policy was modified to allow users to
8802 set lingering for themselves without authentication.
8804 Previous defaults can be restored at compile time by the
8805 --without-kill-user-processes option to "configure".
8807 * systemd-logind gained new configuration settings SessionsMax= and
8808 InhibitorsMax=, both with a default of 8192. It will not register new
8809 user sessions or inhibitors above this limit.
8811 * systemd-logind will now reload configuration on SIGHUP.
8813 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
8814 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
8815 enable. Also, support for the "io" cgroup controller in the unified
8816 hierarchy has been added, so that the "memory", "pids" and "io" are
8817 now the controllers that are supported on the unified hierarchy.
8819 WARNING: it is not possible to use previous systemd versions with
8820 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
8821 is necessary to also update systemd in the initramfs if using the
8822 unified hierarchy. An updated SELinux policy is also required.
8824 * LLDP support has been extended, and both passive (receive-only) and
8825 active (sender) modes are supported. Passive mode ("routers-only") is
8826 enabled by default in systemd-networkd. Active LLDP mode is enabled
8827 by default for containers on the internal network. The "networkctl
8828 lldp" command may be used to list information gathered. "networkctl
8829 status" will also show basic LLDP information on connected peers now.
8831 * The IAID and DUID unique identifier sent in DHCP requests may now be
8832 configured for the system and each .network file managed by
8833 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
8835 * systemd-networkd gained support for configuring proxy ARP support for
8836 each interface, via the ProxyArp= setting in .network files. It also
8837 gained support for configuring the multicast querier feature of
8838 bridge devices, via the new MulticastQuerier= setting in .netdev
8839 files. Similarly, snooping on the IGMP traffic can be controlled
8840 via the new setting MulticastSnooping=.
8842 A new setting PreferredLifetime= has been added for addresses
8843 configured in .network file to configure the lifetime intended for an
8846 The systemd-networkd DHCP server gained the option EmitRouter=, which
8847 defaults to yes, to configure whether the DHCP Option 3 (Router)
8850 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
8851 systemd-socket-activate and installed into /usr/bin. It is now fully
8854 * systemd-journald now uses separate threads to flush changes to disk
8855 when closing journal files, thus reducing impact of slow disk I/O on
8856 logging performance.
8858 * The sd-journal API gained two new calls
8859 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
8860 can be used to open journal files using file descriptors instead of
8861 file or directory paths. sd_journal_open_container() has been
8862 deprecated, sd_journal_open_directory_fd() should be used instead
8863 with the flag SD_JOURNAL_OS_ROOT.
8865 * journalctl learned a new output mode "-o short-unix" that outputs log
8866 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
8867 UTC). It also gained support for a new --no-hostname setting to
8868 suppress the hostname column in the family of "short" output modes.
8870 * systemd-ask-password now optionally skips printing of the password to
8871 stdout with --no-output which can be useful in scripts.
8873 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
8874 (devices tagged with ID_MAKER_TOOL) are now tagged with
8875 "uaccess" and are available to logged in users.
8877 * The DeviceAllow= unit setting now supports specifiers (with "%").
8879 * "systemctl show" gained a new --value switch, which allows print a
8880 only the contents of a specific unit property, without also printing
8881 the property's name. Similar support was added to "show*" verbs
8882 of loginctl and machinectl that output "key=value" lists.
8884 * A new unit type "generated" was added for files dynamically generated
8885 by generator tools. Similarly, a new unit type "transient" is used
8886 for unit files created using the runtime API. "systemctl enable" will
8887 refuse to operate on such files.
8889 * A new command "systemctl revert" has been added that may be used to
8890 revert to the vendor version of a unit file, in case local changes
8891 have been made by adding drop-ins or overriding the unit file.
8893 * "machinectl clean" gained a new verb to automatically remove all or
8894 just hidden container images.
8896 * systemd-tmpfiles gained support for a new line type "e" for emptying
8897 directories, if they exist, without creating them if they don't.
8899 * systemd-nspawn gained support for automatically patching the UID/GIDs
8900 of the owners and the ACLs of all files and directories in a
8901 container tree to match the UID/GID user namespacing range selected
8902 for the container invocation. This mode is enabled via the new
8903 --private-users-chown switch. It also gained support for
8904 automatically choosing a free, previously unused UID/GID range when
8905 starting a container, via the new --private-users=pick setting (which
8906 implies --private-users-chown). Together, these options for the first
8907 time make user namespacing for nspawn containers fully automatic and
8908 thus deployable. The systemd-nspawn@.service template unit file has
8909 been changed to use this functionality by default.
8911 * systemd-nspawn gained a new --network-zone= switch, that allows
8912 creating ad-hoc virtual Ethernet links between multiple containers,
8913 that only exist as long as at least one container referencing them is
8914 running. This allows easy connecting of multiple containers with a
8915 common link that implements an Ethernet broadcast domain. Each of
8916 these network "zones" may be named relatively freely by the user, and
8917 may be referenced by any number of containers, but each container may
8918 only reference one of these "zones". On the lower level, this is
8919 implemented by an automatically managed bridge network interface for
8920 each zone, that is created when the first container referencing its
8921 zone is created and removed when the last one referencing its zone
8924 * The default start timeout may now be configured on the kernel command
8925 line via systemd.default_timeout_start_sec=. It was already
8926 configurable via the DefaultTimeoutStartSec= option in
8927 /etc/systemd/system.conf.
8929 * Socket units gained a new TriggerLimitIntervalSec= and
8930 TriggerLimitBurst= setting to configure a limit on the activation
8931 rate of the socket unit.
8933 * The LimitNICE= setting now optionally takes normal UNIX nice values
8934 in addition to the raw integer limit value. If the specified
8935 parameter is prefixed with "+" or "-" and is in the range -20…19 the
8936 value is understood as UNIX nice value. If not prefixed like this it
8937 is understood as raw RLIMIT_NICE limit.
8939 * Note that the effect of the PrivateDevices= unit file setting changed
8940 slightly with this release: the per-device /dev file system will be
8941 mounted read-only from this version on, and will have "noexec"
8942 set. This (minor) change of behavior might cause some (exceptional)
8943 legacy software to break, when PrivateDevices=yes is set for its
8944 service. Please leave PrivateDevices= off if you run into problems
8947 * systemd-bootchart has been split out to a separate repository:
8948 https://github.com/systemd/systemd-bootchart
8950 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
8951 merged into the kernel in its current form.
8953 * The compatibility libraries libsystemd-daemon.so,
8954 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
8955 which have been deprecated since systemd-209 have been removed along
8956 with the corresponding pkg-config files. All symbols provided by
8957 those libraries are provided by libsystemd.so.
8959 * The Capabilities= unit file setting has been removed (it is ignored
8960 for backwards compatibility). AmbientCapabilities= and
8961 CapabilityBoundingSet= should be used instead.
8963 * A new special target has been added, initrd-root-device.target,
8964 which creates a synchronization point for dependencies of the root
8965 device in early userspace. Initramfs builders must ensure that this
8966 target is now included in early userspace.
8968 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
8969 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
8970 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
8971 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
8972 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
8973 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
8974 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
8975 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
8976 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
8977 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
8978 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
8979 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
8980 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
8981 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
8982 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
8983 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
8984 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
8985 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
8986 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
8987 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
8988 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
8989 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
8990 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
8991 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
8992 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
8995 — Fairfax, 2016-05-21
8999 * The systemd-resolved DNS resolver service has gained a substantial
9000 set of new features, most prominently it may now act as a DNSSEC
9001 validating stub resolver. DNSSEC mode is currently turned off by
9002 default, but is expected to be turned on by default in one of the
9003 next releases. For now, we invite everybody to test the DNSSEC logic
9004 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
9005 service also gained a full set of D-Bus interfaces, including calls
9006 to configure DNS and DNSSEC settings per link (for use by external
9007 network management software). systemd-resolved and systemd-networkd
9008 now distinguish between "search" and "routing" domains. The former
9009 are used to qualify single-label names, the latter are used purely
9010 for routing lookups within certain domains to specific links.
9011 resolved now also synthesizes RRs for all entries from /etc/hosts.
9013 * The systemd-resolve tool (which is a client utility for
9014 systemd-resolved) has been improved considerably and is now fully
9015 supported and documented. Hence it has moved from /usr/lib/systemd to
9018 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
9021 * The coredump collection logic has been reworked: when a coredump is
9022 collected it is now written to disk, compressed and processed
9023 (including stacktrace extraction) from a new instantiated service
9024 systemd-coredump@.service, instead of directly from the
9025 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
9026 processing large coredumps can take up a substantial amount of
9027 resources and time, and this previously happened entirely outside of
9028 systemd's service supervision. With the new logic the core_pattern
9029 hook only does minimal metadata collection before passing off control
9030 to the new instantiated service, which is configured with a time
9031 limit, a nice level and other settings to minimize negative impact on
9032 the rest of the system. Also note that the new logic will honour the
9033 RLIMIT_CORE setting of the crashed process, which now allows users
9034 and processes to turn off coredumping for their processes by setting
9037 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
9038 and all forked processes by default. Previously, PID 1 would leave
9039 the setting at "0" for all processes, as set by the kernel. Note that
9040 the resource limit traditionally has no effect on the generated
9041 coredumps on the system if the /proc/sys/kernel/core_pattern hook
9042 logic is used. Since the limit is now honoured (see above) its
9043 default has been changed so that the coredumping logic is enabled by
9044 default for all processes, while allowing specific opt-out.
9046 * When the stacktrace is extracted from processes of system users, this
9047 is now done as "systemd-coredump" user, in order to sandbox this
9048 potentially security sensitive parsing operation. (Note that when
9049 processing coredumps of normal users this is done under the user ID
9050 of process that crashed, as before.) Packagers should take notice
9051 that it is now necessary to create the "systemd-coredump" system user
9052 and group at package installation time.
9054 * The systemd-activate socket activation testing tool gained support
9055 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
9056 and --seqpacket switches. It also has been extended to support both
9057 new-style and inetd-style file descriptor passing. Use the new
9058 --inetd switch to request inetd-style file descriptor passing.
9060 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
9061 variable, which takes a boolean value. If set to false, ANSI color
9062 output is disabled in the tools even when run on a terminal that
9065 * The VXLAN support in networkd now supports two new settings
9066 DestinationPort= and PortRange=.
9068 * A new systemd.machine_id= kernel command line switch has been added,
9069 that may be used to set the machine ID in /etc/machine-id if it is
9070 not initialized yet. This command line option has no effect if the
9071 file is already initialized.
9073 * systemd-nspawn gained a new --as-pid2 switch that invokes any
9074 specified command line as PID 2 rather than PID 1 in the
9075 container. In this mode PID 1 is a minimal stub init process that
9076 implements the special POSIX and Linux semantics of PID 1 regarding
9077 signal and child process management. Note that this stub init process
9078 is implemented in nspawn itself and requires no support from the
9079 container image. This new logic is useful to support running
9080 arbitrary commands in the container, as normal processes are
9081 generally not prepared to run as PID 1.
9083 * systemd-nspawn gained a new --chdir= switch for setting the current
9084 working directory for the process started in the container.
9086 * "journalctl /dev/sda" will now output all kernel log messages for
9087 specified device from the current boot, in addition to all devices
9088 that are parents of it. This should make log output about devices
9089 pretty useful, as long as kernel drivers attach enough metadata to
9090 the log messages. (The usual SATA drivers do.)
9092 * The sd-journal API gained two new calls
9093 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
9094 that report whether log data from /run or /var has been found.
9096 * journalctl gained a new switch "--fields" that prints all journal
9097 record field names currently in use in the journal. This is backed
9098 by two new sd-journal API calls sd_journal_enumerate_fields() and
9099 sd_journal_restart_fields().
9101 * Most configurable timeouts in systemd now expect an argument of
9102 "infinity" to turn them off, instead of "0" as before. The semantics
9103 from now on is that a timeout of "0" means "now", and "infinity"
9104 means "never". To maintain backwards compatibility, "0" continues to
9105 turn off previously existing timeout settings.
9107 * "systemctl reload-or-try-restart" has been renamed to "systemctl
9108 try-reload-or-restart" to clarify what it actually does: the "try"
9109 logic applies to both reloading and restarting, not just restarting.
9110 The old name continues to be accepted for compatibility.
9112 * On boot-up, when PID 1 detects that the system clock is behind the
9113 release date of the systemd version in use, the clock is now set
9114 to the latter. Previously, this was already done in timesyncd, in order
9115 to avoid running with clocks set to the various clock epochs such as
9116 1902, 1938 or 1970. With this change the logic is now done in PID 1
9117 in addition to timesyncd during early boot-up, so that it is enforced
9118 before the first process is spawned by systemd. Note that the logic
9119 in timesyncd remains, as it is more comprehensive and ensures
9120 clock monotonicity by maintaining a persistent timestamp file in
9121 /var. Since /var is generally not available in earliest boot or the
9122 initrd, this part of the logic remains in timesyncd, and is not done
9125 * Support for tweaking details in net_cls.class_id through the
9126 NetClass= configuration directive has been removed, as the kernel
9127 people have decided to deprecate that controller in cgroup v2.
9128 Userspace tools such as nftables are moving over to setting rules
9129 that are specific to the full cgroup path of a task, which obsoletes
9130 these controllers anyway. The NetClass= directive is kept around for
9131 legacy compatibility reasons. For a more in-depth description of the
9132 kernel change, please refer to the respective upstream commit:
9134 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
9136 * A new service setting RuntimeMaxSec= has been added that may be used
9137 to specify a maximum runtime for a service. If the timeout is hit, the
9138 service is terminated and put into a failure state.
9140 * A new service setting AmbientCapabilities= has been added. It allows
9141 configuration of additional Linux process capabilities that are
9142 passed to the activated processes. This is only available on very
9145 * The process resource limit settings in service units may now be used
9146 to configure hard and soft limits individually.
9148 * The various libsystemd APIs such as sd-bus or sd-event now publicly
9149 expose support for gcc's __attribute__((cleanup())) C extension.
9150 Specifically, for many object destructor functions alternative
9151 versions have been added that have names suffixed with "p" and take a
9152 pointer to a pointer to the object to destroy, instead of just a
9153 pointer to the object itself. This is useful because these destructor
9154 functions may be used directly as parameters to the cleanup
9155 construct. Internally, systemd has been a heavy user of this GCC
9156 extension for a long time, and with this change similar support is
9157 now available to consumers of the library outside of systemd. Note
9158 that by using this extension in your sources compatibility with old
9159 and strictly ANSI compatible C compilers is lost. However, all gcc or
9160 LLVM versions of recent years support this extension.
9162 * Timer units gained support for a new setting RandomizedDelaySec= that
9163 allows configuring some additional randomized delay to the configured
9164 time. This is useful to spread out timer events to avoid load peaks in
9165 clusters or larger setups.
9167 * Calendar time specifications now support sub-second accuracy.
9169 * Socket units now support listening on SCTP and UDP-lite protocol
9172 * The sd-event API now comes with a full set of man pages.
9174 * Older versions of systemd contained experimental support for
9175 compressing journal files and coredumps with the LZ4 compressor that
9176 was not compatible with the lz4 binary (due to API limitations of the
9177 lz4 library). This support has been removed; only support for files
9178 compatible with the lz4 binary remains. This LZ4 logic is now
9179 officially supported and no longer considered experimental.
9181 * The dkr image import logic has been removed again from importd. dkr's
9182 micro-services focus doesn't fit into the machine image focus of
9183 importd, and quickly got out of date with the upstream dkr API.
9185 * Creation of the /run/lock/lockdev/ directory was dropped from
9186 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
9187 been available for many years. If you still need this, you need to
9188 create your own tmpfiles.d config file with:
9190 d /run/lock/lockdev 0775 root lock -
9192 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
9193 and RebootArgument= have been moved from the [Service] section of
9194 unit files to [Unit], and they are now supported on all unit types,
9195 not just service units. Of course, systemd will continue to
9196 understand these settings also at the old location, in order to
9197 maintain compatibility.
9199 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
9200 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
9201 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
9202 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
9203 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
9204 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
9205 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
9206 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
9207 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
9208 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
9209 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
9210 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
9211 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
9212 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
9213 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
9214 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
9215 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
9216 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
9217 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9219 — Berlin, 2016-02-11
9223 * A number of properties previously only settable in unit
9224 files are now also available as properties to set when
9225 creating transient units programmatically via the bus, as it
9226 is exposed with systemd-run's --property=
9227 setting. Specifically, these are: SyslogIdentifier=,
9228 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
9229 EnvironmentFile=, ReadWriteDirectories=,
9230 ReadOnlyDirectories=, InaccessibleDirectories=,
9231 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
9233 * When creating transient services via the bus API it is now
9234 possible to pass in a set of file descriptors to use as
9235 STDIN/STDOUT/STDERR for the invoked process.
9237 * Slice units may now be created transiently via the bus APIs,
9238 similar to the way service and scope units may already be
9239 created transiently.
9241 * Wherever systemd expects a calendar timestamp specification
9242 (like in journalctl's --since= and --until= switches) UTC
9243 timestamps are now supported. Timestamps suffixed with "UTC"
9244 are now considered to be in Universal Time Coordinated
9245 instead of the local timezone. Also, timestamps may now
9246 optionally be specified with sub-second accuracy. Both of
9247 these additions also apply to recurring calendar event
9248 specification, such as OnCalendar= in timer units.
9250 * journalctl gained a new "--sync" switch that asks the
9251 journal daemon to write all so far unwritten log messages to
9252 disk and sync the files, before returning.
9254 * systemd-tmpfiles learned two new line types "q" and "Q" that
9255 operate like "v", but also set up a basic btrfs quota
9256 hierarchy when used on a btrfs file system with quota
9259 * tmpfiles' "v", "q" and "Q" will now create a plain directory
9260 instead of a subvolume (even on a btrfs file system) if the
9261 root directory is a plain directory, and not a
9262 subvolume. This should simplify things with certain chroot()
9263 environments which are not aware of the concept of btrfs
9266 * systemd-detect-virt gained a new --chroot switch to detect
9267 whether execution takes place in a chroot() environment.
9269 * CPUAffinity= now takes CPU index ranges in addition to
9272 * The various memory-related resource limit settings (such as
9273 LimitAS=) now understand the usual K, M, G, … suffixes to
9274 the base of 1024 (IEC). Similar, the time-related resource
9275 limit settings understand the usual min, h, day, … suffixes
9278 * There's a new system.conf setting DefaultTasksMax= to
9279 control the default TasksMax= setting for services and
9280 scopes running on the system. (TasksMax= is the primary
9281 setting that exposes the "pids" cgroup controller on systemd
9282 and was introduced in the previous systemd release.) The
9283 setting now defaults to 512, which means services that are
9284 not explicitly configured otherwise will only be able to
9285 create 512 processes or threads at maximum, from this
9286 version on. Note that this means that thread- or
9287 process-heavy services might need to be reconfigured to set
9288 TasksMax= to a higher value. It is sufficient to set
9289 TasksMax= in these specific unit files to a higher value, or
9290 even "infinity". Similar, there's now a logind.conf setting
9291 UserTasksMax= that defaults to 4096 and limits the total
9292 number of processes or tasks each user may own
9293 concurrently. nspawn containers also have the TasksMax=
9294 value set by default now, to 8192. Note that all of this
9295 only has an effect if the "pids" cgroup controller is
9296 enabled in the kernel. The general benefit of these changes
9297 should be a more robust and safer system, that provides a
9298 certain amount of per-service fork() bomb protection.
9300 * systemd-nspawn gained the new --network-veth-extra= switch
9301 to define additional and arbitrarily-named virtual Ethernet
9302 links between the host and the container.
9304 * A new service execution setting PassEnvironment= has been
9305 added that allows importing select environment variables
9306 from PID1's environment block into the environment block of
9309 * Timer units gained support for a new RemainAfterElapse=
9310 setting which takes a boolean argument. It defaults to on,
9311 exposing behaviour unchanged to previous releases. If set to
9312 off, timer units are unloaded after they elapsed if they
9313 cannot elapse again. This is particularly useful for
9314 transient timer units, which shall not stay around longer
9315 than until they first elapse.
9317 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
9318 default now (the kernel default is 16). This is beneficial
9319 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
9320 allows substantially larger numbers of queued
9321 datagrams. This should increase the capability of systemd to
9322 parallelize boot-up, as logging and sd_notify() are unlikely
9323 to stall execution anymore. If you need to change the value
9324 from the new defaults, use the usual sysctl.d/ snippets.
9326 * The compression framing format used by the journal or
9327 coredump processing has changed to be in line with what the
9328 official LZ4 tools generate. LZ4 compression support in
9329 systemd was considered unsupported previously, as the format
9330 was not compatible with the normal tools. With this release
9331 this has changed now, and it is hence safe for downstream
9332 distributions to turn it on. While not compressing as well
9333 as the XZ, LZ4 is substantially faster, which makes
9334 it a good default choice for the compression logic in the
9335 journal and in coredump handling.
9337 * Any reference to /etc/mtab has been dropped from
9338 systemd. The file has been obsolete since a while, but
9339 systemd refused to work on systems where it was incorrectly
9340 set up (it should be a symlink or non-existent). Please make
9341 sure to update to util-linux 2.27.1 or newer in conjunction
9342 with this systemd release, which also drops any reference to
9343 /etc/mtab. If you maintain a distribution make sure that no
9344 software you package still references it, as this is a
9345 likely source of bugs. There's also a glibc bug pending,
9346 asking for removal of any reference to this obsolete file:
9348 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
9350 Note that only util-linux versions built with
9351 --enable-libmount-force-mountinfo are supported.
9353 * Support for the ".snapshot" unit type has been removed. This
9354 feature turned out to be little useful and little used, and
9355 has now been removed from the core and from systemctl.
9357 * The dependency types RequiresOverridable= and
9358 RequisiteOverridable= have been removed from systemd. They
9359 have been used only very sparingly to our knowledge and
9360 other options that provide a similar effect (such as
9361 systemctl --mode=ignore-dependencies) are much more useful
9362 and commonly used. Moreover, they were only half-way
9363 implemented as the option to control behaviour regarding
9364 these dependencies was never added to systemctl. By removing
9365 these dependency types the execution engine becomes a bit
9366 simpler. Unit files that use these dependencies should be
9367 changed to use the non-Overridable dependency types
9368 instead. In fact, when parsing unit files with these
9369 options, that's what systemd will automatically convert them
9370 too, but it will also warn, asking users to fix the unit
9371 files accordingly. Removal of these dependency types should
9372 only affect a negligible number of unit files in the wild.
9374 * Behaviour of networkd's IPForward= option changed
9375 (again). It will no longer maintain a per-interface setting,
9376 but propagate one way from interfaces where this is enabled
9377 to the global kernel setting. The global setting will be
9378 enabled when requested by a network that is set up, but
9379 never be disabled again. This change was made to make sure
9380 IPv4 and IPv6 behaviour regarding packet forwarding is
9381 similar (as the Linux IPv6 stack does not support
9382 per-interface control of this setting) and to minimize
9385 * In unit files the behaviour of %u, %U, %h, %s has
9386 changed. These specifiers will now unconditionally resolve
9387 to the various user database fields of the user that the
9388 systemd instance is running as, instead of the user
9389 configured in the specific unit via User=. Note that this
9390 effectively doesn't change much, as resolving of these
9391 specifiers was already turned off in the --system instance
9392 of systemd, as we cannot do NSS lookups from PID 1. In the
9393 --user instance of systemd these specifiers where correctly
9394 resolved, but hardly made any sense, since the user instance
9395 lacks privileges to do user switches anyway, and User= is
9396 hence useless. Moreover, even in the --user instance of
9397 systemd behaviour was awkward as it would only take settings
9398 from User= assignment placed before the specifier into
9399 account. In order to unify and simplify the logic around
9400 this the specifiers will now always resolve to the
9401 credentials of the user invoking the manager (which in case
9402 of PID 1 is the root user).
9404 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
9405 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
9406 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
9407 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
9408 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
9409 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
9410 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
9411 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
9412 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
9413 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
9414 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
9415 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
9416 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
9417 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
9420 — Berlin, 2015-11-18
9424 * systemd now depends on util-linux v2.27. More specifically,
9425 the newly added mount monitor feature in libmount now
9426 replaces systemd's former own implementation.
9428 * libmount mandates /etc/mtab not to be regular file, and
9429 systemd now enforces this condition at early boot.
9430 /etc/mtab has been deprecated and warned about for a very
9431 long time, so systems running systemd should already have
9432 stopped having this file around as anything else than a
9433 symlink to /proc/self/mounts.
9435 * Support for the "pids" cgroup controller has been added. It
9436 allows accounting the number of tasks in a cgroup and
9437 enforcing limits on it. This adds two new setting
9438 TasksAccounting= and TasksMax= to each unit, as well as a
9439 global option DefaultTasksAccounting=.
9441 * Support for the "net_cls" cgroup controller has been added.
9442 It allows assigning a net class ID to each task in the
9443 cgroup, which can then be used in firewall rules and traffic
9444 shaping configurations. Note that the kernel netfilter net
9445 class code does not currently work reliably for ingress
9446 packets on unestablished sockets.
9448 This adds a new config directive called NetClass= to CGroup
9449 enabled units. Allowed values are positive numbers for fixed
9450 assignments and "auto" for picking a free value
9453 * 'systemctl is-system-running' now returns 'offline' if the
9454 system is not booted with systemd. This command can now be
9455 used as a substitute for 'systemd-notify --booted'.
9457 * Watchdog timeouts have been increased to 3 minutes for all
9458 in-tree service files. Apparently, disk IO issues are more
9459 frequent than we hoped, and user reported >1 minute waiting
9462 * 'machine-id-commit' functionality has been merged into
9463 'machine-id-setup --commit'. The separate binary has been
9466 * The WorkingDirectory= directive in unit files may now be set
9467 to the special value '~'. In this case, the working
9468 directory is set to the home directory of the user
9469 configured in User=.
9471 * "machinectl shell" will now open the shell in the home
9472 directory of the selected user by default.
9474 * The CrashChVT= configuration file setting is renamed to
9475 CrashChangeVT=, following our usual logic of not
9476 abbreviating unnecessarily. The old directive is still
9477 supported for compat reasons. Also, this directive now takes
9478 an integer value between 1 and 63, or a boolean value. The
9479 formerly supported '-1' value for disabling stays around for
9482 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
9483 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
9484 RootDirectory= properties can now be set for transient
9487 * The systemd-analyze tool gained a new "set-log-target" verb
9488 to change the logging target the system manager logs to
9489 dynamically during runtime. This is similar to how
9490 "systemd-analyze set-log-level" already changes the log
9493 * In nspawn /sys is now mounted as tmpfs, with only a selected
9494 set of subdirectories mounted in from the real sysfs. This
9495 enhances security slightly, and is useful for ensuring user
9496 namespaces work correctly.
9498 * Support for USB FunctionFS activation has been added. This
9499 allows implementation of USB gadget services that are
9500 activated as soon as they are requested, so that they don't
9501 have to run continuously, similar to classic socket
9504 * The "systemctl exit" command now optionally takes an
9505 additional parameter that sets the exit code to return from
9506 the systemd manager when exiting. This is only relevant when
9507 running the systemd user instance, or when running the
9508 system instance in a container.
9510 * sd-bus gained the new API calls sd_bus_path_encode_many()
9511 and sd_bus_path_decode_many() that allow easy encoding and
9512 decoding of multiple identifier strings inside a D-Bus
9513 object path. Another new call sd_bus_default_flush_close()
9514 has been added to flush and close per-thread default
9517 * systemd-cgtop gained support for a -M/--machine= switch to
9518 show the control groups within a certain container only.
9520 * "systemctl kill" gained support for an optional --fail
9521 switch. If specified the requested operation will fail of no
9522 processes have been killed, because the unit had no
9523 processes attached, or similar.
9525 * A new systemd.crash_reboot=1 kernel command line option has
9526 been added that triggers a reboot after crashing. This can
9527 also be set through CrashReboot= in systemd.conf.
9529 * The RuntimeDirectory= setting now understands unit
9530 specifiers like %i or %f.
9532 * A new (still internal) library API sd-ipv4acd has been added,
9533 that implements address conflict detection for IPv4. It's
9534 based on code from sd-ipv4ll, and will be useful for
9535 detecting DHCP address conflicts.
9537 * File descriptors passed during socket activation may now be
9538 named. A new API sd_listen_fds_with_names() is added to
9539 access the names. The default names may be overridden,
9540 either in the .socket file using the FileDescriptorName=
9541 parameter, or by passing FDNAME= when storing the file
9542 descriptors using sd_notify().
9544 * systemd-networkd gained support for:
9546 - Setting the IPv6 Router Advertisement settings via
9547 IPv6AcceptRouterAdvertisements= in .network files.
9549 - Configuring the HelloTimeSec=, MaxAgeSec= and
9550 ForwardDelaySec= bridge parameters in .netdev files.
9552 - Configuring PreferredSource= for static routes in
9555 * The "ask-password" framework used to query for LUKS harddisk
9556 passwords or SSL passwords during boot gained support for
9557 caching passwords in the kernel keyring, if it is
9558 available. This makes sure that the user only has to type in
9559 a passphrase once if there are multiple objects to unlock
9560 with the same one. Previously, such password caching was
9561 available only when Plymouth was used; this moves the
9562 caching logic into the systemd codebase itself. The
9563 "systemd-ask-password" utility gained a new --keyname=
9564 switch to control which kernel keyring key to use for
9565 caching a password in. This functionality is also useful for
9566 enabling display managers such as gdm to automatically
9567 unlock the user's GNOME keyring if its passphrase, the
9568 user's password and the harddisk password are the same, if
9569 gdm-autologin is used.
9571 * When downloading tar or raw images using "machinectl
9572 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
9573 file is now also downloaded, if it is available and stored
9574 next to the image file.
9576 * Units of type ".socket" gained a new boolean setting
9577 Writable= which is only useful in conjunction with
9578 ListenSpecial=. If true, enables opening the specified
9579 special file in O_RDWR mode rather than O_RDONLY mode.
9581 * systemd-rfkill has been reworked to become a singleton
9582 service that is activated through /dev/rfkill on each rfkill
9583 state change and saves the settings to disk. This way,
9584 systemd-rfkill is now compatible with devices that exist
9585 only intermittendly, and even restores state if the previous
9586 system shutdown was abrupt rather than clean.
9588 * The journal daemon gained support for vacuuming old journal
9589 files controlled by the number of files that shall remain,
9590 in addition to the already existing control by size and by
9591 date. This is useful as journal interleaving performance
9592 degrades with too many separate journal files, and allows
9593 putting an effective limit on them. The new setting defaults
9594 to 100, but this may be changed by setting SystemMaxFiles=
9595 and RuntimeMaxFiles= in journald.conf. Also, the
9596 "journalctl" tool gained the new --vacuum-files= switch to
9597 manually vacuum journal files to leave only the specified
9598 number of files in place.
9600 * udev will now create /dev/disk/by-path links for ATA devices
9601 on kernels where that is supported.
9603 * Galician, Serbian, Turkish and Korean translations were added.
9605 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
9606 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
9607 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
9608 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
9609 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
9610 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
9611 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
9612 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
9613 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
9614 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
9615 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
9616 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
9617 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
9618 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
9619 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
9620 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
9621 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
9622 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
9624 — Berlin, 2015-10-07
9628 * The DHCP implementation of systemd-networkd gained a set of
9631 - The DHCP server now supports emitting DNS and NTP
9632 information. It may be enabled and configured via
9633 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
9634 and NTP information is enabled, but no servers are
9635 configured, the corresponding uplink information (if there
9636 is any) is propagated.
9638 - Server and client now support transmission and reception
9639 of timezone information. It can be configured via the
9640 newly introduced network options UseTimezone=,
9641 EmitTimezone=, and Timezone=. Transmission of timezone
9642 information is enabled between host and containers by
9643 default now: the container will change its local timezone
9644 to what the host has set.
9646 - Lease timeouts can now be configured via
9647 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
9649 - The DHCP server improved on the stability of
9650 leases. Clients are more likely to get the same lease
9651 information back, even if the server loses state.
9653 - The DHCP server supports two new configuration options to
9654 control the lease address pool metrics, PoolOffset= and
9657 * The encapsulation limit of tunnels in systemd-networkd may
9658 now be configured via 'EncapsulationLimit='. It allows
9659 modifying the maximum additional levels of encapsulation
9660 that are permitted to be prepended to a packet.
9662 * systemd now supports the concept of user buses replacing
9663 session buses, if used with dbus-1.10 (and enabled via dbus
9664 --enable-user-session). It previously only supported this on
9665 kdbus-enabled systems, and this release expands this to
9666 'dbus-daemon' systems.
9668 * systemd-networkd now supports predictable interface names
9671 * systemd now optionally supports the new Linux kernel
9672 "unified" control group hierarchy. If enabled via the kernel
9673 command-line option 'systemd.unified_cgroup_hierarchy=1',
9674 systemd will try to mount the unified cgroup hierarchy
9675 directly on /sys/fs/cgroup. If not enabled, or not
9676 available, systemd will fall back to the legacy cgroup
9677 hierarchy setup, as before. Host system and containers can
9678 mix and match legacy and unified hierarchies as they
9679 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
9680 environment variable to individually select the hierarchy to
9681 use for executed containers. By default, nspawn will use the
9682 unified hierarchy for the containers if the host uses the
9683 unified hierarchy, and the legacy hierarchy otherwise.
9684 Please note that at this point the unified hierarchy is an
9685 experimental kernel feature and is likely to change in one
9686 of the next kernel releases. Therefore, it should not be
9687 enabled by default in downstream distributions yet. The
9688 minimum required kernel version for the unified hierarchy to
9689 work is 4.2. Note that when the unified hierarchy is used
9690 for the first time delegated access to controllers is
9691 safe. Because of this systemd-nspawn containers will get
9692 access to controllers now, as will systemd user
9693 sessions. This means containers and user sessions may now
9694 manage their own resources, partitioning up what the system
9697 * A new special scope unit "init.scope" has been introduced
9698 that encapsulates PID 1 of the system. It may be used to
9699 determine resource usage and enforce resource limits on PID
9700 1 itself. PID 1 hence moved out of the root of the control
9703 * The cgtop tool gained support for filtering out kernel
9704 threads when counting tasks in a control group. Also, the
9705 count of processes is now recursively summed up by
9706 default. Two options -k and --recursive= have been added to
9707 revert to old behaviour. The tool has also been updated to
9708 work correctly in containers now.
9710 * systemd-nspawn's --bind= and --bind-ro= options have been
9711 extended to allow creation of non-recursive bind mounts.
9713 * libsystemd gained two new calls sd_pid_get_cgroup() and
9714 sd_peer_get_cgroup() which return the control group path of
9715 a process or peer of a connected AF_UNIX socket. This
9716 function call is particularly useful when implementing
9717 delegated subtrees support in the control group hierarchy.
9719 * The "sd-event" event loop API of libsystemd now supports
9720 correct dequeuing of real-time signals, without losing
9723 * When systemd requests a polkit decision when managing units it
9724 will now add additional fields to the request, including unit
9725 name and desired operation. This enables more powerful polkit
9726 policies, that make decisions depending on these parameters.
9728 * nspawn learnt support for .nspawn settings files, that may
9729 accompany the image files or directories of containers, and
9730 may contain additional settings for the container. This is
9731 an alternative to configuring container parameters via the
9732 nspawn command line.
9734 Contributions from: Cristian Rodríguez, Daniel Mack, David
9735 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
9736 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
9737 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
9738 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
9739 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
9740 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
9741 Andersen, Tom Gundersen, Torstein Husebø
9743 — Berlin, 2015-09-08
9747 * machinectl gained a new verb 'shell' which opens a fresh
9748 shell on the target container or the host. It is similar to
9749 the existing 'login' command of machinectl, but spawns the
9750 shell directly without prompting for username or
9751 password. The pseudo machine '.host' now refers to the local
9752 host and is used by default. Hence, 'machinectl shell' can
9753 be used as replacement for 'su -' which spawns a session as
9754 a fresh systemd unit in a way that is fully isolated from
9755 the originating session.
9757 * systemd-networkd learned to cope with private-zone DHCP
9758 options and allows other programs to query the values.
9760 * SELinux access control when enabling/disabling units is no
9761 longer enforced with this release. The previous implementation
9762 was incorrect, and a new corrected implementation is not yet
9763 available. As unit file operations are still protected via
9764 polkit and D-Bus policy this is not a security problem. Yet,
9765 distributions which care about optimal SELinux support should
9766 probably not stabilize on this release.
9768 * sd-bus gained support for matches of type "arg0has=", that
9769 test for membership of strings in string arrays sent in bus
9772 * systemd-resolved now dumps the contents of its DNS and LLMNR
9773 caches to the logs on reception of the SIGUSR1 signal. This
9774 is useful to debug DNS behaviour.
9776 * The coredumpctl tool gained a new --directory= option to
9777 operate on journal files in a specific directory.
9779 * "systemctl reboot" and related commands gained a new
9780 "--message=" option which may be used to set a free-text
9781 wall message when shutting down or rebooting the
9782 system. This message is also logged, which is useful for
9783 figuring out the reason for a reboot or shutdown a
9786 * The "systemd-resolve-host" tool's -i switch now takes
9787 network interface numbers as alternative to interface names.
9789 * A new unit file setting for services has been introduced:
9790 UtmpMode= allows configuration of how precisely systemd
9791 handles utmp and wtmp entries for the service if this is
9792 enabled. This allows writing services that appear similar to
9793 user sessions in the output of the "w", "who", "last" and
9796 * systemd-resolved will now locally synthesize DNS resource
9797 records for the "localhost" and "gateway" domains as well as
9798 the local hostname. This should ensure that clients querying
9799 RRs via resolved will get similar results as those going via
9800 NSS, if nss-myhostname is enabled.
9802 Contributions from: Alastair Hughes, Alex Crawford, Daniel
9803 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
9804 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
9805 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
9806 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
9807 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
9808 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
9809 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
9810 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
9811 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
9812 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
9813 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
9815 — Berlin, 2015-08-27
9819 * The systemd-efi-boot-generator functionality was merged into
9820 systemd-gpt-auto-generator.
9822 * systemd-networkd now supports Group Policy for vxlan
9823 devices. It can be enabled via the new boolean configuration
9824 option called 'GroupPolicyExtension='.
9826 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
9827 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
9828 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
9830 — Berlin, 2015-07-31
9834 * The python-systemd code has been removed from the systemd repository.
9835 A new repository has been created which accommodates the code from
9836 now on, and we kindly ask distributions to create a separate package
9837 for this: https://github.com/systemd/python-systemd
9839 * The systemd daemon will now reload its main configuration
9840 (/etc/systemd/system.conf) on daemon-reload.
9842 * sd-dhcp now exposes vendor specific extensions via
9843 sd_dhcp_lease_get_vendor_specific().
9845 * systemd-networkd gained a number of new configuration options.
9847 - A new boolean configuration option for TAP devices called
9848 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
9849 device, thus allowing to send and receive GSO packets.
9851 - A new tunnel configuration option called 'CopyDSCP='.
9852 If enabled, the DSCP field of ip6 tunnels is copied into the
9853 decapsulated packet.
9855 - A set of boolean bridge configuration options were added.
9856 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
9857 and 'UnicastFlood=' are now parsed by networkd and applied to the
9858 respective bridge link device via the respective IFLA_BRPORT_*
9861 - A new string configuration option to override the hostname sent
9862 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
9863 is true, networkd will use the configured hostname instead of the
9864 system hostname when sending DHCP requests.
9866 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
9867 networkd will configure the IPv6 flow-label of the tunnel device
9868 according to RFC2460.
9870 - The 'macvtap' virtual network devices are now supported, similar to
9871 the already supported 'macvlan' devices.
9873 * systemd-resolved now implements RFC5452 to improve resilience against
9874 cache poisoning. Additionally, source port randomization is enabled
9875 by default to further protect against DNS spoofing attacks.
9877 * nss-mymachines now supports translating UIDs and GIDs of running
9878 containers with user-namespaces enabled. If a container 'foo'
9879 translates a host uid 'UID' to the container uid 'TUID', then
9880 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
9881 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
9882 mapped as 'vg-foo-TGID'.
9884 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
9885 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
9886 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
9887 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
9888 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
9889 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
9890 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
9891 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
9892 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
9893 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
9895 — Berlin, 2015-07-29
9899 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
9900 There are no known issues with current sysfs, and udev does not need
9901 or should be used to work around such bugs.
9903 * udev does no longer enable USB HID power management. Several reports
9904 indicate, that some devices cannot handle that setting.
9906 * The udev accelerometer helper was removed. The functionality
9907 is now fully included in iio-sensor-proxy. But this means,
9908 older iio-sensor-proxy versions will no longer provide
9909 accelerometer/orientation data with this systemd version.
9910 Please upgrade iio-sensor-proxy to version 1.0.
9912 * networkd gained a new configuration option IPv6PrivacyExtensions=
9913 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
9914 for Stateless Address") on selected networks.
9916 * For the sake of fewer build-time dependencies and less code in the
9917 main repository, the python bindings are about to be removed in the
9918 next release. A new repository has been created which accommodates
9919 the code from now on, and we kindly ask distributions to create a
9920 separate package for this. The removal will take place in v223.
9922 https://github.com/systemd/python-systemd
9924 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
9925 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
9926 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
9927 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
9928 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
9929 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
9930 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
9931 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
9932 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
9933 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
9935 — Berlin, 2015-07-07
9939 * The sd-bus.h and sd-event.h APIs have now been declared
9940 stable and have been added to the official interface of
9941 libsystemd.so. sd-bus implements an alternative D-Bus client
9942 library, that is relatively easy to use, very efficient and
9943 supports both classic D-Bus as well as kdbus as transport
9944 backend. sd-event is a generic event loop abstraction that
9945 is built around Linux epoll, but adds features such as event
9946 prioritization or efficient timer handling. Both APIs are good
9947 choices for C programs looking for a bus and/or event loop
9948 implementation that is minimal and does not have to be
9949 portable to other kernels.
9951 * kdbus support is no longer compile-time optional. It is now
9952 always built-in. However, it can still be disabled at
9953 runtime using the kdbus=0 kernel command line setting, and
9954 that setting may be changed to default to off, by specifying
9955 --disable-kdbus at build-time. Note though that the kernel
9956 command line setting has no effect if the kdbus.ko kernel
9957 module is not installed, in which case kdbus is (obviously)
9958 also disabled. We encourage all downstream distributions to
9959 begin testing kdbus by adding it to the kernel images in the
9960 development distributions, and leaving kdbus support in
9963 * The minimal required util-linux version has been bumped to
9966 * Support for chkconfig (--enable-chkconfig) was removed in
9967 favor of calling an abstraction tool
9968 /lib/systemd/systemd-sysv-install. This needs to be
9969 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
9970 in README for details.
9972 * If there's a systemd unit and a SysV init script for the
9973 same service name, and the user executes "systemctl enable"
9974 for it (or a related call), then this will now enable both
9975 (or execute the related operation on both), not just the
9978 * The libudev API documentation has been converted from gtkdoc
9981 * gudev has been removed from the systemd tree, it is now an
9984 * The systemd-cgtop tool learnt a new --raw switch to generate
9985 "raw" (machine parsable) output.
9987 * networkd's IPForwarding= .network file setting learnt the
9988 new setting "kernel", which ensures that networkd does not
9989 change the IP forwarding sysctl from the default kernel
9992 * The systemd-logind bus API now exposes a new boolean
9993 property "Docked" that reports whether logind considers the
9994 system "docked", i.e. connected to a docking station or not.
9996 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
9997 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
9998 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
9999 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
10000 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
10001 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
10002 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
10003 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
10004 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
10005 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
10006 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
10007 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
10008 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
10009 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
10010 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
10011 Fink, Zbigniew Jędrzejewski-Szmek
10013 — Berlin, 2015-06-19
10017 * The gudev library has been extracted into a separate repository
10018 available at: https://git.gnome.org/browse/libgudev/
10019 It is now managed as part of the Gnome project. Distributions
10020 are recommended to pass --disable-gudev to systemd and use
10021 gudev from the Gnome project instead. gudev is still included
10022 in systemd, for now. It will be removed soon, though. Please
10023 also see the announcement-thread on systemd-devel:
10024 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
10026 * systemd now exposes a CPUUsageNSec= property for each
10027 service unit on the bus, that contains the overall consumed
10028 CPU time of a service (the sum of what each process of the
10029 service consumed). This value is only available if
10030 CPUAccounting= is turned on for a service, and is then shown
10031 in the "systemctl status" output.
10033 * Support for configuring alternative mappings of the old SysV
10034 runlevels to systemd targets has been removed. They are now
10035 hardcoded in a way that runlevels 2, 3, 4 all map to
10036 multi-user.target and 5 to graphical.target (which
10037 previously was already the default behaviour).
10039 * The auto-mounter logic gained support for mount point
10040 expiry, using a new TimeoutIdleSec= setting in .automount
10041 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
10043 * The EFI System Partition (ESP) as mounted to /boot by
10044 systemd-efi-boot-generator will now be unmounted
10045 automatically after 2 minutes of not being used. This should
10046 minimize the risk of ESP corruptions.
10048 * New /etc/fstab options x-systemd.requires= and
10049 x-systemd.requires-mounts-for= are now supported to express
10050 additional dependencies for mounts. This is useful for
10051 journaling file systems that support external journal
10052 devices or overlay file systems that require underlying file
10053 systems to be mounted.
10055 * systemd does not support direct live-upgrades (via systemctl
10056 daemon-reexec) from versions older than v44 anymore. As no
10057 distribution we are aware of shipped such old versions in a
10058 stable release this should not be problematic.
10060 * When systemd forks off a new per-connection service instance
10061 it will now set the $REMOTE_ADDR environment variable to the
10062 remote IP address, and $REMOTE_PORT environment variable to
10063 the remote IP port. This behaviour is similar to the
10064 corresponding environment variables defined by CGI.
10066 * systemd-networkd gained support for uplink failure
10067 detection. The BindCarrier= option allows binding interface
10068 configuration dynamically to the link sense of other
10069 interfaces. This is useful to achieve behaviour like in
10072 * systemd-networkd gained support for configuring the DHCP
10073 client identifier to use when requesting leases.
10075 * systemd-networkd now has a per-network UseNTP= option to
10076 configure whether NTP server information acquired via DHCP
10077 is passed on to services like systemd-timesyncd.
10079 * systemd-networkd gained support for vti6 tunnels.
10081 * Note that systemd-networkd manages the sysctl variable
10082 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
10083 it is configured for since v219. The variable controls IP
10084 forwarding, and is a per-interface alternative to the global
10085 /proc/sys/net/ipv[46]/ip_forward. This setting is
10086 configurable in the IPForward= option, which defaults to
10087 "no". This means if networkd is used for an interface it is
10088 no longer sufficient to set the global sysctl option to turn
10089 on IP forwarding! Instead, the .network file option
10090 IPForward= needs to be turned on! Note that the
10091 implementation of this behaviour was broken in v219 and has
10092 been fixed in v220.
10094 * Many bonding and vxlan options are now configurable in
10097 * systemd-nspawn gained a new --property= setting to set unit
10098 properties for the container scope. This is useful for
10099 setting resource parameters (e.g. "CPUShares=500") on
10100 containers started from the command line.
10102 * systemd-nspawn gained a new --private-users= switch to make
10103 use of user namespacing available on recent Linux kernels.
10105 * systemd-nspawn may now be called as part of a shell pipeline
10106 in which case the pipes used for stdin and stdout are passed
10107 directly to the process invoked in the container, without
10108 indirection via a pseudo tty.
10110 * systemd-nspawn gained a new switch to control the UNIX
10111 signal to use when killing the init process of the container
10112 when shutting down.
10114 * systemd-nspawn gained a new --overlay= switch for mounting
10115 overlay file systems into the container using the new kernel
10118 * When a container image is imported via systemd-importd and
10119 the host file system is not btrfs, a loopback block device
10120 file is created in /var/lib/machines.raw with a btrfs file
10121 system inside. It is then mounted to /var/lib/machines to
10122 enable btrfs features for container management. The loopback
10123 file and btrfs file system is grown as needed when container
10124 images are imported via systemd-importd.
10126 * systemd-machined/systemd-importd gained support for btrfs
10127 quota, to enforce container disk space limits on disk. This
10128 is exposed in "machinectl set-limit".
10130 * systemd-importd now can import containers from local .tar,
10131 .raw and .qcow2 images, and export them to .tar and .raw. It
10132 can also import dkr v2 images now from the network (on top
10135 * systemd-importd gained support for verifying downloaded
10136 images with gpg2 (previously only gpg1 was supported).
10138 * systemd-machined, systemd-logind, systemd: most bus calls are
10139 now accessible to unprivileged processes via polkit. Also,
10140 systemd-logind will now allow users to kill their own sessions
10141 without further privileges or authorization.
10143 * systemd-shutdownd has been removed. This service was
10144 previously responsible for implementing scheduled shutdowns
10145 as exposed in /usr/bin/shutdown's time parameter. This
10146 functionality has now been moved into systemd-logind and is
10147 accessible via a bus interface.
10149 * "systemctl reboot" gained a new switch --firmware-setup that
10150 can be used to reboot into the EFI firmware setup, if that
10151 is available. systemd-logind now exposes an API on the bus
10152 to trigger such reboots, in case graphical desktop UIs want
10153 to cover this functionality.
10155 * "systemctl enable", "systemctl disable" and "systemctl mask"
10156 now support a new "--now" switch. If specified the units
10157 that are enabled will also be started, and the ones
10158 disabled/masked also stopped.
10160 * The Gummiboot EFI boot loader tool has been merged into
10161 systemd, and renamed to "systemd-boot". The bootctl tool has been
10162 updated to support systemd-boot.
10164 * An EFI kernel stub has been added that may be used to create
10165 kernel EFI binaries that contain not only the actual kernel,
10166 but also an initrd, boot splash, command line and OS release
10167 information. This combined binary can then be signed as a
10168 single image, so that the firmware can verify it all in one
10169 step. systemd-boot has special support for EFI binaries created
10170 like this and can extract OS release information from them
10171 and show them in the boot menu. This functionality is useful
10172 to implement cryptographically verified boot schemes.
10174 * Optional support has been added to systemd-fsck to pass
10175 fsck's progress report to an AF_UNIX socket in the file
10178 * udev will no longer create device symlinks for all block devices by
10179 default. A deny list for excluding special block devices from this
10180 logic has been turned into an allow list that requires picking block
10181 devices explicitly that require device symlinks.
10183 * A new (currently still internal) API sd-device.h has been
10184 added to libsystemd. This modernized API is supposed to
10185 replace libudev eventually. In fact, already much of libudev
10186 is now just a wrapper around sd-device.h.
10188 * A new hwdb database for storing metadata about pointing
10189 stick devices has been added.
10191 * systemd-tmpfiles gained support for setting file attributes
10192 similar to the "chattr" tool with new 'h' and 'H' lines.
10194 * systemd-journald will no longer unconditionally set the
10195 btrfs NOCOW flag on new journal files. This is instead done
10196 with tmpfiles snippet using the new 'h' line type. This
10197 allows easy disabling of this logic, by masking the
10198 journal-nocow.conf tmpfiles file.
10200 * systemd-journald will now translate audit message types to
10201 human readable identifiers when writing them to the
10202 journal. This should improve readability of audit messages.
10204 * The LUKS logic gained support for the offset= and skip=
10205 options in /etc/crypttab, as previously implemented by
10208 * /usr/lib/os-release gained a new optional field VARIANT= for
10209 distributions that support multiple variants (such as a
10210 desktop edition, a server edition, …)
10212 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
10213 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
10214 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
10215 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
10216 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
10217 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
10218 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
10219 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
10220 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
10221 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
10222 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
10223 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
10224 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
10225 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
10226 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
10227 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
10228 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
10229 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
10230 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
10231 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
10232 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
10233 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
10234 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
10235 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
10236 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
10237 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
10238 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
10240 — Berlin, 2015-05-22
10244 * Introduce a new API "sd-hwdb.h" for querying the hardware
10245 metadata database. With this minimal interface one can query
10246 and enumerate the udev hwdb, decoupled from the old libudev
10247 library. libudev's interface for this is now only a wrapper
10248 around sd-hwdb. A new tool systemd-hwdb has been added to
10249 interface with and update the database.
10251 * When any of systemd's tools copies files (for example due to
10252 tmpfiles' C lines) a btrfs reflink will attempted first,
10253 before bytewise copying is done.
10255 * systemd-nspawn gained a new --ephemeral switch. When
10256 specified a btrfs snapshot is taken of the container's root
10257 directory, and immediately removed when the container
10258 terminates again. Thus, a container can be started whose
10259 changes never alter the container's root directory, and are
10260 lost on container termination. This switch can also be used
10261 for starting a container off the root file system of the
10262 host without affecting the host OS. This switch is only
10263 available on btrfs file systems.
10265 * systemd-nspawn gained a new --template= switch. It takes the
10266 path to a container tree to use as template for the tree
10267 specified via --directory=, should that directory be
10268 missing. This allows instantiating containers dynamically,
10269 on first run. This switch is only available on btrfs file
10272 * When a .mount unit refers to a mount point on which multiple
10273 mounts are stacked, and the .mount unit is stopped all of
10274 the stacked mount points will now be unmounted until no
10275 mount point remains.
10277 * systemd now has an explicit notion of supported and
10278 unsupported unit types. Jobs enqueued for unsupported unit
10279 types will now fail with an "unsupported" error code. More
10280 specifically .swap, .automount and .device units are not
10281 supported in containers, .busname units are not supported on
10282 non-kdbus systems. .swap and .automount are also not
10283 supported if their respective kernel compile time options
10286 * machinectl gained support for two new "copy-from" and
10287 "copy-to" commands for copying files from a running
10288 container to the host or vice versa.
10290 * machinectl gained support for a new "bind" command to bind
10291 mount host directories into local containers. This is
10292 currently only supported for nspawn containers.
10294 * networkd gained support for configuring bridge forwarding
10295 database entries (fdb) from .network files.
10297 * A new tiny daemon "systemd-importd" has been added that can
10298 download container images in tar, raw, qcow2 or dkr formats,
10299 and make them available locally in /var/lib/machines, so
10300 that they can run as nspawn containers. The daemon can GPG
10301 verify the downloads (not supported for dkr, since it has no
10302 provisions for verifying downloads). It will transparently
10303 decompress bz2, xz, gzip compressed downloads if necessary,
10304 and restore sparse files on disk. The daemon uses privilege
10305 separation to ensure the actual download logic runs with
10306 fewer privileges than the daemon itself. machinectl has
10307 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
10308 make the functionality of importd available to the
10309 user. With this in place the Fedora and Ubuntu "Cloud"
10310 images can be downloaded and booted as containers unmodified
10311 (the Fedora images lack the appropriate GPG signature files
10312 currently, so they cannot be verified, but this will change
10313 soon, hopefully). Note that downloading images is currently
10314 only fully supported on btrfs.
10316 * machinectl is now able to list container images found in
10317 /var/lib/machines, along with some metadata about sizes of
10318 disk and similar. If the directory is located on btrfs and
10319 quota is enabled, this includes quota display. A new command
10320 "image-status" has been added that shows additional
10321 information about images.
10323 * machinectl is now able to clone container images
10324 efficiently, if the underlying file system (btrfs) supports
10325 it, with the new "machinectl clone" command. It also
10326 gained commands for renaming and removing images, as well as
10327 marking them read-only or read-write (supported also on
10328 legacy file systems).
10330 * networkd gained support for collecting LLDP network
10331 announcements, from hardware that supports this. This is
10332 shown in networkctl output.
10334 * systemd-run gained support for a new -t (--pty) switch for
10335 invoking a binary on a pty whose input and output is
10336 connected to the invoking terminal. This allows executing
10337 processes as system services while interactively
10338 communicating with them via the terminal. Most interestingly
10339 this is supported across container boundaries. Invoking
10340 "systemd-run -t /bin/bash" is an alternative to running a
10341 full login session, the difference being that the former
10342 will not register a session, nor go through the PAM session
10345 * tmpfiles gained support for a new "v" line type for creating
10346 btrfs subvolumes. If the underlying file system is a legacy
10347 file system, this automatically degrades to creating a
10348 normal directory. Among others /var/lib/machines is now
10349 created like this at boot, should it be missing.
10351 * The directory /var/lib/containers/ has been deprecated and
10352 been replaced by /var/lib/machines. The term "machines" has
10353 been used in the systemd context as generic term for both
10354 VMs and containers, and hence appears more appropriate for
10355 this, as the directory can also contain raw images bootable
10358 * systemd-nspawn when invoked with -M but without --directory=
10359 or --image= is now capable of searching for the container
10360 root directory, subvolume or disk image automatically, in
10361 /var/lib/machines. systemd-nspawn@.service has been updated
10362 to make use of this, thus allowing it to be used for raw
10365 * A new machines.target unit has been introduced that is
10366 supposed to group all containers/VMs invoked as services on
10367 the system. systemd-nspawn@.service has been updated to
10368 integrate with that.
10370 * machinectl gained a new "start" command, for invoking a
10371 container as a service. "machinectl start foo" is mostly
10372 equivalent to "systemctl start systemd-nspawn@foo.service",
10373 but handles escaping in a nicer way.
10375 * systemd-nspawn will now mount most of the cgroupfs tree
10376 read-only into each container, with the exception of the
10377 container's own subtree in the name=systemd hierarchy.
10379 * journald now sets the special FS_NOCOW file flag for its
10380 journal files. This should improve performance on btrfs, by
10381 avoiding heavy fragmentation when journald's write-pattern
10382 is used on COW file systems. It degrades btrfs' data
10383 integrity guarantees for the files to the same levels as for
10384 ext3/ext4 however. This should be OK though as journald does
10385 its own data integrity checks and all its objects are
10386 checksummed on disk. Also, journald should handle btrfs disk
10387 full events a lot more gracefully now, by processing SIGBUS
10388 errors, and not relying on fallocate() anymore.
10390 * When journald detects that journal files it is writing to
10391 have been deleted it will immediately start new journal
10394 * systemd now provides a way to store file descriptors
10395 per-service in PID 1. This is useful for daemons to ensure
10396 that fds they require are not lost during a daemon
10397 restart. The fds are passed to the daemon on the next
10398 invocation in the same way socket activation fds are
10399 passed. This is now used by journald to ensure that the
10400 various sockets connected to all the system's stdout/stderr
10401 are not lost when journald is restarted. File descriptors
10402 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
10403 an extension to sd_notify(). Note that a limit is enforced
10404 on the number of fds a service can store in PID 1, and it
10405 defaults to 0, so that no fds may be stored, unless this is
10406 explicitly turned on.
10408 * The default TERM variable to use for units connected to a
10409 terminal, when no other value is explicitly is set is now
10410 vt220 rather than vt102. This should be fairly safe still,
10411 but allows PgUp/PgDn work.
10413 * The /etc/crypttab option header= as known from Debian is now
10416 * "loginctl user-status" and "loginctl session-status" will
10417 now show the last 10 lines of log messages of the
10418 user/session following the status output. Similar,
10419 "machinectl status" will show the last 10 log lines
10420 associated with a virtual machine or container
10421 service. (Note that this is usually not the log messages
10422 done in the VM/container itself, but simply what the
10423 container manager logs. For nspawn this includes all console
10426 * "loginctl session-status" without further argument will now
10427 show the status of the session of the caller. Similar,
10428 "lock-session", "unlock-session", "activate",
10429 "enable-linger", "disable-linger" may now be called without
10430 session/user parameter in which case they apply to the
10431 caller's session/user.
10433 * An X11 session scriptlet is now shipped that uploads
10434 $DISPLAY and $XAUTHORITY into the environment of the systemd
10435 --user daemon if a session begins. This should improve
10436 compatibility with X11 enabled applications run as systemd
10439 * Generators are now subject to masking via /etc and /run, the
10440 same way as unit files.
10442 * networkd .network files gained support for configuring
10443 per-link IPv4/IPv6 packet forwarding as well as IPv4
10444 masquerading. This is by default turned on for veth links to
10445 containers, as registered by systemd-nspawn. This means that
10446 nspawn containers run with --network-veth will now get
10447 automatic routed access to the host's networks without any
10448 further configuration or setup, as long as networkd runs on
10451 * systemd-nspawn gained the --port= (-p) switch to expose TCP
10452 or UDP posts of a container on the host. With this in place
10453 it is possible to run containers with private veth links
10454 (--network-veth), and have their functionality exposed on
10455 the host as if their services were running directly on the
10458 * systemd-nspawn's --network-veth switch now gained a short
10459 version "-n", since with the changes above it is now truly
10460 useful out-of-the-box. The systemd-nspawn@.service has been
10461 updated to make use of it too by default.
10463 * systemd-nspawn will now maintain a per-image R/W lock, to
10464 ensure that the same image is not started more than once
10465 writable. (It's OK to run an image multiple times
10466 simultaneously in read-only mode.)
10468 * systemd-nspawn's --image= option is now capable of
10469 dissecting and booting MBR and GPT disk images that contain
10470 only a single active Linux partition. Previously it
10471 supported only GPT disk images with proper GPT type
10472 IDs. This allows running cloud images from major
10473 distributions directly with systemd-nspawn, without
10476 * In addition to collecting mouse dpi data in the udev
10477 hardware database, there's now support for collecting angle
10478 information for mouse scroll wheels. The database is
10479 supposed to guarantee similar scrolling behavior on mice
10480 that it knows about. There's also support for collecting
10481 information about Touchpad types.
10483 * udev's input_id built-in will now also collect touch screen
10484 dimension data and attach it to probed devices.
10486 * /etc/os-release gained support for a Distribution Privacy
10489 * networkd gained support for creating "ipvlan", "gretap",
10490 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
10492 * systemd-tmpfiles gained support for "a" lines for setting
10495 * systemd-nspawn will now mount /tmp in the container to
10496 tmpfs, automatically.
10498 * systemd now exposes the memory.usage_in_bytes cgroup
10499 attribute and shows it for each service in the "systemctl
10500 status" output, if available.
10502 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
10503 immediate reboot is triggered. This useful if shutdown is
10504 hung and is unable to complete, to expedite the
10505 operation. Note that this kind of reboot will still unmount
10506 all file systems, and hence should not result in fsck being
10507 run on next reboot.
10509 * A .device unit for an optical block device will now be
10510 considered active only when a medium is in the drive. Also,
10511 mount units are now bound to their backing devices thus
10512 triggering automatic unmounting when devices become
10513 unavailable. With this in place systemd will now
10514 automatically unmount left-over mounts when a CD-ROM is
10515 ejected or an USB stick is yanked from the system.
10517 * networkd-wait-online now has support for waiting for
10518 specific interfaces only (with globbing), and for giving up
10519 after a configurable timeout.
10521 * networkd now exits when idle. It will be automatically
10522 restarted as soon as interfaces show up, are removed or
10523 change state. networkd will stay around as long as there is
10524 at least one DHCP state machine or similar around, that keep
10527 * networkd may now configure IPv6 link-local addressing in
10528 addition to IPv4 link-local addressing.
10530 * The IPv6 "token" for use in SLAAC may now be configured for
10531 each .network interface in networkd.
10533 * Routes configured with networkd may now be assigned a scope
10536 * networkd's [Match] sections now support globbing and lists
10537 of multiple space-separated matches per item.
10539 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
10540 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
10541 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
10542 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
10543 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
10544 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
10545 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
10546 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
10547 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
10548 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
10549 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
10550 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
10551 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
10552 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
10553 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
10554 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
10555 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
10556 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
10557 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
10558 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
10559 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
10560 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
10561 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
10562 Hoffmann, Zbigniew Jędrzejewski-Szmek
10564 — Berlin, 2015-02-16
10568 * When querying unit file enablement status (for example via
10569 "systemctl is-enabled"), a new state "indirect" is now known
10570 which indicates that a unit might not be enabled itself, but
10571 another unit listed in its Also= setting might be.
10573 * Similar to the various existing ConditionXYZ= settings for
10574 units, there are now matching AssertXYZ= settings. While
10575 failing conditions cause a unit to be skipped, but its job
10576 to succeed, failing assertions declared like this will cause
10577 a unit start operation and its job to fail.
10579 * hostnamed now knows a new chassis type "embedded".
10581 * systemctl gained a new "edit" command. When used on a unit
10582 file, this allows extending unit files with .d/ drop-in
10583 configuration snippets or editing the full file (after
10584 copying it from /usr/lib to /etc). This will invoke the
10585 user's editor (as configured with $EDITOR), and reload the
10586 modified configuration after editing.
10588 * "systemctl status" now shows the suggested enablement state
10589 for a unit, as declared in the (usually vendor-supplied)
10590 system preset files.
10592 * nss-myhostname will now resolve the single-label hostname
10593 "gateway" to the locally configured default IP routing
10594 gateways, ordered by their metrics. This assigns a stable
10595 name to the used gateways, regardless which ones are
10596 currently configured. Note that the name will only be
10597 resolved after all other name sources (if nss-myhostname is
10598 configured properly) and should hence not negatively impact
10599 systems that use the single-label hostname "gateway" in
10602 * systemd-inhibit now allows filtering by mode when listing
10605 * Scope and service units gained a new "Delegate" boolean
10606 property, which, when set, allows processes running inside the
10607 unit to further partition resources. This is primarily
10608 useful for systemd user instances as well as container
10611 * journald will now pick up audit messages directly from
10612 the kernel, and log them like any other log message. The
10613 audit fields are split up and fully indexed. This means that
10614 journalctl in many ways is now a (nicer!) alternative to
10615 ausearch, the traditional audit client. Note that this
10616 implements only a minimal audit client. If you want the
10617 special audit modes like reboot-on-log-overflow, please use
10618 the traditional auditd instead, which can be used in
10619 parallel to journald.
10621 * The ConditionSecurity= unit file option now understands the
10622 special string "audit" to check whether auditing is
10625 * journalctl gained two new commands --vacuum-size= and
10626 --vacuum-time= to delete old journal files until the
10627 remaining ones take up no more than the specified size on disk,
10628 or are not older than the specified time.
10630 * A new, native PPPoE library has been added to sd-network,
10631 systemd's library of light-weight networking protocols. This
10632 library will be used in a future version of networkd to
10633 enable PPPoE communication without an external pppd daemon.
10635 * The busctl tool now understands a new "capture" verb that
10636 works similar to "monitor", but writes a packet capture
10637 trace to STDOUT that can be redirected to a file which is
10638 compatible with libcap's capture file format. This can then
10639 be loaded in Wireshark and similar tools to inspect bus
10642 * The busctl tool now understands a new "tree" verb that shows
10643 the object trees of a specific service on the bus, or of all
10646 * The busctl tool now understands a new "introspect" verb that
10647 shows all interfaces and members of objects on the bus,
10648 including their signature and values. This is particularly
10649 useful to get more information about bus objects shown by
10650 the new "busctl tree" command.
10652 * The busctl tool now understands new verbs "call",
10653 "set-property" and "get-property" for invoking bus method
10654 calls, setting and getting bus object properties in a
10657 * busctl gained a new --augment-creds= argument that controls
10658 whether the tool shall augment credential information it
10659 gets from the bus with data from /proc, in a possibly
10662 * nspawn's --link-journal= switch gained two new values
10663 "try-guest" and "try-host" that work like "guest" and
10664 "host", but do not fail if the host has no persistent
10665 journaling enabled. -j is now equivalent to
10666 --link-journal=try-guest.
10668 * macvlan network devices created by nspawn will now have
10669 stable MAC addresses.
10671 * A new SmackProcessLabel= unit setting has been added, which
10672 controls the SMACK security label processes forked off by
10673 the respective unit shall use.
10675 * If compiled with --enable-xkbcommon, systemd-localed will
10676 verify x11 keymap settings by compiling the given keymap. It
10677 will spew out warnings if the compilation fails. This
10678 requires libxkbcommon to be installed.
10680 * When a coredump is collected, a larger number of metadata
10681 fields is now collected and included in the journal records
10682 created for it. More specifically, control group membership,
10683 environment variables, memory maps, working directory,
10684 chroot directory, /proc/$PID/status, and a list of open file
10685 descriptors is now stored in the log entry.
10687 * The udev hwdb now contains DPI information for mice. For
10690 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
10692 * All systemd programs that read standalone configuration
10693 files in /etc now also support a corresponding series of
10694 .conf.d configuration directories in /etc/, /run/,
10695 /usr/local/lib/, /usr/lib/, and (if configured with
10696 --enable-split-usr) /lib/. In particular, the following
10697 configuration files now have corresponding configuration
10698 directories: system.conf user.conf, logind.conf,
10699 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
10700 resolved.conf, timesyncd.conf, journal-remote.conf, and
10701 journal-upload.conf. Note that distributions should use the
10702 configuration directories in /usr/lib/; the directories in
10703 /etc/ are reserved for the system administrator.
10705 * systemd-rfkill will no longer take the rfkill device name
10706 into account when storing rfkill state on disk, as the name
10707 might be dynamically assigned and not stable. Instead, the
10708 ID_PATH udev variable combined with the rfkill type (wlan,
10709 bluetooth, …) is used.
10711 * A new service systemd-machine-id-commit.service has been
10712 added. When used on systems where /etc is read-only during
10713 boot, and /etc/machine-id is not initialized (but an empty
10714 file), this service will copy the temporary machine ID
10715 created as replacement into /etc after the system is fully
10716 booted up. This is useful for systems that are freshly
10717 installed with a non-initialized machine ID, but should get
10718 a fixed machine ID for subsequent boots.
10720 * networkd's .netdev files now provide a large set of
10721 configuration parameters for VXLAN devices. Similarly, the
10722 bridge port cost parameter is now configurable in .network
10723 files. There's also new support for configuring IP source
10724 routing. networkd .link files gained support for a new
10725 OriginalName= match that is useful to match against the
10726 original interface name the kernel assigned. .network files
10727 may include MTU= and MACAddress= fields for altering the MTU
10728 and MAC address while being connected to a specific network
10731 * The LUKS logic gained supported for configuring
10732 UUID-specific key files. There's also new support for naming
10733 LUKS device from the kernel command line, using the new
10734 luks.name= argument.
10736 * Timer units may now be transiently created via the bus API
10737 (this was previously already available for scope and service
10738 units). In addition it is now possible to create multiple
10739 transient units at the same time with a single bus call. The
10740 "systemd-run" tool has been updated to make use of this for
10741 running commands on a specified time, in at(1)-style.
10743 * tmpfiles gained support for "t" lines, for assigning
10744 extended attributes to files. Among other uses this may be
10745 used to assign SMACK labels to files.
10747 Contributions from: Alin Rauta, Alison Chaiken, Andrej
10748 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
10749 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
10750 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
10751 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
10752 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
10753 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
10754 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
10755 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
10756 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
10757 Michael Biebl, Michael Chapman, Michael Marineau, Michal
10758 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
10759 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
10760 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
10761 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
10762 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10763 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
10764 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
10766 — Berlin, 2014-12-10
10770 * journalctl gained the new options -t/--identifier= to match
10771 on the syslog identifier (aka "tag"), as well as --utc to
10772 show log timestamps in the UTC timezone. journalctl now also
10773 accepts -n/--lines=all to disable line capping in a pager.
10775 * journalctl gained a new switch, --flush, that synchronously
10776 flushes logs from /run/log/journal to /var/log/journal if
10777 persistent storage is enabled. systemd-journal-flush.service
10778 now waits until the operation is complete.
10780 * Services can notify the manager before they start a reload
10781 (by sending RELOADING=1) or shutdown (by sending
10782 STOPPING=1). This allows the manager to track and show the
10783 internal state of daemons and closes a race condition when
10784 the process is still running but has closed its D-Bus
10787 * Services with Type=oneshot do not have to have any ExecStart
10790 * User units are now loaded also from
10791 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
10792 /run/systemd/user directory that was already previously
10793 supported, but is under the control of the user.
10795 * Job timeouts (i.e. timeouts on the time a job that is
10796 queued stays in the run queue) can now optionally result in
10797 immediate reboot or power-off actions (JobTimeoutAction= and
10798 JobTimeoutRebootArgument=). This is useful on ".target"
10799 units, to limit the maximum time a target remains
10800 undispatched in the run queue, and to trigger an emergency
10801 operation in such a case. This is now used by default to
10802 turn off the system if boot-up (as defined by everything in
10803 basic.target) hangs and does not complete for at least
10804 15min. Also, if power-off or reboot hang for at least 30min
10805 an immediate power-off/reboot operation is triggered. This
10806 functionality is particularly useful to increase reliability
10807 on embedded devices, but also on laptops which might
10808 accidentally get powered on when carried in a backpack and
10809 whose boot stays stuck in a hard disk encryption passphrase
10812 * systemd-logind can be configured to also handle lid switch
10813 events even when the machine is docked or multiple displays
10814 are attached (HandleLidSwitchDocked= option).
10816 * A helper binary and a service have been added which can be
10817 used to resume from hibernation in the initramfs. A
10818 generator will parse the resume= option on the kernel
10819 command line to trigger resume.
10821 * A user console daemon systemd-consoled has been
10822 added. Currently, it is a preview, and will so far open a
10823 single terminal on each session of the user marked as
10824 Desktop=systemd-console.
10826 * Route metrics can be specified for DHCP routes added by
10829 * The SELinux context of socket-activated services can be set
10830 from the information provided by the networking stack
10831 (SELinuxContextFromNet= option).
10833 * Userspace firmware loading support has been removed and
10834 the minimum supported kernel version is thus bumped to 3.7.
10836 * Timeout for udev workers has been increased from 1 to 3
10837 minutes, but a warning will be printed after 1 minute to
10838 help diagnose kernel modules that take a long time to load.
10840 * Udev rules can now remove tags on devices with TAG-="foobar".
10842 * systemd's readahead implementation has been removed. In many
10843 circumstances it didn't give expected benefits even for
10844 rotational disk drives and was becoming less relevant in the
10845 age of SSDs. As none of the developers has been using
10846 rotating media anymore, and nobody stepped up to actively
10847 maintain this component of systemd it has now been removed.
10849 * Swap units can use Options= to specify discard options.
10850 Discard options specified for swaps in /etc/fstab are now
10853 * Docker containers are now detected as a separate type of
10856 * The Password Agent protocol gained support for queries where
10857 the user input is shown, useful e.g. for user names.
10858 systemd-ask-password gained a new --echo option to turn that
10861 * The default sysctl.d/ snippets will now set:
10863 net.core.default_qdisc = fq_codel
10865 This selects Fair Queuing Controlled Delay as the default
10866 queuing discipline for network interfaces. fq_codel helps
10867 fight the network bufferbloat problem. It is believed to be
10868 a good default with no tuning required for most workloads.
10869 Downstream distributions may override this choice. On 10Gbit
10870 servers that do not do forwarding, "fq" may perform better.
10871 Systems without a good clocksource should use "pfifo_fast".
10873 * If kdbus is enabled during build a new option BusPolicy= is
10874 available for service units, that allows locking all service
10875 processes into a stricter bus policy, in order to limit
10876 access to various bus services, or even hide most of them
10877 from the service's view entirely.
10879 * networkctl will now show the .network and .link file
10880 networkd has applied to a specific interface.
10882 * sd-login gained a new API call sd_session_get_desktop() to
10883 query which desktop environment has been selected for a
10886 * UNIX utmp support is now compile-time optional to support
10887 legacy-free systems.
10889 * systemctl gained two new commands "add-wants" and
10890 "add-requires" for pulling in units from specific targets
10893 * If the word "rescue" is specified on the kernel command line
10894 the system will now boot into rescue mode (aka
10895 rescue.target), which was previously available only by
10896 specifying "1" or "systemd.unit=rescue.target" on the kernel
10897 command line. This new kernel command line option nicely
10898 mirrors the already existing "emergency" kernel command line
10901 * New kernel command line options mount.usr=, mount.usrflags=,
10902 mount.usrfstype= have been added that match root=, rootflags=,
10903 rootfstype= but allow mounting a specific file system to
10906 * The $NOTIFY_SOCKET is now also passed to control processes of
10907 services, not only the main process.
10909 * This version reenables support for fsck's -l switch. This
10910 means at least version v2.25 of util-linux is required for
10911 operation, otherwise dead-locks on device nodes may
10912 occur. Again: you need to update util-linux to at least
10913 v2.25 when updating systemd to v217.
10915 * The "multi-seat-x" tool has been removed from systemd, as
10916 its functionality has been integrated into X servers 1.16,
10917 and the tool is hence redundant. It is recommended to update
10918 display managers invoking this tool to simply invoke X
10919 directly from now on, again.
10921 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
10922 message flag has been added for all of systemd's polkit
10923 authenticated method calls has been added. In particular this
10924 now allows optional interactive authorization via polkit for
10925 many of PID1's privileged operations such as unit file
10926 enabling and disabling.
10928 * "udevadm hwdb --update" learnt a new switch "--usr" for
10929 placing the rebuilt hardware database in /usr instead of
10930 /etc. When used only hardware database entries stored in
10931 /usr will be used, and any user database entries in /etc are
10932 ignored. This functionality is useful for vendors to ship a
10933 pre-built database on systems where local configuration is
10934 unnecessary or unlikely.
10936 * Calendar time specifications in .timer units now also
10937 understand the strings "semi-annually", "quarterly" and
10938 "minutely" as shortcuts (in addition to the preexisting
10939 "annually", "hourly", …).
10941 * systemd-tmpfiles will now correctly create files in /dev
10942 at boot which are marked for creation only at boot. It is
10943 recommended to always create static device nodes with 'c!'
10944 and 'b!', so that they are created only at boot and not
10945 overwritten at runtime.
10947 * When the watchdog logic is used for a service (WatchdogSec=)
10948 and the watchdog timeout is hit the service will now be
10949 terminated with SIGABRT (instead of just SIGTERM), in order
10950 to make sure a proper coredump and backtrace is
10951 generated. This ensures that hanging services will result in
10952 similar coredump/backtrace behaviour as services that hit a
10953 segmentation fault.
10955 Contributions from: Andreas Henriksson, Andrei Borzenkov,
10956 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
10957 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
10958 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
10959 Herrmann, David Sommerseth, David Strauss, Emil Renner
10960 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
10961 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
10962 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
10963 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
10964 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
10965 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
10966 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
10967 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
10968 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
10969 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
10970 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
10971 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
10972 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
10973 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
10974 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
10975 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
10978 — Berlin, 2014-10-28
10982 * timedated no longer reads NTP implementation unit names from
10983 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
10984 implementations should add a
10986 Conflicts=systemd-timesyncd.service
10988 to their unit files to take over and replace systemd's NTP
10989 default functionality.
10991 * systemd-sysusers gained a new line type "r" for configuring
10992 which UID/GID ranges to allocate system users/groups
10993 from. Lines of type "u" may now add an additional column
10994 that specifies the home directory for the system user to be
10995 created. Also, systemd-sysusers may now optionally read user
10996 information from STDIN instead of a file. This is useful for
10997 invoking it from RPM preinst scriptlets that need to create
10998 users before the first RPM file is installed since these
10999 files might need to be owned by them. A new
11000 %sysusers_create_inline RPM macro has been introduced to do
11001 just that. systemd-sysusers now updates the shadow files as
11002 well as the user/group databases, which should enhance
11003 compatibility with certain tools like grpck.
11005 * A number of bus APIs of PID 1 now optionally consult polkit to
11006 permit access for otherwise unprivileged clients under certain
11007 conditions. Note that this currently doesn't support
11008 interactive authentication yet, but this is expected to be
11009 added eventually, too.
11011 * /etc/machine-info now has new fields for configuring the
11012 deployment environment of the machine, as well as the
11013 location of the machine. hostnamectl has been updated with
11014 new command to update these fields.
11016 * systemd-timesyncd has been updated to automatically acquire
11017 NTP server information from systemd-networkd, which might
11018 have been discovered via DHCP.
11020 * systemd-resolved now includes a caching DNS stub resolver
11021 and a complete LLMNR name resolution implementation. A new
11022 NSS module "nss-resolve" has been added which can be used
11023 instead of glibc's own "nss-dns" to resolve hostnames via
11024 systemd-resolved. Hostnames, addresses and arbitrary RRs may
11025 be resolved via systemd-resolved D-Bus APIs. In contrast to
11026 the glibc internal resolver systemd-resolved is aware of
11027 multi-homed system, and keeps DNS server and caches separate
11028 and per-interface. Queries are sent simultaneously on all
11029 interfaces that have DNS servers configured, in order to
11030 properly handle VPNs and local LANs which might resolve
11031 separate sets of domain names. systemd-resolved may acquire
11032 DNS server information from systemd-networkd automatically,
11033 which in turn might have discovered them via DHCP. A tool
11034 "systemd-resolve-host" has been added that may be used to
11035 query the DNS logic in resolved. systemd-resolved implements
11036 IDNA and automatically uses IDNA or UTF-8 encoding depending
11037 on whether classic DNS or LLMNR is used as transport. In the
11038 next releases we intend to add a DNSSEC and mDNS/DNS-SD
11039 implementation to systemd-resolved.
11041 * A new NSS module nss-mymachines has been added, that
11042 automatically resolves the names of all local registered
11043 containers to their respective IP addresses.
11045 * A new client tool "networkctl" for systemd-networkd has been
11046 added. It currently is entirely passive and will query
11047 networking configuration from udev, rtnetlink and networkd,
11048 and present it to the user in a very friendly
11049 way. Eventually, we hope to extend it to become a full
11050 control utility for networkd.
11052 * .socket units gained a new DeferAcceptSec= setting that
11053 controls the kernels' TCP_DEFER_ACCEPT sockopt for
11054 TCP. Similarly, support for controlling TCP keep-alive
11055 settings has been added (KeepAliveTimeSec=,
11056 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
11057 turning off Nagle's algorithm on TCP has been added
11060 * logind learned a new session type "web", for use in projects
11061 like Cockpit which register web clients as PAM sessions.
11063 * timer units with at least one OnCalendar= setting will now
11064 be started only after time-sync.target has been
11065 reached. This way they will not elapse before the system
11066 clock has been corrected by a local NTP client or
11067 similar. This is particular useful on RTC-less embedded
11068 machines, that come up with an invalid system clock.
11070 * systemd-nspawn's --network-veth= switch should now result in
11071 stable MAC addresses for both the outer and the inner side
11074 * systemd-nspawn gained a new --volatile= switch for running
11075 container instances with /etc or /var unpopulated.
11077 * The kdbus client code has been updated to use the new Linux
11078 3.17 memfd subsystem instead of the old kdbus-specific one.
11080 * systemd-networkd's DHCP client and server now support
11081 FORCERENEW. There are also new configuration options to
11082 configure the vendor client identifier and broadcast mode
11085 * systemd will no longer inform the kernel about the current
11086 timezone, as this is necessarily incorrect and racy as the
11087 kernel has no understanding of DST and similar
11088 concepts. This hence means FAT timestamps will be always
11089 considered UTC, similar to what Android is already
11090 doing. Also, when the RTC is configured to the local time
11091 (rather than UTC) systemd will never synchronize back to it,
11092 as this might confuse Windows at a later boot.
11094 * systemd-analyze gained a new command "verify" for offline
11095 validation of unit files.
11097 * systemd-networkd gained support for a couple of additional
11098 settings for bonding networking setups. Also, the metric for
11099 statically configured routes may now be configured. For
11100 network interfaces where this is appropriate the peer IP
11101 address may now be configured.
11103 * systemd-networkd's DHCP client will no longer request
11104 broadcasting by default, as this tripped up some networks.
11105 For hardware where broadcast is required the feature should
11106 be switched back on using RequestBroadcast=yes.
11108 * systemd-networkd will now set up IPv4LL addresses (when
11109 enabled) even if DHCP is configured successfully.
11111 * udev will now default to respect network device names given
11112 by the kernel when the kernel indicates that these are
11113 predictable. This behavior can be tweaked by changing
11114 NamePolicy= in the relevant .link file.
11116 * A new library systemd-terminal has been added that
11117 implements full TTY stream parsing and rendering. This
11118 library is supposed to be used later on for implementing a
11119 full userspace VT subsystem, replacing the current kernel
11122 * A new tool systemd-journal-upload has been added to push
11123 journal data to a remote system running
11124 systemd-journal-remote.
11126 * journald will no longer forward all local data to another
11127 running syslog daemon. This change has been made because
11128 rsyslog (which appears to be the most commonly used syslog
11129 implementation these days) no longer makes use of this, and
11130 instead pulls the data out of the journal on its own. Since
11131 forwarding the messages to a non-existent syslog server is
11132 more expensive than we assumed we have now turned this
11133 off. If you run a syslog server that is not a recent rsyslog
11134 version, you have to turn this option on again
11135 (ForwardToSyslog= in journald.conf).
11137 * journald now optionally supports the LZ4 compressor for
11138 larger journal fields. This compressor should perform much
11139 better than XZ which was the previous default.
11141 * machinectl now shows the IP addresses of local containers,
11142 if it knows them, plus the interface name of the container.
11144 * A new tool "systemd-escape" has been added that makes it
11145 easy to escape strings to build unit names and similar.
11147 * sd_notify() messages may now include a new ERRNO= field
11148 which is parsed and collected by systemd and shown among the
11149 "systemctl status" output for a service.
11151 * A new component "systemd-firstboot" has been added that
11152 queries the most basic systemd information (timezone,
11153 hostname, root password) interactively on first
11154 boot. Alternatively it may also be used to provision these
11155 things offline on OS images installed into directories.
11157 * The default sysctl.d/ snippets will now set
11159 net.ipv4.conf.default.promote_secondaries=1
11161 This has the benefit of no flushing secondary IP addresses
11162 when primary addresses are removed.
11164 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
11165 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
11166 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
11167 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
11168 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
11169 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
11170 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
11171 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
11172 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
11173 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
11174 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
11175 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
11176 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
11177 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
11178 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
11180 — Berlin, 2014-08-19
11184 * A new tool systemd-sysusers has been added. This tool
11185 creates system users and groups in /etc/passwd and
11186 /etc/group, based on static declarative system user/group
11187 definitions in /usr/lib/sysusers.d/. This is useful to
11188 enable factory resets and volatile systems that boot up with
11189 an empty /etc directory, and thus need system users and
11190 groups created during early boot. systemd now also ships
11191 with two default sysusers.d/ files for the most basic
11192 users and groups systemd and the core operating system
11195 * A new tmpfiles snippet has been added that rebuilds the
11196 essential files in /etc on boot, should they be missing.
11198 * A directive for ensuring automatic clean-up of
11199 /var/cache/man/ has been removed from the default
11200 configuration. This line should now be shipped by the man
11201 implementation. The necessary change has been made to the
11202 man-db implementation. Note that you need to update your man
11203 implementation to one that ships this line, otherwise no
11204 automatic clean-up of /var/cache/man will take place.
11206 * A new condition ConditionNeedsUpdate= has been added that
11207 may conditionalize services to only run when /etc or /var
11208 are "older" than the vendor operating system resources in
11209 /usr. This is useful for reconstructing or updating /etc
11210 after an offline update of /usr or a factory reset, on the
11211 next reboot. Services that want to run once after such an
11212 update or reset should use this condition and order
11213 themselves before the new systemd-update-done.service, which
11214 will mark the two directories as fully updated. A number of
11215 service files have been added making use of this, to rebuild
11216 the udev hardware database, the journald message catalog and
11217 dynamic loader cache (ldconfig). The systemd-sysusers tool
11218 described above also makes use of this now. With this in
11219 place it is now possible to start up a minimal operating
11220 system with /etc empty cleanly. For more information on the
11221 concepts involved see this recent blog story:
11223 http://0pointer.de/blog/projects/stateless.html
11225 * A new system group "input" has been introduced, and all
11226 input device nodes get this group assigned. This is useful
11227 for system-level software to get access to input devices. It
11228 complements what is already done for "audio" and "video".
11230 * systemd-networkd learnt minimal DHCPv4 server support in
11231 addition to the existing DHCPv4 client support. It also
11232 learnt DHCPv6 client and IPv6 Router Solicitation client
11233 support. The DHCPv4 client gained support for static routes
11234 passed in from the server. Note that the [DHCPv4] section
11235 known in older systemd-networkd versions has been renamed to
11236 [DHCP] and is now also used by the DHCPv6 client. Existing
11237 .network files using settings of this section should be
11238 updated, though compatibility is maintained. Optionally, the
11239 client hostname may now be sent to the DHCP server.
11241 * networkd gained support for vxlan virtual networks as well
11242 as tun/tap and dummy devices.
11244 * networkd gained support for automatic allocation of address
11245 ranges for interfaces from a system-wide pool of
11246 addresses. This is useful for dynamically managing a large
11247 number of interfaces with a single network configuration
11248 file. In particular this is useful to easily assign
11249 appropriate IP addresses to the veth links of a large number
11250 of nspawn instances.
11252 * RPM macros for processing sysusers, sysctl and binfmt
11253 drop-in snippets at package installation time have been
11256 * The /etc/os-release file should now be placed in
11257 /usr/lib/os-release. The old location is automatically
11258 created as symlink. /usr/lib is the more appropriate
11259 location of this file, since it shall actually describe the
11260 vendor operating system shipped in /usr, and not the
11261 configuration stored in /etc.
11263 * .mount units gained a new boolean SloppyOptions= setting
11264 that maps to mount(8)'s -s option which enables permissive
11265 parsing of unknown mount options.
11267 * tmpfiles learnt a new "L+" directive which creates a symlink
11268 but (unlike "L") deletes a pre-existing file first, should
11269 it already exist and not already be the correct
11270 symlink. Similarly, "b+", "c+" and "p+" directives have been
11271 added as well, which create block and character devices, as
11272 well as fifos in the filesystem, possibly removing any
11273 pre-existing files of different types.
11275 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
11276 'argument' field (which so far specified the source to
11277 symlink/copy the files from) is now optional. If omitted the
11278 same file os copied from /usr/share/factory/ suffixed by the
11279 full destination path. This is useful for populating /etc
11280 with essential files, by copying them from vendor defaults
11281 shipped in /usr/share/factory/etc.
11283 * A new command "systemctl preset-all" has been added that
11284 applies the service preset settings to all installed unit
11285 files. A new switch --preset-mode= has been added that
11286 controls whether only enable or only disable operations
11289 * A new command "systemctl is-system-running" has been added
11290 that allows checking the overall state of the system, for
11291 example whether it is fully up and running.
11293 * When the system boots up with an empty /etc, the equivalent
11294 to "systemctl preset-all" is executed during early boot, to
11295 make sure all default services are enabled after a factory
11298 * systemd now contains a minimal preset file that enables the
11299 most basic services systemd ships by default.
11301 * Unit files' [Install] section gained a new DefaultInstance=
11302 field for defining the default instance to create if a
11303 template unit is enabled with no instance specified.
11305 * A new passive target cryptsetup-pre.target has been added
11306 that may be used by services that need to make they run and
11307 finish before the first LUKS cryptographic device is set up.
11309 * The /dev/loop-control and /dev/btrfs-control device nodes
11310 are now owned by the "disk" group by default, opening up
11311 access to this group.
11313 * systemd-coredump will now automatically generate a
11314 stack trace of all core dumps taking place on the system,
11315 based on elfutils' libdw library. This stack trace is logged
11318 * systemd-coredump may now optionally store coredumps directly
11319 on disk (in /var/lib/systemd/coredump, possibly compressed),
11320 instead of storing them unconditionally in the journal. This
11321 mode is the new default. A new configuration file
11322 /etc/systemd/coredump.conf has been added to configure this
11323 and other parameters of systemd-coredump.
11325 * coredumpctl gained a new "info" verb to show details about a
11326 specific coredump. A new switch "-1" has also been added
11327 that makes sure to only show information about the most
11328 recent entry instead of all entries. Also, as the tool is
11329 generally useful now the "systemd-" prefix of the binary
11330 name has been removed. Distributions that want to maintain
11331 compatibility with the old name should add a symlink from
11332 the old name to the new name.
11334 * journald's SplitMode= now defaults to "uid". This makes sure
11335 that unprivileged users can access their own coredumps with
11336 coredumpctl without restrictions.
11338 * New kernel command line options "systemd.wants=" (for
11339 pulling an additional unit during boot), "systemd.mask="
11340 (for masking a specific unit for the boot), and
11341 "systemd.debug-shell" (for enabling the debug shell on tty9)
11342 have been added. This is implemented in the new generator
11343 "systemd-debug-generator".
11345 * systemd-nspawn will now by default filter a couple of
11346 syscalls for containers, among them those required for
11347 kernel module loading, direct x86 IO port access, swap
11348 management, and kexec. Most importantly though
11349 open_by_handle_at() is now prohibited for containers,
11350 closing a hole similar to a recently discussed vulnerability
11351 in docker regarding access to files on file hierarchies the
11352 container should normally not have access to. Note that, for
11353 nspawn, we generally make no security claims anyway (and
11354 this is explicitly documented in the man page), so this is
11355 just a fix for one of the most obvious problems.
11357 * A new man page file-hierarchy(7) has been added that
11358 contains a minimized, modernized version of the file system
11359 layout systemd expects, similar in style to the FHS
11360 specification or hier(5). A new tool systemd-path(1) has
11361 been added to query many of these paths for the local
11364 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
11365 longer done. Since the directory now has a per-user size
11366 limit, and is cleaned on logout this appears unnecessary,
11367 in particular since this now brings the lifecycle of this
11368 directory closer in line with how IPC objects are handled.
11370 * systemd.pc now exports a number of additional directories,
11371 including $libdir (which is useful to identify the library
11372 path for the primary architecture of the system), and a
11373 couple of drop-in directories.
11375 * udev's predictable network interface names now use the dev_port
11376 sysfs attribute, introduced in linux 3.15 instead of dev_id to
11377 distinguish between ports of the same PCI function. dev_id should
11378 only be used for ports using the same HW address, hence the need
11381 * machined has been updated to export the OS version of a
11382 container (read from /etc/os-release and
11383 /usr/lib/os-release) on the bus. This is now shown in
11384 "machinectl status" for a machine.
11386 * A new service setting RestartForceExitStatus= has been
11387 added. If configured to a set of exit signals or process
11388 return values, the service will be restarted when the main
11389 daemon process exits with any of them, regardless of the
11392 * systemctl's -H switch for connecting to remote systemd
11393 machines has been extended so that it may be used to
11394 directly connect to a specific container on the
11395 host. "systemctl -H root@foobar:waldi" will now connect as
11396 user "root" to host "foobar", and then proceed directly to
11397 the container named "waldi". Note that currently you have to
11398 authenticate as user "root" for this to work, as entering
11399 containers is a privileged operation.
11401 Contributions from: Andreas Henriksson, Benjamin Steinwender,
11402 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
11403 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
11404 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
11405 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
11406 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
11407 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
11408 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
11409 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
11410 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
11411 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
11412 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
11414 — Berlin, 2014-07-03
11418 * As an experimental feature, udev now tries to lock the
11419 disk device node (flock(LOCK_SH|LOCK_NB)) while it
11420 executes events for the disk or any of its partitions.
11421 Applications like partitioning programs can lock the
11422 disk device node (flock(LOCK_EX)) and claim temporary
11423 device ownership that way; udev will entirely skip all event
11424 handling for this disk and its partitions. If the disk
11425 was opened for writing, the close will trigger a partition
11426 table rescan in udev's "watch" facility, and if needed
11427 synthesize "change" events for the disk and all its partitions.
11428 This is now unconditionally enabled, and if it turns out to
11429 cause major problems, we might turn it on only for specific
11430 devices, or might need to disable it entirely. Device Mapper
11431 devices are excluded from this logic.
11433 * We temporarily dropped the "-l" switch for fsck invocations,
11434 since they collide with the flock() logic above. util-linux
11435 upstream has been changed already to avoid this conflict,
11436 and we will re-add "-l" as soon as util-linux with this
11437 change has been released.
11439 * The dependency on libattr has been removed. Since a long
11440 time, the extended attribute calls have moved to glibc, and
11441 libattr is thus unnecessary.
11443 * Virtualization detection works without privileges now. This
11444 means the systemd-detect-virt binary no longer requires
11445 CAP_SYS_PTRACE file capabilities, and our daemons can run
11446 with fewer privileges.
11448 * systemd-networkd now runs under its own "systemd-network"
11449 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
11450 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
11451 loses the ability to write to files owned by root this way.
11453 * Similarly, systemd-resolved now runs under its own
11454 "systemd-resolve" user with no capabilities remaining.
11456 * Similarly, systemd-bus-proxyd now runs under its own
11457 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
11459 * systemd-networkd gained support for setting up "veth"
11460 virtual Ethernet devices for container connectivity, as well
11461 as GRE and VTI tunnels.
11463 * systemd-networkd will no longer automatically attempt to
11464 manually load kernel modules necessary for certain tunnel
11465 transports. Instead, it is assumed the kernel loads them
11466 automatically when required. This only works correctly on
11467 very new kernels. On older kernels, please consider adding
11468 the kernel modules to /etc/modules-load.d/ as a work-around.
11470 * The resolv.conf file systemd-resolved generates has been
11471 moved to /run/systemd/resolve/. If you have a symlink from
11472 /etc/resolv.conf, it might be necessary to correct it.
11474 * Two new service settings, ProtectHome= and ProtectSystem=,
11475 have been added. When enabled, they will make the user data
11476 (such as /home) inaccessible or read-only and the system
11477 (such as /usr) read-only, for specific services. This allows
11478 very light-weight per-service sandboxing to avoid
11479 modifications of user data or system files from
11480 services. These two new switches have been enabled for all
11481 of systemd's long-running services, where appropriate.
11483 * Socket units gained new SocketUser= and SocketGroup=
11484 settings to set the owner user and group of AF_UNIX sockets
11485 and FIFOs in the file system.
11487 * Socket units gained a new RemoveOnStop= setting. If enabled,
11488 all FIFOS and sockets in the file system will be removed
11489 when the specific socket unit is stopped.
11491 * Socket units gained a new Symlinks= setting. It takes a list
11492 of symlinks to create to file system sockets or FIFOs
11493 created by the specific Unix sockets. This is useful to
11494 manage symlinks to socket nodes with the same lifecycle as
11497 * The /dev/log socket and /dev/initctl FIFO have been moved to
11498 /run, and have been replaced by symlinks. This allows
11499 connecting to these facilities even if PrivateDevices=yes is
11500 used for a service (which makes /dev/log itself unavailable,
11501 but /run is left). This also has the benefit of ensuring
11502 that /dev only contains device nodes, directories and
11503 symlinks, and nothing else.
11505 * sd-daemon gained two new calls sd_pid_notify() and
11506 sd_pid_notifyf(). They are similar to sd_notify() and
11507 sd_notifyf(), but allow overriding of the source PID of
11508 notification messages if permissions permit this. This is
11509 useful to send notify messages on behalf of a different
11510 process (for example, the parent process). The
11511 systemd-notify tool has been updated to make use of this
11512 when sending messages (so that notification messages now
11513 originate from the shell script invoking systemd-notify and
11514 not the systemd-notify process itself. This should minimize
11515 a race where systemd fails to associate notification
11516 messages to services when the originating process already
11519 * A new "on-abnormal" setting for Restart= has been added. If
11520 set, it will result in automatic restarts on all "abnormal"
11521 reasons for a process to exit, which includes unclean
11522 signals, core dumps, timeouts and watchdog timeouts, but
11523 does not include clean and unclean exit codes or clean
11524 signals. Restart=on-abnormal is an alternative for
11525 Restart=on-failure for services that shall be able to
11526 terminate and avoid restarts on certain errors, by
11527 indicating so with an unclean exit code. Restart=on-failure
11528 or Restart=on-abnormal is now the recommended setting for
11529 all long-running services.
11531 * If the InaccessibleDirectories= service setting points to a
11532 mount point (or if there are any submounts contained within
11533 it), it is now attempted to completely unmount it, to make
11534 the file systems truly unavailable for the respective
11537 * The ReadOnlyDirectories= service setting and
11538 systemd-nspawn's --read-only parameter are now recursively
11539 applied to all submounts, too.
11541 * Mount units may now be created transiently via the bus APIs.
11543 * The support for SysV and LSB init scripts has been removed
11544 from the systemd daemon itself. Instead, it is now
11545 implemented as a generator that creates native systemd units
11546 from these scripts when needed. This enables us to remove a
11547 substantial amount of legacy code from PID 1, following the
11548 fact that many distributions only ship a very small number
11549 of LSB/SysV init scripts nowadays.
11551 * Privileged Xen (dom0) domains are not considered
11552 virtualization anymore by the virtualization detection
11553 logic. After all, they generally have unrestricted access to
11554 the hardware and usually are used to manage the unprivileged
11557 * systemd-tmpfiles gained a new "C" line type, for copying
11558 files or entire directories.
11560 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
11561 lines. So far, they have been non-globbing versions of the
11562 latter, and have thus been redundant. In future, it is
11563 recommended to only use "z". "m" has hence been removed
11564 from the documentation, even though it stays supported.
11566 * A tmpfiles snippet to recreate the most basic structure in
11567 /var has been added. This is enough to create the /var/run →
11568 /run symlink and create a couple of structural
11569 directories. This allows systems to boot up with an empty or
11570 volatile /var. Of course, while with this change, the core OS
11571 now is capable with dealing with a volatile /var, not all
11572 user services are ready for it. However, we hope that sooner
11573 or later, many service daemons will be changed upstream so
11574 that they are able to automatically create their necessary
11575 directories in /var at boot, should they be missing. This is
11576 the first step to allow state-less systems that only require
11577 the vendor image for /usr to boot.
11579 * systemd-nspawn has gained a new --tmpfs= switch to mount an
11580 empty tmpfs instance to a specific directory. This is
11581 particularly useful for making use of the automatic
11582 reconstruction of /var (see above), by passing --tmpfs=/var.
11584 * Access modes specified in tmpfiles snippets may now be
11585 prefixed with "~", which indicates that they shall be masked
11586 by whether the existing file or directory is currently
11587 writable, readable or executable at all. Also, if specified,
11588 the sgid/suid/sticky bits will be masked for all
11591 * A new passive target unit "network-pre.target" has been
11592 added which is useful for services that shall run before any
11593 network is configured, for example firewall scripts.
11595 * The "floppy" group that previously owned the /dev/fd*
11596 devices is no longer used. The "disk" group is now used
11597 instead. Distributions should probably deprecate usage of
11600 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
11601 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
11602 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
11603 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
11604 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
11605 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
11608 — Berlin, 2014-06-11
11612 * A new "systemd-timesyncd" daemon has been added for
11613 synchronizing the system clock across the network. It
11614 implements an SNTP client. In contrast to NTP
11615 implementations such as chrony or the NTP reference server,
11616 this only implements a client side, and does not bother with
11617 the full NTP complexity, focusing only on querying time from
11618 one remote server and synchronizing the local clock to
11619 it. Unless you intend to serve NTP to networked clients or
11620 want to connect to local hardware clocks, this simple NTP
11621 client should be more than appropriate for most
11622 installations. The daemon runs with minimal privileges, and
11623 has been hooked up with networkd to only operate when
11624 network connectivity is available. The daemon saves the
11625 current clock to disk every time a new NTP sync has been
11626 acquired, and uses this to possibly correct the system clock
11627 early at bootup, in order to accommodate for systems that
11628 lack an RTC such as the Raspberry Pi and embedded devices,
11629 and to make sure that time monotonically progresses on these
11630 systems, even if it is not always correct. To make use of
11631 this daemon, a new system user and group "systemd-timesync"
11632 needs to be created on installation of systemd.
11634 * The queue "seqnum" interface of libudev has been disabled, as
11635 it was generally incompatible with device namespacing as
11636 sequence numbers of devices go "missing" if the devices are
11637 part of a different namespace.
11639 * "systemctl list-timers" and "systemctl list-sockets" gained
11640 a --recursive switch for showing units of these types also
11641 for all local containers, similar in style to the already
11642 supported --recursive switch for "systemctl list-units".
11644 * A new RebootArgument= setting has been added for service
11645 units, which may be used to specify a kernel reboot argument
11646 to use when triggering reboots with StartLimitAction=.
11648 * A new FailureAction= setting has been added for service
11649 units which may be used to specify an operation to trigger
11650 when a service fails. This works similarly to
11651 StartLimitAction=, but unlike it, controls what is done
11652 immediately rather than only after several attempts to
11653 restart the service in question.
11655 * hostnamed got updated to also expose the kernel name,
11656 release, and version on the bus. This is useful for
11657 executing commands like hostnamectl with the -H switch.
11658 systemd-analyze makes use of this to properly display
11659 details when running non-locally.
11661 * The bootchart tool can now show cgroup information in the
11662 graphs it generates.
11664 * The CFS CPU quota cgroup attribute is now exposed for
11665 services. The new CPUQuota= switch has been added for this
11666 which takes a percentage value. Setting this will have the
11667 result that a service may never get more CPU time than the
11668 specified percentage, even if the machine is otherwise idle.
11670 * systemd-networkd learned IPIP and SIT tunnel support.
11672 * LSB init scripts exposing a dependency on $network will now
11673 get a dependency on network-online.target rather than simply
11674 network.target. This should bring LSB handling closer to
11675 what it was on SysV systems.
11677 * A new fsck.repair= kernel option has been added to control
11678 how fsck shall deal with unclean file systems at boot.
11680 * The (.ini) configuration file parser will now silently ignore
11681 sections whose names begin with "X-". This may be used to maintain
11682 application-specific extension sections in unit files.
11684 * machined gained a new API to query the IP addresses of
11685 registered containers. "machinectl status" has been updated
11686 to show these addresses in its output.
11688 * A new call sd_uid_get_display() has been added to the
11689 sd-login APIs for querying the "primary" session of a
11690 user. The "primary" session of the user is elected from the
11691 user's sessions and generally a graphical session is
11692 preferred over a text one.
11694 * A minimal systemd-resolved daemon has been added. It
11695 currently simply acts as a companion to systemd-networkd and
11696 manages resolv.conf based on per-interface DNS
11697 configuration, possibly supplied via DHCP. In the long run
11698 we hope to extend this into a local DNSSEC enabled DNS and
11701 * The systemd-networkd-wait-online tool is now enabled by
11702 default. It will delay network-online.target until a network
11703 connection has been configured. The tool primarily integrates
11704 with networkd, but will also make a best effort to make sense
11705 of network configuration performed in some other way.
11707 * Two new service options StartupCPUShares= and
11708 StartupBlockIOWeight= have been added that work similarly to
11709 CPUShares= and BlockIOWeight= however only apply during
11710 system startup. This is useful to prioritize certain services
11711 differently during bootup than during normal runtime.
11713 * hostnamed has been changed to prefer the statically
11714 configured hostname in /etc/hostname (unless set to
11715 'localhost' or empty) over any dynamic one supplied by
11716 dhcp. With this change, the rules for picking the hostname
11717 match more closely the rules of other configuration settings
11718 where the local administrator's configuration in /etc always
11719 overrides any other settings.
11721 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
11722 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
11723 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
11724 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
11725 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
11726 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
11727 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
11728 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
11729 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
11730 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
11731 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
11732 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
11733 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
11734 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
11735 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
11736 Lindskog, WaLyong Cho, Will Woods, Zbigniew
11739 — Beijing, 2014-05-28
11743 * When restoring the screen brightness at boot, stay away from
11744 the darkest setting or from the lowest 5% of the available
11745 range, depending on which is the larger value of both. This
11746 should effectively protect the user from rebooting into a
11747 black screen, should the brightness have been set to minimum
11750 * sd-login gained a new sd_machine_get_class() call to
11751 determine the class ("vm" or "container") of a machine
11752 registered with machined.
11754 * sd-login gained new calls
11755 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
11756 to query the identity of the peer of a local AF_UNIX
11757 connection. They operate similarly to their sd_pid_get_xyz()
11760 * PID 1 will now maintain a system-wide system state engine
11761 with the states "starting", "running", "degraded",
11762 "maintenance", "stopping". These states are bound to system
11763 startup, normal runtime, runtime with at least one failed
11764 service, rescue/emergency mode and system shutdown. This
11765 state is shown in the "systemctl status" output when no unit
11766 name is passed. It is useful to determine system state, in
11767 particularly when doing so for many systems or containers at
11770 * A new command "list-machines" has been added to "systemctl"
11771 that lists all local OS containers and shows their system
11772 state (see above), if systemd runs inside of them.
11774 * systemctl gained a new "-r" switch to recursively enumerate
11775 units on all local containers, when used with the
11776 "list-unit" command (which is the default one that is
11777 executed when no parameters are specified).
11779 * The GPT automatic partition discovery logic will now honour
11780 two GPT partition flags: one may be set on a partition to
11781 cause it to be mounted read-only, and the other may be set
11782 on a partition to ignore it during automatic discovery.
11784 * Two new GPT type UUIDs have been added for automatic root
11785 partition discovery, for 32-bit and 64-bit ARM. This is not
11786 particularly useful for discovering the root directory on
11787 these architectures during bare-metal boots (since UEFI is
11788 not common there), but still very useful to allow booting of
11789 ARM disk images in nspawn with the -i option.
11791 * MAC addresses of interfaces created with nspawn's
11792 --network-interface= switch will now be generated from the
11793 machine name, and thus be stable between multiple invocations
11796 * logind will now automatically remove all IPC objects owned
11797 by a user if she or he fully logs out. This makes sure that
11798 users who are logged out cannot continue to consume IPC
11799 resources. This covers SysV memory, semaphores and message
11800 queues as well as POSIX shared memory and message
11801 queues. Traditionally, SysV and POSIX IPC had no lifecycle
11802 limits. With this functionality, that is corrected. This may
11803 be turned off by using the RemoveIPC= switch of logind.conf.
11805 * The systemd-machine-id-setup and tmpfiles tools gained a
11806 --root= switch to operate on a specific root directory,
11809 * journald can now forward logged messages to the TTYs of all
11810 logged in users ("wall"). This is the default for all
11811 emergency messages now.
11813 * A new tool systemd-journal-remote has been added to stream
11814 journal log messages across the network.
11816 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
11817 controller trees are mounted into it. Note that the
11818 directories mounted beneath it are not read-only. This is a
11819 security measure and is particularly useful because glibc
11820 actually includes a search logic to pick any tmpfs it can
11821 find to implement shm_open() if /dev/shm is not available
11822 (which it might very well be in namespaced setups).
11824 * machinectl gained a new "poweroff" command to cleanly power
11825 down a local OS container.
11827 * The PrivateDevices= unit file setting will now also drop the
11828 CAP_MKNOD capability from the capability bound set, and
11829 imply DevicePolicy=closed.
11831 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
11832 comprehensively on all long-running systemd services where
11833 this is appropriate.
11835 * systemd-udevd will now run in a disassociated mount
11836 namespace. To mount directories from udev rules, make sure to
11837 pull in mount units via SYSTEMD_WANTS properties.
11839 * The kdbus support gained support for uploading policy into
11840 the kernel. sd-bus gained support for creating "monitoring"
11841 connections that can eavesdrop into all bus communication
11842 for debugging purposes.
11844 * Timestamps may now be specified in seconds since the UNIX
11845 epoch Jan 1st, 1970 by specifying "@" followed by the value
11848 * Native tcpwrap support in systemd has been removed. tcpwrap
11849 is old code, not really maintained anymore and has serious
11850 shortcomings, and better options such as firewalls
11851 exist. For setups that require tcpwrap usage, please
11852 consider invoking your socket-activated service via tcpd,
11853 like on traditional inetd.
11855 * A new system.conf configuration option
11856 DefaultTimerAccuracySec= has been added that controls the
11857 default AccuracySec= setting of .timer units.
11859 * Timer units gained a new WakeSystem= switch. If enabled,
11860 timers configured this way will cause the system to resume
11861 from system suspend (if the system supports that, which most
11864 * Timer units gained a new Persistent= switch. If enabled,
11865 timers configured this way will save to disk when they have
11866 been last triggered. This information is then used on next
11867 reboot to possible execute overdue timer events, that
11868 could not take place because the system was powered off.
11869 This enables simple anacron-like behaviour for timer units.
11871 * systemctl's "list-timers" will now also list the time a
11872 timer unit was last triggered in addition to the next time
11873 it will be triggered.
11875 * systemd-networkd will now assign predictable IPv4LL
11876 addresses to its local interfaces.
11878 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
11879 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
11880 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
11881 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
11882 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
11883 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
11884 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
11885 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
11888 — Berlin, 2014-03-25
11892 * A new unit file setting RestrictAddressFamilies= has been
11893 added to restrict which socket address families unit
11894 processes gain access to. This takes address family names
11895 like "AF_INET" or "AF_UNIX", and is useful to minimize the
11896 attack surface of services via exotic protocol stacks. This
11897 is built on seccomp system call filters.
11899 * Two new unit file settings RuntimeDirectory= and
11900 RuntimeDirectoryMode= have been added that may be used to
11901 manage a per-daemon runtime directories below /run. This is
11902 an alternative for setting up directory permissions with
11903 tmpfiles snippets, and has the advantage that the runtime
11904 directory's lifetime is bound to the daemon runtime and that
11905 the daemon starts up with an empty directory each time. This
11906 is particularly useful when writing services that drop
11907 privileges using the User= or Group= setting.
11909 * The DeviceAllow= unit setting now supports globbing for
11910 matching against device group names.
11912 * The systemd configuration file system.conf gained new
11913 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
11914 DefaultMemoryAccounting= to globally turn on/off accounting
11915 for specific resources (cgroups) for all units. These
11916 settings may still be overridden individually in each unit
11919 * systemd-gpt-auto-generator is now able to discover /srv and
11920 root partitions in addition to /home and swap partitions. It
11921 also supports LUKS-encrypted partitions now. With this in
11922 place, automatic discovery of partitions to mount following
11923 the Discoverable Partitions Specification
11924 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
11925 is now a lot more complete. This allows booting without
11926 /etc/fstab and without root= on the kernel command line on
11927 systems prepared appropriately.
11929 * systemd-nspawn gained a new --image= switch which allows
11930 booting up disk images and Linux installations on any block
11931 device that follow the Discoverable Partitions Specification
11932 (see above). This means that installations made with
11933 appropriately updated installers may now be started and
11934 deployed using container managers, completely
11935 unmodified. (We hope that libvirt-lxc will add support for
11936 this feature soon, too.)
11938 * systemd-nspawn gained a new --network-macvlan= setting to
11939 set up a private macvlan interface for the
11940 container. Similarly, systemd-networkd gained a new
11941 Kind=macvlan setting in .netdev files.
11943 * systemd-networkd now supports configuring local addresses
11946 * A new tool systemd-network-wait-online has been added to
11947 synchronously wait for network connectivity using
11950 * The sd-bus.h bus API gained a new sd_bus_track object for
11951 tracking the lifecycle of bus peers. Note that sd-bus.h is
11952 still not a public API though (unless you specify
11953 --enable-kdbus on the configure command line, which however
11954 voids your warranty and you get no API stability guarantee).
11956 * The $XDG_RUNTIME_DIR runtime directories for each user are
11957 now individual tmpfs instances, which has the benefit of
11958 introducing separate pools for each user, with individual
11959 size limits, and thus making sure that unprivileged clients
11960 can no longer negatively impact the system or other users by
11961 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
11962 RuntimeDirectorySize= has been introduced that allows
11963 controlling the default size limit for all users. It
11964 defaults to 10% of the available physical memory. This is no
11965 replacement for quotas on tmpfs though (which the kernel
11966 still does not support), as /dev/shm and /tmp are still
11967 shared resources used by both the system and unprivileged
11970 * logind will now automatically turn off automatic suspending
11971 on laptop lid close when more than one display is
11972 connected. This was previously expected to be implemented
11973 individually in desktop environments (such as GNOME),
11974 however has been added to logind now, in order to fix a
11975 boot-time race where a desktop environment might not have
11976 been started yet and thus not been able to take an inhibitor
11977 lock at the time where logind already suspends the system
11978 due to a closed lid.
11980 * logind will now wait at least 30s after each system
11981 suspend/resume cycle, and 3min after system boot before
11982 suspending the system due to a closed laptop lid. This
11983 should give USB docking stations and similar enough time to
11984 be probed and configured after system resume and boot in
11985 order to then act as suspend blocker.
11987 * systemd-run gained a new --property= setting which allows
11988 initialization of resource control properties (and others)
11989 for the created scope or service unit. Example: "systemd-run
11990 --property=BlockIOWeight=10 updatedb" may be used to run
11991 updatedb at a low block IO scheduling weight.
11993 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
11994 now also work in --scope mode.
11996 * When systemd is compiled with kdbus support, basic support
11997 for enforced policies is now in place. (Note that enabling
11998 kdbus still voids your warranty and no API compatibility
11999 promises are made.)
12001 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
12002 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
12003 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
12004 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
12005 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
12006 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
12007 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
12008 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
12009 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
12010 Zbigniew Jędrzejewski-Szmek
12012 — Berlin, 2014-03-12
12016 * systemd will now relabel /dev after loading the SMACK policy
12017 according to SMACK rules.
12019 * A new unit file option AppArmorProfile= has been added to
12020 set the AppArmor profile for the processes of a unit.
12022 * A new condition check ConditionArchitecture= has been added
12023 to conditionalize units based on the system architecture, as
12024 reported by uname()'s "machine" field.
12026 * systemd-networkd now supports matching on the system
12027 virtualization, architecture, kernel command line, hostname
12030 * logind is now a lot more aggressive when suspending the
12031 machine due to a closed laptop lid. Instead of acting only
12032 on the lid close action, it will continuously watch the lid
12033 status and act on it. This is useful for laptops where the
12034 power button is on the outside of the chassis so that it can
12035 be reached without opening the lid (such as the Lenovo
12036 Yoga). On those machines, logind will now immediately
12037 re-suspend the machine if the power button has been
12038 accidentally pressed while the laptop was suspended and in a
12039 backpack or similar.
12041 * logind will now watch SW_DOCK switches and inhibit reaction
12042 to the lid switch if it is pressed. This means that logind
12043 will not suspend the machine anymore if the lid is closed
12044 and the system is docked, if the laptop supports SW_DOCK
12045 notifications via the input layer. Note that ACPI docking
12046 stations do not generate this currently. Also note that this
12047 logic is usually not fully sufficient and Desktop
12048 Environments should take a lid switch inhibitor lock when an
12049 external display is connected, as systemd will not watch
12052 * nspawn will now make use of the devices cgroup controller by
12053 default, and only permit creation of and access to the usual
12054 API device nodes like /dev/null or /dev/random, as well as
12055 access to (but not creation of) the pty devices.
12057 * We will now ship a default .network file for
12058 systemd-networkd that automatically configures DHCP for
12059 network interfaces created by nspawn's --network-veth or
12060 --network-bridge= switches.
12062 * systemd will now understand the usual M, K, G, T suffixes
12063 according to SI conventions (i.e. to the base 1000) when
12064 referring to throughput and hardware metrics. It will stay
12065 with IEC conventions (i.e. to the base 1024) for software
12066 metrics, according to what is customary according to
12067 Wikipedia. We explicitly document which base applies for
12068 each configuration option.
12070 * The DeviceAllow= setting in unit files now supports a syntax to
12071 allow-list an entire group of devices node majors at once, based on
12072 the /proc/devices listing. For example, with the string "char-pts",
12073 it is now possible to allow-list all current and future pseudo-TTYs
12076 * sd-event learned a new "post" event source. Event sources of
12077 this type are triggered by the dispatching of any event
12078 source of a type that is not "post". This is useful for
12079 implementing clean-up and check event sources that are
12080 triggered by other work being done in the program.
12082 * systemd-networkd is no longer statically enabled, but uses
12083 the usual [Install] sections so that it can be
12084 enabled/disabled using systemctl. It still is enabled by
12087 * When creating a veth interface pair with systemd-nspawn, the
12088 host side will now be prefixed with "vb-" if
12089 --network-bridge= is used, and with "ve-" if --network-veth
12090 is used. This way, it is easy to distinguish these cases on
12091 the host, for example to apply different configuration to
12092 them with systemd-networkd.
12094 * The compatibility libraries for libsystemd-journal.so,
12095 libsystem-id128.so, libsystemd-login.so and
12096 libsystemd-daemon.so do not make use of IFUNC
12097 anymore. Instead, we now build libsystemd.so multiple times
12098 under these alternative names. This means that the footprint
12099 is drastically increased, but given that these are
12100 transitional compatibility libraries, this should not matter
12101 much. This change has been made necessary to support the ARM
12102 platform for these compatibility libraries, as the ARM
12103 toolchain is not really at the same level as the toolchain
12104 for other architectures like x86 and does not support
12105 IFUNC. Please make sure to use --enable-compat-libs only
12106 during a transitional period!
12108 * The .include syntax has been deprecated and is not documented
12109 anymore. Drop-in files in .d directories should be used instead.
12111 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
12112 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
12113 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
12114 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
12115 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
12116 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
12117 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
12118 Zbigniew Jędrzejewski-Szmek
12120 — Berlin, 2014-02-24
12124 * A new component "systemd-networkd" has been added that can
12125 be used to configure local network interfaces statically or
12126 via DHCP. It is capable of bringing up bridges, VLANs, and
12127 bonding. Currently, no hook-ups for interactive network
12128 configuration are provided. Use this for your initrd,
12129 container, embedded, or server setup if you need a simple,
12130 yet powerful, network configuration solution. This
12131 configuration subsystem is quite nifty, as it allows wildcard
12132 hotplug matching in interfaces. For example, with a single
12133 configuration snippet, you can configure that all Ethernet
12134 interfaces showing up are automatically added to a bridge,
12135 or similar. It supports link-sensing and more.
12137 * A new tool "systemd-socket-proxyd" has been added which can
12138 act as a bidirectional proxy for TCP sockets. This is
12139 useful for adding socket activation support to services that
12140 do not actually support socket activation, including virtual
12141 machines and the like.
12143 * Add a new tool to save/restore rfkill state on
12146 * Save/restore state of keyboard backlights in addition to
12147 display backlights on shutdown/boot.
12149 * udev learned a new SECLABEL{} construct to label device
12150 nodes with a specific security label when they appear. For
12151 now, only SECLABEL{selinux} is supported, but the syntax is
12152 prepared for additional security frameworks.
12154 * udev gained a new scheme to configure link-level attributes
12155 from files in /etc/systemd/network/*.link. These files can
12156 match against MAC address, device path, driver name and type,
12157 and will apply attributes like the naming policy, link speed,
12158 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
12159 address assignment policy (randomized, …).
12161 * The configuration of network interface naming rules for
12162 "permanent interface names" has changed: a new NamePolicy=
12163 setting in the [Link] section of .link files determines the
12164 priority of possible naming schemes (onboard, slot, MAC,
12165 path). The default value of this setting is determined by
12166 /usr/lib/net/links/99-default.link. Old
12167 80-net-name-slot.rules udev configuration file has been
12168 removed, so local configuration overriding this file should
12169 be adapted to override 99-default.link instead.
12171 * When the User= switch is used in a unit file, also
12172 initialize $SHELL= based on the user database entry.
12174 * systemd no longer depends on libdbus. All communication is
12175 now done with sd-bus, systemd's low-level bus library
12178 * kdbus support has been added to PID 1 itself. When kdbus is
12179 enabled, this causes PID 1 to set up the system bus and
12180 enable support for a new ".busname" unit type that
12181 encapsulates bus name activation on kdbus. It works a little
12182 bit like ".socket" units, except for bus names. A new
12183 generator has been added that converts classic dbus1 service
12184 activation files automatically into native systemd .busname
12185 and .service units.
12187 * sd-bus: add a light-weight vtable implementation that allows
12188 defining objects on the bus with a simple static const
12189 vtable array of its methods, signals and properties.
12191 * systemd will not generate or install static dbus
12192 introspection data anymore to /usr/share/dbus-1/interfaces,
12193 as the precise format of these files is unclear, and
12194 nothing makes use of it.
12196 * A proxy daemon is now provided to proxy clients connecting
12197 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
12198 compatibility with classic D-Bus.
12200 * A bus driver implementation has been added that supports the
12201 classic D-Bus bus driver calls on kdbus, also for
12202 compatibility purposes.
12204 * A new API "sd-event.h" has been added that implements a
12205 minimal event loop API built around epoll. It provides a
12206 couple of features that direct epoll usage is lacking:
12207 prioritization of events, scales to large numbers of timer
12208 events, per-event timer slack (accuracy), system-wide
12209 coalescing of timer events, exit handlers, watchdog
12210 supervision support using systemd's sd_notify() API, child
12213 * A new API "sd-rntl.h" has been added that provides an API
12214 around the route netlink interface of the kernel, similar in
12215 style to "sd-bus.h".
12217 * A new API "sd-dhcp-client.h" has been added that provides a
12218 small DHCPv4 client-side implementation. This is used by
12219 "systemd-networkd".
12221 * There is a new kernel command line option
12222 "systemd.restore_state=0|1". When set to "0", none of the
12223 systemd tools will restore saved runtime state to hardware
12224 devices. More specifically, the rfkill and backlight states
12227 * The FsckPassNo= compatibility option in mount/service units
12228 has been removed. The fstab generator will now add the
12229 necessary dependencies automatically, and does not require
12230 PID1's support for that anymore.
12232 * journalctl gained a new switch, --list-boots, that lists
12233 recent boots with their times and boot IDs.
12235 * The various tools like systemctl, loginctl, timedatectl,
12236 busctl, systemd-run, … have gained a new switch "-M" to
12237 connect to a specific, local OS container (as direct
12238 connection, without requiring SSH). This works on any
12239 container that is registered with machined, such as those
12240 created by libvirt-lxc or nspawn.
12242 * systemd-run and systemd-analyze also gained support for "-H"
12243 to connect to remote hosts via SSH. This is particularly
12244 useful for systemd-run because it enables queuing of jobs
12245 onto remote systems.
12247 * machinectl gained a new command "login" to open a getty
12248 login in any local container. This works with any container
12249 that is registered with machined (such as those created by
12250 libvirt-lxc or nspawn), and which runs systemd inside.
12252 * machinectl gained a new "reboot" command that may be used to
12253 trigger a reboot on a specific container that is registered
12254 with machined. This works on any container that runs an init
12255 system of some kind.
12257 * systemctl gained a new "list-timers" command to print a nice
12258 listing of installed timer units with the times they elapse
12261 * Alternative reboot() parameters may now be specified on the
12262 "systemctl reboot" command line and are passed to the
12263 reboot() system call.
12265 * systemctl gained a new --job-mode= switch to configure the
12266 mode to queue a job with. This is a more generic version of
12267 --fail, --irreversible, and --ignore-dependencies, which are
12268 still available but not advertised anymore.
12270 * /etc/systemd/system.conf gained new settings to configure
12271 various default timeouts of units, as well as the default
12272 start limit interval and burst. These may still be overridden
12275 * PID1 will now export on the bus profile data of the security
12276 policy upload process (such as the SELinux policy upload to
12279 * journald: when forwarding logs to the console, include
12280 timestamps (following the setting in
12281 /sys/module/printk/parameters/time).
12283 * OnCalendar= in timer units now understands the special
12284 strings "yearly" and "annually". (Both are equivalent)
12286 * The accuracy of timer units is now configurable with the new
12287 AccuracySec= setting. It defaults to 1min.
12289 * A new dependency type JoinsNamespaceOf= has been added that
12290 allows running two services within the same /tmp and network
12291 namespace, if PrivateNetwork= or PrivateTmp= are used.
12293 * A new command "cat" has been added to systemctl. It outputs
12294 the original unit file of a unit, and concatenates the
12295 contents of additional "drop-in" unit file snippets, so that
12296 the full configuration is shown.
12298 * systemctl now supports globbing on the various "list-xyz"
12299 commands, like "list-units" or "list-sockets", as well as on
12300 those commands which take multiple unit names.
12302 * journalctl's --unit= switch gained support for globbing.
12304 * All systemd daemons now make use of the watchdog logic so
12305 that systemd automatically notices when they hang.
12307 * If the $container_ttys environment variable is set,
12308 getty-generator will automatically spawn a getty for each
12309 listed tty. This is useful for container managers to request
12310 login gettys to be spawned on as many ttys as needed.
12312 * %h, %s, %U specifier support is not available anymore when
12313 used in unit files for PID 1. This is because NSS calls are
12314 not safe from PID 1. They stay available for --user
12315 instances of systemd, and as special case for the root user.
12317 * loginctl gained a new "--no-legend" switch to turn off output
12318 of the legend text.
12320 * The "sd-login.h" API gained three new calls:
12321 sd_session_is_remote(), sd_session_get_remote_user(),
12322 sd_session_get_remote_host() to query information about
12325 * The udev hardware database now also carries vendor/product
12326 information of SDIO devices.
12328 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
12329 determine whether watchdog notifications are requested by
12330 the system manager.
12332 * Socket-activated per-connection services now include a
12333 short description of the connection parameters in the
12336 * tmpfiles gained a new "--boot" option. When this is not used,
12337 only lines where the command character is not suffixed with
12338 "!" are executed. When this option is specified, those
12339 options are executed too. This partitions tmpfiles
12340 directives into those that can be safely executed at any
12341 time, and those which should be run only at boot (for
12342 example, a line that creates /run/nologin).
12344 * A new API "sd-resolve.h" has been added which provides a simple
12345 asynchronous wrapper around glibc NSS hostname resolution
12346 calls, such as getaddrinfo(). In contrast to glibc's
12347 getaddrinfo_a(), it does not use signals. In contrast to most
12348 other asynchronous name resolution libraries, this one does
12349 not reimplement DNS, but reuses NSS, so that alternate
12350 hostname resolution systems continue to work, such as mDNS,
12351 LDAP, etc. This API is based on libasyncns, but it has been
12352 cleaned up for inclusion in systemd.
12354 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
12355 "sd-daemon.h" are no longer found in individual libraries
12356 libsystemd-journal.so, libsystemd-login.so,
12357 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
12358 merged them into a single library, libsystemd.so, which
12359 provides all symbols. The reason for this is cyclic
12360 dependencies, as these libraries tend to use each other's
12361 symbols. So far, we have managed to workaround that by linking
12362 a copy of a good part of our code into each of these
12363 libraries again and again, which, however, makes certain
12364 things hard to do, like sharing static variables. Also, it
12365 substantially increases footprint. With this change, there
12366 is only one library for the basic APIs systemd
12367 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
12368 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
12369 library as well, however are subject to the --enable-kdbus
12370 switch (see below). Note that "sd-dhcp-client.h" is not part
12371 of this library (this is because it only consumes, never
12372 provides, services of/to other APIs). To make the transition
12373 easy from the separate libraries to the unified one, we
12374 provide the --enable-compat-libs compile-time switch which
12375 will generate stub libraries that are compatible with the
12376 old ones but redirect all calls to the new one.
12378 * All of the kdbus logic and the new APIs "sd-bus.h",
12379 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
12380 and "sd-utf8.h" are compile-time optional via the
12381 "--enable-kdbus" switch, and they are not compiled in by
12382 default. To make use of kdbus, you have to explicitly enable
12383 the switch. Note however, that neither the kernel nor the
12384 userspace API for all of this is considered stable yet. We
12385 want to maintain the freedom to still change the APIs for
12386 now. By specifying this build-time switch, you acknowledge
12387 that you are aware of the instability of the current
12390 * Also, note that while kdbus is pretty much complete,
12391 it lacks one thing: proper policy support. This means you
12392 can build a fully working system with all features; however,
12393 it will be highly insecure. Policy support will be added in
12394 one of the next releases, at the same time that we will
12395 declare the APIs stable.
12397 * When the kernel command line argument "kdbus" is specified,
12398 systemd will automatically load the kdbus.ko kernel module. At
12399 this stage of development, it is only useful for testing kdbus
12400 and should not be used in production. Note: if "--enable-kdbus"
12401 is specified, and the kdbus.ko kernel module is available, and
12402 "kdbus" is added to the kernel command line, the entire system
12403 runs with kdbus instead of dbus-daemon, with the above mentioned
12404 problem of missing the system policy enforcement. Also a future
12405 version of kdbus.ko or a newer systemd will not be compatible with
12406 each other, and will unlikely be able to boot the machine if only
12407 one of them is updated.
12409 * systemctl gained a new "import-environment" command which
12410 uploads the caller's environment (or parts thereof) into the
12411 service manager so that it is inherited by services started
12412 by the manager. This is useful to upload variables like
12413 $DISPLAY into the user service manager.
12415 * A new PrivateDevices= switch has been added to service units
12416 which allows running a service with a namespaced /dev
12417 directory that does not contain any device nodes for
12418 physical devices. More specifically, it only includes devices
12419 such as /dev/null, /dev/urandom, and /dev/zero which are API
12422 * logind has been extended to support behaviour like VT
12423 switching on seats that do not support a VT. This makes
12424 multi-session available on seats that are not the first seat
12425 (seat0), and on systems where kernel support for VTs has
12426 been disabled at compile-time.
12428 * If a process holds a delay lock for system sleep or shutdown
12429 and fails to release it in time, we will now log its
12430 identity. This makes it easier to identify processes that
12431 cause slow suspends or power-offs.
12433 * When parsing /etc/crypttab, support for a new key-slot=
12434 option as supported by Debian is added. It allows indicating
12435 which LUKS slot to use on disk, speeding up key loading.
12437 * The sd_journal_sendv() API call has been checked and
12438 officially declared to be async-signal-safe so that it may
12439 be invoked from signal handlers for logging purposes.
12441 * Boot-time status output is now enabled automatically after a
12442 short timeout if boot does not progress, in order to give
12443 the user an indication what she or he is waiting for.
12445 * The boot-time output has been improved to show how much time
12446 remains until jobs expire.
12448 * The KillMode= switch in service units gained a new possible
12449 value "mixed". If set, and the unit is shut down, then the
12450 initial SIGTERM signal is sent only to the main daemon
12451 process, while the following SIGKILL signal is sent to
12452 all remaining processes of the service.
12454 * When a scope unit is registered, a new property "Controller"
12455 may be set. If set to a valid bus name, systemd will send a
12456 RequestStop() signal to this name when it would like to shut
12457 down the scope. This may be used to hook manager logic into
12458 the shutdown logic of scope units. Also, scope units may now
12459 be put in a special "abandoned" state, in which case the
12460 manager process which created them takes no further
12461 responsibilities for it.
12463 * When reading unit files, systemd will now verify
12464 the access mode of these files, and warn about certain
12465 suspicious combinations. This has been added to make it
12466 easier to track down packaging bugs where unit files are
12467 marked executable or world-writable.
12469 * systemd-nspawn gained a new "--setenv=" switch to set
12470 container-wide environment variables. The similar option in
12471 systemd-activate was renamed from "--environment=" to
12472 "--setenv=" for consistency.
12474 * systemd-nspawn has been updated to create a new kdbus domain
12475 for each container that is invoked, thus allowing each
12476 container to have its own set of system and user buses,
12477 independent of the host.
12479 * systemd-nspawn gained a new --drop-capability= switch to run
12480 the container with less capabilities than the default. Both
12481 --drop-capability= and --capability= now take the special
12482 string "all" for dropping or keeping all capabilities.
12484 * systemd-nspawn gained new switches for executing containers
12485 with specific SELinux labels set.
12487 * systemd-nspawn gained a new --quiet switch to not generate
12488 any additional output but the container's own console
12491 * systemd-nspawn gained a new --share-system switch to run a
12492 container without PID namespacing enabled.
12494 * systemd-nspawn gained a new --register= switch to control
12495 whether the container is registered with systemd-machined or
12496 not. This is useful for containers that do not run full
12497 OS images, but only specific apps.
12499 * systemd-nspawn gained a new --keep-unit which may be used
12500 when invoked as the only program from a service unit, and
12501 results in registration of the unit service itself in
12502 systemd-machined, instead of a newly opened scope unit.
12504 * systemd-nspawn gained a new --network-interface= switch for
12505 moving arbitrary interfaces to the container. The new
12506 --network-veth switch creates a virtual Ethernet connection
12507 between host and container. The new --network-bridge=
12508 switch then allows assigning the host side of this virtual
12509 Ethernet connection to a bridge device.
12511 * systemd-nspawn gained a new --personality= switch for
12512 setting the kernel personality for the container. This is
12513 useful when running a 32-bit container on a 64-bit host. A
12514 similar option Personality= is now also available for service
12517 * logind will now also track a "Desktop" identifier for each
12518 session which encodes the desktop environment of it. This is
12519 useful for desktop environments that want to identify
12520 multiple running sessions of itself easily.
12522 * A new SELinuxContext= setting for service units has been
12523 added that allows setting a specific SELinux execution
12524 context for a service.
12526 * Most systemd client tools will now honour $SYSTEMD_LESS for
12527 settings of the "less" pager. By default, these tools will
12528 override $LESS to allow certain operations to work, such as
12529 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
12530 influence this logic.
12532 * systemd's "seccomp" hook-up has been changed to make use of
12533 the libseccomp library instead of using its own
12534 implementation. This has benefits for portability among
12537 * For usage together with SystemCallFilter=, a new
12538 SystemCallErrorNumber= setting has been introduced that
12539 allows configuration of a system error number to be returned
12540 on filtered system calls, instead of immediately killing the
12541 process. Also, SystemCallArchitectures= has been added to
12542 limit access to system calls of a particular architecture
12543 (in order to turn off support for unused secondary
12544 architectures). There is also a global
12545 SystemCallArchitectures= setting in system.conf now to turn
12546 off support for non-native system calls system-wide.
12548 * systemd requires a kernel with a working name_to_handle_at(),
12549 please see the kernel config requirements in the README file.
12551 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
12552 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
12553 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
12554 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
12555 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
12556 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
12557 Elia Pinto, Florian Weimer, George McCollister, Goffredo
12558 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
12559 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
12560 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
12561 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
12562 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
12563 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
12564 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
12565 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
12566 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
12567 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
12568 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
12569 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
12570 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
12571 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
12572 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
12573 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
12574 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
12576 — Berlin, 2014-02-20
12580 * logind has gained support for facilitating privileged input
12581 and drm device access for unprivileged clients. This work is
12582 useful to allow Wayland display servers (and similar
12583 programs, such as kmscon) to run under the user's ID and
12584 access input and drm devices which are normally
12585 protected. When this is used (and the kernel is new enough)
12586 logind will "mute" IO on the file descriptors passed to
12587 Wayland as long as it is in the background and "unmute" it
12588 if it returns into the foreground. This allows secure
12589 session switching without allowing background sessions to
12590 eavesdrop on input and display data. This also introduces
12591 session switching support if VT support is turned off in the
12592 kernel, and on seats that are not seat0.
12594 * A new kernel command line option luks.options= is understood
12595 now which allows specifying LUKS options for usage for LUKS
12596 encrypted partitions specified with luks.uuid=.
12598 * tmpfiles.d(5) snippets may now use specifier expansion in
12599 path names. More specifically %m, %b, %H, %v, are now
12600 replaced by the local machine id, boot id, hostname, and
12601 kernel version number.
12603 * A new tmpfiles.d(5) command "m" has been introduced which
12604 may be used to change the owner/group/access mode of a file
12605 or directory if it exists, but do nothing if it does not.
12607 * This release removes high-level support for the
12608 MemorySoftLimit= cgroup setting. The underlying kernel
12609 cgroup attribute memory.soft_limit= is currently badly
12610 designed and likely to be removed from the kernel API in its
12611 current form, hence we should not expose it for now.
12613 * The memory.use_hierarchy cgroup attribute is now enabled for
12614 all cgroups systemd creates in the memory cgroup
12615 hierarchy. This option is likely to be come the built-in
12616 default in the kernel anyway, and the non-hierarchical mode
12617 never made much sense in the intrinsically hierarchical
12620 * A new field _SYSTEMD_SLICE= is logged along with all journal
12621 messages containing the slice a message was generated
12622 from. This is useful to allow easy per-customer filtering of
12623 logs among other things.
12625 * systemd-journald will no longer adjust the group of journal
12626 files it creates to the "systemd-journal" group. Instead we
12627 rely on the journal directory to be owned by the
12628 "systemd-journal" group, and its setgid bit set, so that the
12629 kernel file system layer will automatically enforce that
12630 journal files inherit this group assignment. The reason for
12631 this change is that we cannot allow NSS look-ups from
12632 journald which would be necessary to resolve
12633 "systemd-journal" to a numeric GID, because this might
12634 create deadlocks if NSS involves synchronous queries to
12635 other daemons (such as nscd, or sssd) which in turn are
12636 logging clients of journald and might block on it, which
12637 would then dead lock. A tmpfiles.d(5) snippet included in
12638 systemd will make sure the setgid bit and group are
12639 properly set on the journal directory if it exists on every
12640 boot. However, we recommend adjusting it manually after
12641 upgrades too (or from RPM scriptlets), so that the change is
12642 not delayed until next reboot.
12644 * Backlight and random seed files in /var/lib/ have moved into
12645 the /var/lib/systemd/ directory, in order to centralize all
12646 systemd generated files in one directory.
12648 * Boot time performance measurements (as displayed by
12649 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
12650 performance information if that's available to determine how
12651 much time BIOS and boot loader initialization required. With
12652 a sufficiently new BIOS you hence no longer need to boot
12653 with Gummiboot to get access to such information.
12655 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
12656 Cristian Rodríguez, Dave Reisner, David Herrmann, David
12657 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
12658 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
12659 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
12660 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
12661 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
12663 — Berlin, 2013-10-02
12667 * The Restart= option for services now understands a new
12668 on-watchdog setting, which will restart the service
12669 automatically if the service stops sending out watchdog keep
12670 alive messages (as configured with WatchdogSec=).
12672 * The getty generator (which is responsible for bringing up a
12673 getty on configured serial consoles) will no longer only
12674 start a getty on the primary kernel console but on all
12675 others, too. This makes the order in which console= is
12676 specified on the kernel command line less important.
12678 * libsystemd-logind gained a new sd_session_get_vt() call to
12679 retrieve the VT number of a session.
12681 * If the option "tries=0" is set for an entry of /etc/crypttab
12682 its passphrase is queried indefinitely instead of any
12683 maximum number of tries.
12685 * If a service with a configure PID file terminates its PID
12686 file will now be removed automatically if it still exists
12687 afterwards. This should put an end to stale PID files.
12689 * systemd-run will now also take relative binary path names
12690 for execution and no longer insists on absolute paths.
12692 * InaccessibleDirectories= and ReadOnlyDirectories= now take
12693 paths that are optionally prefixed with "-" to indicate that
12694 it should not be considered a failure if they do not exist.
12696 * journalctl -o (and similar commands) now understands a new
12697 output mode "short-precise", it is similar to "short" but
12698 shows timestamps with usec accuracy.
12700 * The option "discard" (as known from Debian) is now
12701 synonymous to "allow-discards" in /etc/crypttab. In fact,
12702 "discard" is preferred now (since it is easier to remember
12705 * Some licensing clean-ups were made, so that more code is now
12706 LGPL-2.1 licensed than before.
12708 * A minimal tool to save/restore the display backlight
12709 brightness across reboots has been added. It will store the
12710 backlight setting as late as possible at shutdown, and
12711 restore it as early as possible during reboot.
12713 * A logic to automatically discover and enable home and swap
12714 partitions on GPT disks has been added. With this in place
12715 /etc/fstab becomes optional for many setups as systemd can
12716 discover certain partitions located on the root disk
12717 automatically. Home partitions are recognized under their
12718 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
12719 partitions are recognized under their GPT type ID
12720 0657fd6da4ab43c484e50933c84b4f4f.
12722 * systemd will no longer pass any environment from the kernel
12723 or initrd to system services. If you want to set an
12724 environment for all services, do so via the kernel command
12725 line systemd.setenv= assignment.
12727 * The systemd-sysctl tool no longer natively reads the file
12728 /etc/sysctl.conf. If desired, the file should be symlinked
12729 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
12730 legacy support by a symlink rather than built-in code, it
12731 also makes the otherwise hidden order of application of the
12732 different files visible. (Note that this partly reverts to a
12733 pre-198 application order of sysctl knobs!)
12735 * The "systemctl set-log-level" and "systemctl dump" commands
12736 have been moved to systemd-analyze.
12738 * systemd-run learned the new --remain-after-exit switch,
12739 which causes the scope unit not to be cleaned up
12740 automatically after the process terminated.
12742 * tmpfiles learned a new --exclude-prefix= switch to exclude
12743 certain paths from operation.
12745 * journald will now automatically flush all messages to disk
12746 as soon as a message at the log level CRIT, ALERT or EMERG
12749 Contributions from: Andrew Cook, Brandon Philips, Christian
12750 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
12751 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
12752 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
12753 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
12754 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
12755 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
12756 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
12757 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
12758 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
12759 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
12760 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
12761 William Giokas, Zbigniew Jędrzejewski-Szmek
12763 — Berlin, 2013-09-13
12767 * The documentation has been updated to cover the various new
12768 concepts introduced with 205.
12770 * Unit files now understand the new %v specifier which
12771 resolves to the kernel version string as returned by "uname
12774 * systemctl now supports filtering the unit list output by
12775 load state, active state and sub state, using the new
12776 --state= parameter.
12778 * "systemctl status" will now show the results of the
12779 condition checks (like ConditionPathExists= and similar) of
12780 the last start attempts of the unit. They are also logged to
12783 * "journalctl -b" may now be used to look for boot output of a
12784 specific boot. Try "journalctl -b -1" for the previous boot,
12785 but the syntax is substantially more powerful.
12787 * "journalctl --show-cursor" has been added which prints the
12788 cursor string the last shown log line. This may then be used
12789 with the new "journalctl --after-cursor=" switch to continue
12790 browsing logs from that point on.
12792 * "journalctl --force" may now be used to force regeneration
12795 * Creation of "dead" device nodes has been moved from udev
12796 into kmod and tmpfiles. Previously, udev would read the kmod
12797 databases to pre-generate dead device nodes based on meta
12798 information contained in kernel modules, so that these would
12799 be auto-loaded on access rather then at boot. As this
12800 does not really have much to do with the exposing actual
12801 kernel devices to userspace this has always been slightly
12802 alien in the udev codebase. Following the new scheme kmod
12803 will now generate a runtime snippet for tmpfiles from the
12804 module meta information and it now is tmpfiles' job to the
12805 create the nodes. This also allows overriding access and
12806 other parameters for the nodes using the usual tmpfiles
12807 facilities. As side effect this allows us to remove the
12808 CAP_SYS_MKNOD capability bit from udevd entirely.
12810 * logind's device ACLs may now be applied to these "dead"
12811 devices nodes too, thus finally allowing managed access to
12812 devices such as /dev/snd/sequencer without loading the
12813 backing module right-away.
12815 * A new RPM macro has been added that may be used to apply
12816 tmpfiles configuration during package installation.
12818 * systemd-detect-virt and ConditionVirtualization= now can
12819 detect User-Mode-Linux machines (UML).
12821 * journald will now implicitly log the effective capabilities
12822 set of processes in the message metadata.
12824 * systemd-cryptsetup has gained support for TrueCrypt volumes.
12826 * The initrd interface has been simplified (more specifically,
12827 support for passing performance data via environment
12828 variables and fsck results via files in /run has been
12829 removed). These features were non-essential, and are
12830 nowadays available in a much nicer way by having systemd in
12831 the initrd serialize its state and have the hosts systemd
12832 deserialize it again.
12834 * The udev "keymap" data files and tools to apply keyboard
12835 specific mappings of scan to key codes, and force-release
12836 scan code lists have been entirely replaced by a udev
12837 "keyboard" builtin and a hwdb data file.
12839 * systemd will now honour the kernel's "quiet" command line
12840 argument also during late shutdown, resulting in a
12841 completely silent shutdown when used.
12843 * There's now an option to control the SO_REUSEPORT socket
12844 option in .socket units.
12846 * Instance units will now automatically get a per-template
12847 subslice of system.slice unless something else is explicitly
12848 configured. For example, instances of sshd@.service will now
12849 implicitly be placed in system-sshd.slice rather than
12850 system.slice as before.
12852 * Test coverage support may now be enabled at build time.
12854 Contributions from: Dave Reisner, Frederic Crozat, Harald
12855 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
12856 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
12857 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
12858 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
12859 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
12860 Giokas, Zbigniew Jędrzejewski-Szmek
12862 — Berlin, 2013-07-23
12866 * Two new unit types have been introduced:
12868 Scope units are very similar to service units, however, are
12869 created out of pre-existing processes — instead of PID 1
12870 forking off the processes. By using scope units it is
12871 possible for system services and applications to group their
12872 own child processes (worker processes) in a powerful way
12873 which then maybe used to organize them, or kill them
12874 together, or apply resource limits on them.
12876 Slice units may be used to partition system resources in an
12877 hierarchical fashion and then assign other units to them. By
12878 default there are now three slices: system.slice (for all
12879 system services), user.slice (for all user sessions),
12880 machine.slice (for VMs and containers).
12882 Slices and scopes have been introduced primarily in
12883 context of the work to move cgroup handling to a
12884 single-writer scheme, where only PID 1
12885 creates/removes/manages cgroups.
12887 * There's a new concept of "transient" units. In contrast to
12888 normal units these units are created via an API at runtime,
12889 not from configuration from disk. More specifically this
12890 means it is now possible to run arbitrary programs as
12891 independent services, with all execution parameters passed
12892 in via bus APIs rather than read from disk. Transient units
12893 make systemd substantially more dynamic then it ever was,
12894 and useful as a general batch manager.
12896 * logind has been updated to make use of scope and slice units
12897 for managing user sessions. As a user logs in he will get
12898 his own private slice unit, to which all sessions are added
12899 as scope units. We also added support for automatically
12900 adding an instance of user@.service for the user into the
12901 slice. Effectively logind will no longer create cgroup
12902 hierarchies on its own now, it will defer entirely to PID 1
12903 for this by means of scope, service and slice units. Since
12904 user sessions this way become entities managed by PID 1
12905 the output of "systemctl" is now a lot more comprehensive.
12907 * A new mini-daemon "systemd-machined" has been added which
12908 may be used by virtualization managers to register local
12909 VMs/containers. nspawn has been updated accordingly, and
12910 libvirt will be updated shortly. machined will collect a bit
12911 of meta information about the VMs/containers, and assign
12912 them their own scope unit (see above). The collected
12913 meta-data is then made available via the "machinectl" tool,
12914 and exposed in "ps" and similar tools. machined/machinectl
12915 is compile-time optional.
12917 * As discussed earlier, the low-level cgroup configuration
12918 options ControlGroup=, ControlGroupModify=,
12919 ControlGroupPersistent=, ControlGroupAttribute= have been
12920 removed. Please use high-level attribute settings instead as
12921 well as slice units.
12923 * A new bus call SetUnitProperties() has been added to alter
12924 various runtime parameters of a unit. This is primarily
12925 useful to alter cgroup parameters dynamically in a nice way,
12926 but will be extended later on to make more properties
12927 modifiable at runtime. systemctl gained a new set-properties
12928 command that wraps this call.
12930 * A new tool "systemd-run" has been added which can be used to
12931 run arbitrary command lines as transient services or scopes,
12932 while configuring a number of settings via the command
12933 line. This tool is currently very basic, however already
12934 very useful. We plan to extend this tool to even allow
12935 queuing of execution jobs with time triggers from the
12936 command line, similar in fashion to "at".
12938 * nspawn will now inform the user explicitly that kernels with
12939 audit enabled break containers, and suggest the user to turn
12942 * Support for detecting the IMA and AppArmor security
12943 frameworks with ConditionSecurity= has been added.
12945 * journalctl gained a new "-k" switch for showing only kernel
12946 messages, mimicking dmesg output; in addition to "--user"
12947 and "--system" switches for showing only user's own logs
12950 * systemd-delta can now show information about drop-in
12951 snippets extending unit files.
12953 * libsystemd-bus has been substantially updated but is still
12954 not available as public API.
12956 * systemd will now look for the "debug" argument on the kernel
12957 command line and enable debug logging, similar to what
12958 "systemd.log_level=debug" already did before.
12960 * "systemctl set-default", "systemctl get-default" has been
12961 added to configure the default.target symlink, which
12962 controls what to boot into by default.
12964 * "systemctl set-log-level" has been added as a convenient
12965 way to raise and lower systemd logging threshold.
12967 * "systemd-analyze plot" will now show the time the various
12968 generators needed for execution, as well as information
12969 about the unit file loading.
12971 * libsystemd-journal gained a new sd_journal_open_files() call
12972 for opening specific journal files. journactl also gained a
12973 new switch to expose this new functionality. Previously we
12974 only supported opening all files from a directory, or all
12975 files from the system, as opening individual files only is
12976 racy due to journal file rotation.
12978 * systemd gained the new DefaultEnvironment= setting in
12979 /etc/systemd/system.conf to set environment variables for
12982 * If a privileged process logs a journal message with the
12983 OBJECT_PID= field set, then journald will automatically
12984 augment this with additional OBJECT_UID=, OBJECT_GID=,
12985 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
12986 system services want to log events about specific client
12987 processes. journactl/systemctl has been updated to make use
12988 of this information if all log messages regarding a specific
12991 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
12992 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
12993 Reisner, David Coppa, David King, David Strauss, Eelco
12994 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
12995 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
12996 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
12997 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
12998 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
12999 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
13000 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
13001 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
13002 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
13003 Łukasz Stelmach, 장동준
13007 * The Python bindings gained some minimal support for the APIs
13008 exposed by libsystemd-logind.
13010 * ConditionSecurity= gained support for detecting SMACK. Since
13011 this condition already supports SELinux and AppArmor we only
13012 miss IMA for this. Patches welcome!
13014 Contributions from: Karol Lewandowski, Lennart Poettering,
13015 Zbigniew Jędrzejewski-Szmek
13019 * systemd-nspawn will now create /etc/resolv.conf if
13020 necessary, before bind-mounting the host's file onto it.
13022 * systemd-nspawn will now store meta information about a
13023 container on the container's cgroup as extended attribute
13024 fields, including the root directory.
13026 * The cgroup hierarchy has been reworked in many ways. All
13027 objects any of the components systemd creates in the cgroup
13028 tree are now suffixed. More specifically, user sessions are
13029 now placed in cgroups suffixed with ".session", users in
13030 cgroups suffixed with ".user", and nspawn containers in
13031 cgroups suffixed with ".nspawn". Furthermore, all cgroup
13032 names are now escaped in a simple scheme to avoid collision
13033 of userspace object names with kernel filenames. This work
13034 is preparation for making these objects relocatable in the
13035 cgroup tree, in order to allow easy resource partitioning of
13036 these objects without causing naming conflicts.
13038 * systemctl list-dependencies gained the new switches
13039 --plain, --reverse, --after and --before.
13041 * systemd-inhibit now shows the process name of processes that
13042 have taken an inhibitor lock.
13044 * nss-myhostname will now also resolve "localhost"
13045 implicitly. This makes /etc/hosts an optional file and
13046 nicely handles that on IPv6 ::1 maps to both "localhost" and
13047 the local hostname.
13049 * libsystemd-logind.so gained a new call
13050 sd_get_machine_names() to enumerate running containers and
13051 VMs (currently only supported by very new libvirt and
13052 nspawn). sd_login_monitor can now be used to watch
13053 VMs/containers coming and going.
13055 * .include is not allowed recursively anymore, and only in
13056 unit files. Usually it is better to use drop-in snippets in
13057 .d/*.conf anyway, as introduced with systemd 198.
13059 * systemd-analyze gained a new "critical-chain" command that
13060 determines the slowest chain of units run during system
13061 boot-up. It is very useful for tracking down where
13062 optimizing boot time is the most beneficial.
13064 * systemd will no longer allow manipulating service paths in
13065 the name=systemd:/system cgroup tree using ControlGroup= in
13066 units. (But is still fine with it in all other dirs.)
13068 * There's a new systemd-nspawn@.service service file that may
13069 be used to easily run nspawn containers as system
13070 services. With the container's root directory in
13071 /var/lib/container/foobar it is now sufficient to run
13072 "systemctl start systemd-nspawn@foobar.service" to boot it.
13074 * systemd-cgls gained a new parameter "--machine" to list only
13075 the processes within a certain container.
13077 * ConditionSecurity= now can check for "apparmor". We still
13078 are lacking checks for SMACK and IMA for this condition
13079 check though. Patches welcome!
13081 * A new configuration file /etc/systemd/sleep.conf has been
13082 added that may be used to configure which kernel operation
13083 systemd is supposed to execute when "suspend", "hibernate"
13084 or "hybrid-sleep" is requested. This makes the new kernel
13085 "freeze" state accessible to the user.
13087 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
13088 the passed argument if applicable.
13090 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
13091 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
13092 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
13093 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
13094 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
13095 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
13096 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
13101 * The output of 'systemctl list-jobs' got some polishing. The
13102 '--type=' argument may now be passed more than once. A new
13103 command 'systemctl list-sockets' has been added which shows
13104 a list of kernel sockets systemd is listening on with the
13105 socket units they belong to, plus the units these socket
13108 * The experimental libsystemd-bus library got substantial
13109 updates to work in conjunction with the (also experimental)
13110 kdbus kernel project. It works well enough to exchange
13111 messages with some sophistication. Note that kdbus is not
13112 ready yet, and the library is mostly an elaborate test case
13113 for now, and not installable.
13115 * systemd gained a new unit 'systemd-static-nodes.service'
13116 that generates static device nodes earlier during boot, and
13117 can run in conjunction with udev.
13119 * libsystemd-login gained a new call sd_pid_get_user_unit()
13120 to retrieve the user systemd unit a process is running
13121 in. This is useful for systems where systemd is used as
13124 * systemd-nspawn now places all containers in the new /machine
13125 top-level cgroup directory in the name=systemd
13126 hierarchy. libvirt will soon do the same, so that we get a
13127 uniform separation of /system, /user and /machine for system
13128 services, user processes and containers/virtual
13129 machines. This new cgroup hierarchy is also useful to stick
13130 stable names to specific container instances, which can be
13131 recognized later this way (this name may be controlled
13132 via systemd-nspawn's new -M switch). libsystemd-login also
13133 gained a new call sd_pid_get_machine_name() to retrieve the
13134 name of the container/VM a specific process belongs to.
13136 * bootchart can now store its data in the journal.
13138 * libsystemd-journal gained a new call
13139 sd_journal_add_conjunction() for AND expressions to the
13140 matching logic. This can be used to express more complex
13141 logical expressions.
13143 * journactl can now take multiple --unit= and --user-unit=
13146 * The cryptsetup logic now understands the "luks.key=" kernel
13147 command line switch for specifying a file to read the
13148 decryption key from. Also, if a configured key file is not
13149 found the tool will now automatically fall back to prompting
13152 * Python systemd.journal module was updated to wrap recently
13153 added functions from libsystemd-journal. The interface was
13154 changed to bring the low level interface in s.j._Reader
13155 closer to the C API, and the high level interface in
13156 s.j.Reader was updated to wrap and convert all data about
13159 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
13160 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
13161 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
13162 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
13163 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
13164 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
13168 * journalctl --update-catalog now understands a new --root=
13169 option to operate on catalogs found in a different root
13172 * During shutdown after systemd has terminated all running
13173 services a final killing loop kills all remaining left-over
13174 processes. We will now print the name of these processes
13175 when we send SIGKILL to them, since this usually indicates a
13178 * If /etc/crypttab refers to password files stored on
13179 configured mount points automatic dependencies will now be
13180 generated to ensure the specific mount is established first
13181 before the key file is attempted to be read.
13183 * 'systemctl status' will now show information about the
13184 network sockets a socket unit is listening on.
13186 * 'systemctl status' will also shown information about any
13187 drop-in configuration file for units. (Drop-In configuration
13188 files in this context are files such as
13189 /etc/systemd/system/foobar.service.d/*.conf)
13191 * systemd-cgtop now optionally shows summed up CPU times of
13192 cgroups. Press '%' while running cgtop to switch between
13193 percentage and absolute mode. This is useful to determine
13194 which cgroups use up the most CPU time over the entire
13195 runtime of the system. systemd-cgtop has also been updated
13196 to be 'pipeable' for processing with further shell tools.
13198 * 'hostnamectl set-hostname' will now allow setting of FQDN
13201 * The formatting and parsing of time span values has been
13202 changed. The parser now understands fractional expressions
13203 such as "5.5h". The formatter will now output fractional
13204 expressions for all time spans under 1min, i.e. "5.123456s"
13205 rather than "5s 123ms 456us". For time spans under 1s
13206 millisecond values are shown, for those under 1ms
13207 microsecond values are shown. This should greatly improve
13208 all time-related output of systemd.
13210 * libsystemd-login and libsystemd-journal gained new
13211 functions for querying the poll() events mask and poll()
13212 timeout value for integration into arbitrary event
13215 * localectl gained the ability to list available X11 keymaps
13216 (models, layouts, variants, options).
13218 * 'systemd-analyze dot' gained the ability to filter for
13219 specific units via shell-style globs, to create smaller,
13220 more useful graphs. I.e. it is now possible to create simple
13221 graphs of all the dependencies between only target units, or
13222 of all units that Avahi has dependencies with.
13224 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
13225 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
13226 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
13227 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
13228 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
13229 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
13230 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
13234 * The boot-time readahead implementation for rotating media
13235 will now read the read-ahead data in multiple passes which
13236 consist of all read requests made in equidistant time
13237 intervals. This means instead of strictly reading read-ahead
13238 data in its physical order on disk we now try to find a
13239 middle ground between physical and access time order.
13241 * /etc/os-release files gained a new BUILD_ID= field for usage
13242 on operating systems that provide continuous builds of OS
13245 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
13246 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
13247 William Douglas, Zbigniew Jędrzejewski-Szmek
13251 * systemd-python gained an API exposing libsystemd-daemon.
13253 * The SMACK setup logic gained support for uploading CIPSO
13256 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
13257 ReadOnlyDirectories= and InaccessibleDirectories= has
13258 changed. The private /tmp and /var/tmp directories are now
13259 shared by all processes of a service (which means
13260 ExecStartPre= may now leave data in /tmp that ExecStart= of
13261 the same service can still access). When a service is
13262 stopped its temporary directories are immediately deleted
13263 (normal clean-up with tmpfiles is still done in addition to
13266 * By default, systemd will now set a couple of sysctl
13267 variables in the kernel: the safe sysrq options are turned
13268 on, IP route verification is turned on, and source routing
13269 disabled. The recently added hardlink and softlink
13270 protection of the kernel is turned on. These settings should
13271 be reasonably safe, and good defaults for all new systems.
13273 * The predictable network naming logic may now be turned off
13274 with a new kernel command line switch: net.ifnames=0.
13276 * A new libsystemd-bus module has been added that implements a
13277 pretty complete D-Bus client library. For details see:
13279 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
13281 * journald will now explicitly flush the journal files to disk
13282 at the latest 5min after each write. The file will then also
13283 be marked offline until the next write. This should increase
13284 reliability in case of a crash. The synchronization delay
13285 can be configured via SyncIntervalSec= in journald.conf.
13287 * There's a new remote-fs-setup.target unit that can be used
13288 to pull in specific services when at least one remote file
13289 system is to be mounted.
13291 * There are new targets timers.target and paths.target as
13292 canonical targets to pull user timer and path units in
13293 from. This complements sockets.target with a similar
13294 purpose for socket units.
13296 * libudev gained a new call udev_device_set_attribute_value()
13297 to set sysfs attributes of a device.
13299 * The udev daemon now sets the default number of worker
13300 processes executed in parallel based on the number of available
13301 CPUs instead of the amount of available RAM. This is supposed
13302 to provide a more reliable default and limit a too aggressive
13303 parallelism for setups with 1000s of devices connected.
13305 Contributions from: Auke Kok, Colin Walters, Cristian
13306 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
13307 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
13308 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
13309 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
13310 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
13311 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
13312 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
13313 Zbigniew Jędrzejewski-Szmek
13317 * Configuration of unit files may now be extended via drop-in
13318 files without having to edit/override the unit files
13319 themselves. More specifically, if the administrator wants to
13320 change one value for a service file foobar.service he can
13321 now do so by dropping in a configuration snippet into
13322 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
13323 will load all these snippets and apply them on top of the
13324 main unit configuration file, possibly extending or
13325 overriding its settings. Using these drop-in snippets is
13326 generally nicer than the two earlier options for changing
13327 unit files locally: copying the files from
13328 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
13329 them there; or creating a new file in /etc/systemd/system/
13330 that incorporates the original one via ".include". Drop-in
13331 snippets into these .d/ directories can be placed in any
13332 directory systemd looks for units in, and the usual
13333 overriding semantics between /usr/lib, /etc and /run apply
13336 * Most unit file settings which take lists of items can now be
13337 reset by assigning the empty string to them. For example,
13338 normally, settings such as Environment=FOO=BAR append a new
13339 environment variable assignment to the environment block,
13340 each time they are used. By assigning Environment= the empty
13341 string the environment block can be reset to empty. This is
13342 particularly useful with the .d/*.conf drop-in snippets
13343 mentioned above, since this adds the ability to reset list
13344 settings from vendor unit files via these drop-ins.
13346 * systemctl gained a new "list-dependencies" command for
13347 listing the dependencies of a unit recursively.
13349 * Inhibitors are now honored and listed by "systemctl
13350 suspend", "systemctl poweroff" (and similar) too, not only
13351 GNOME. These commands will also list active sessions by
13354 * Resource limits (as exposed by the various control group
13355 controllers) can now be controlled dynamically at runtime
13356 for all units. More specifically, you can now use a command
13357 like "systemctl set-cgroup-attr foobar.service cpu.shares
13358 2000" to alter the CPU shares a specific service gets. These
13359 settings are stored persistently on disk, and thus allow the
13360 administrator to easily adjust the resource usage of
13361 services with a few simple commands. This dynamic resource
13362 management logic is also available to other programs via the
13363 bus. Almost any kernel cgroup attribute and controller is
13366 * systemd-vconsole-setup will now copy all font settings to
13367 all allocated VTs, where it previously applied them only to
13370 * libsystemd-login gained the new sd_session_get_tty() API
13373 * This release drops support for a few legacy or
13374 distribution-specific LSB facility names when parsing init
13375 scripts: $x-display-manager, $mail-transfer-agent,
13376 $mail-transport-agent, $mail-transfer-agent, $smtp,
13377 $null. Also, the mail-transfer-agent.target unit backing
13378 this has been removed. Distributions which want to retain
13379 compatibility with this should carry the burden for
13380 supporting this themselves and patch support for these back
13381 in, if they really need to. Also, the facilities $syslog and
13382 $local_fs are now ignored, since systemd does not support
13383 early-boot LSB init scripts anymore, and these facilities
13384 are implied anyway for normal services. syslog.target has
13387 * There are new bus calls on PID1's Manager object for
13388 cancelling jobs, and removing snapshot units. Previously,
13389 both calls were only available on the Job and Snapshot
13390 objects themselves.
13392 * systemd-journal-gatewayd gained SSL support.
13394 * The various "environment" files, such as /etc/locale.conf
13395 now support continuation lines with a backslash ("\") as
13396 last character in the line, similarly in style (but different)
13397 to how this is supported in shells.
13399 * For normal user processes the _SYSTEMD_USER_UNIT= field is
13400 now implicitly appended to every log entry logged. systemctl
13401 has been updated to filter by this field when operating on a
13402 user systemd instance.
13404 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
13405 CAP_AUDIT_CONTROL capabilities to the capabilities set for
13406 the container. This makes it easier to boot unmodified
13407 Fedora systems in a container, which however still requires
13408 audit=0 to be passed on the kernel command line. Auditing in
13409 kernel and userspace is unfortunately still too broken in
13410 context of containers, hence we recommend compiling it out
13411 of the kernel or using audit=0. Hopefully this will be fixed
13412 one day for good in the kernel.
13414 * nspawn gained the new --bind= and --bind-ro= parameters to
13415 bind mount specific directories from the host into the
13418 * nspawn will now mount its own devpts file system instance
13419 into the container, in order not to leak pty devices from
13420 the host into the container.
13422 * systemd will now read the firmware boot time performance
13423 information from the EFI variables, if the used boot loader
13424 supports this, and takes it into account for boot performance
13425 analysis via "systemd-analyze". This is currently supported
13426 only in conjunction with Gummiboot, but could be supported
13427 by other boot loaders too. For details see:
13429 https://systemd.io/BOOT_LOADER_INTERFACE
13431 * A new generator has been added that automatically mounts the
13432 EFI System Partition (ESP) to /boot, if that directory
13433 exists, is empty, and no other file system has been
13434 configured to be mounted there.
13436 * logind will now send out PrepareForSleep(false) out
13437 unconditionally, after coming back from suspend. This may be
13438 used by applications as asynchronous notification for
13439 system resume events.
13441 * "systemctl unlock-sessions" has been added, that allows
13442 unlocking the screens of all user sessions at once, similar
13443 to how "systemctl lock-sessions" already locked all users
13444 sessions. This is backed by a new D-Bus call UnlockSessions().
13446 * "loginctl seat-status" will now show the master device of a
13447 seat. (i.e. the device of a seat that needs to be around for
13448 the seat to be considered available, usually the graphics
13451 * tmpfiles gained a new "X" line type, that allows
13452 configuration of files and directories (with wildcards) that
13453 shall be excluded from automatic cleanup ("aging").
13455 * udev default rules set the device node permissions now only
13456 at "add" events, and do not change them any longer with a
13457 later "change" event.
13459 * The log messages for lid events and power/sleep keypresses
13460 now carry a message ID.
13462 * We now have a substantially larger unit test suite, but this
13463 continues to be work in progress.
13465 * udevadm hwdb gained a new --root= parameter to change the
13466 root directory to operate relative to.
13468 * logind will now issue a background sync() request to the kernel
13469 early at shutdown, so that dirty buffers are flushed to disk early
13470 instead of at the last moment, in order to optimize shutdown
13473 * A new bootctl tool has been added that is an interface for
13474 certain boot loader operations. This is currently a preview
13475 and is likely to be extended into a small mechanism daemon
13476 like timedated, localed, hostnamed, and can be used by
13477 graphical UIs to enumerate available boot options, and
13478 request boot into firmware operations.
13480 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
13481 the rest of the package. It also has been updated to work
13482 correctly in initrds.
13484 * polkit previously has been runtime optional, and is now also
13485 compile time optional via a configure switch.
13487 * systemd-analyze has been reimplemented in C. Also "systemctl
13488 dot" has moved into systemd-analyze.
13490 * "systemctl status" with no further parameters will now print
13491 the status of all active or failed units.
13493 * Operations such as "systemctl start" can now be executed
13494 with a new mode "--irreversible" which may be used to queue
13495 operations that cannot accidentally be reversed by a later
13496 job queuing. This is by default used to make shutdown
13497 requests more robust.
13499 * The Python API of systemd now gained a new module for
13500 reading journal files.
13502 * A new tool kernel-install has been added that can install
13503 kernel images according to the Boot Loader Specification:
13505 https://systemd.io/BOOT_LOADER_SPECIFICATION
13507 * Boot time console output has been improved to provide
13508 animated boot time output for hanging jobs.
13510 * A new tool systemd-activate has been added which can be used
13511 to test socket activation with, directly from the command
13512 line. This should make it much easier to test and debug
13513 socket activation in daemons.
13515 * journalctl gained a new "--reverse" (or -r) option to show
13516 journal output in reverse order (i.e. newest line first).
13518 * journalctl gained a new "--pager-end" (or -e) option to jump
13519 to immediately jump to the end of the journal in the
13520 pager. This is only supported in conjunction with "less".
13522 * journalctl gained a new "--user-unit=" option, that works
13523 similarly to "--unit=" but filters for user units rather than
13526 * A number of unit files to ease adoption of systemd in
13527 initrds has been added. This moves some minimal logic from
13528 the various initrd implementations into systemd proper.
13530 * The journal files are now owned by a new group
13531 "systemd-journal", which exists specifically to allow access
13532 to the journal, and nothing else. Previously, we used the
13533 "adm" group for that, which however possibly covers more
13534 than just journal/log file access. This new group is now
13535 already used by systemd-journal-gatewayd to ensure this
13536 daemon gets access to the journal files and as little else
13537 as possible. Note that "make install" will also set FS ACLs
13538 up for /var/log/journal to give "adm" and "wheel" read
13539 access to it, in addition to "systemd-journal" which owns
13540 the journal files. We recommend that packaging scripts also
13541 add read access to "adm" + "wheel" to /var/log/journal, and
13542 all existing/future journal files. To normal users and
13543 administrators little changes, however packagers need to
13544 ensure to create the "systemd-journal" system group at
13545 package installation time.
13547 * The systemd-journal-gatewayd now runs as unprivileged user
13548 systemd-journal-gateway:systemd-journal-gateway. Packaging
13549 scripts need to create these system user/group at
13552 * timedated now exposes a new boolean property CanNTP that
13553 indicates whether a local NTP service is available or not.
13555 * systemd-detect-virt will now also detect xen PVs
13557 * The pstore file system is now mounted by default, if it is
13560 * In addition to the SELinux and IMA policies we will now also
13561 load SMACK policies at early boot.
13563 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
13564 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
13565 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
13566 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
13567 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
13568 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
13569 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
13570 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
13571 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
13572 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
13573 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
13574 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
13575 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
13576 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
13580 * Timer units now support calendar time events in addition to
13581 monotonic time events. That means you can now trigger a unit
13582 based on a calendar time specification such as "Thu,Fri
13583 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
13584 or fifth day of any month of the year 2013, given that it is
13585 a Thursday or a Friday. This brings timer event support
13586 considerably closer to cron's capabilities. For details on
13587 the supported calendar time specification language see
13590 * udev now supports a number of different naming policies for
13591 network interfaces for predictable names, and a combination
13592 of these policies is now the default. Please see this wiki
13593 document for details:
13595 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
13597 * Auke Kok's bootchart implementation has been added to the
13598 systemd tree. It is an optional component that can graph the
13599 boot in quite some detail. It is one of the best bootchart
13600 implementations around and minimal in its code and
13603 * nss-myhostname has been integrated into the systemd source
13604 tree. nss-myhostname guarantees that the local hostname
13605 always stays resolvable via NSS. It has been a weak
13606 requirement of systemd-hostnamed since a long time, and
13607 since its code is actually trivial we decided to just
13608 include it in systemd's source tree. It can be turned off
13609 with a configure switch.
13611 * The read-ahead logic is now capable of properly detecting
13612 whether a btrfs file system is on SSD or rotating media, in
13613 order to optimize the read-ahead scheme. Previously, it was
13614 only capable of detecting this on traditional file systems
13617 * In udev, additional device properties are now read from the
13618 IAB in addition to the OUI database. Also, Bluetooth company
13619 identities are attached to the devices as well.
13621 * In service files %U may be used as specifier that is
13622 replaced by the configured user name of the service.
13624 * nspawn may now be invoked without a controlling TTY. This
13625 makes it suitable for invocation as its own service. This
13626 may be used to set up a simple containerized server system
13627 using only core OS tools.
13629 * systemd and nspawn can now accept socket file descriptors
13630 when they are started for socket activation. This enables
13631 implementation of socket activated nspawn
13632 containers. i.e. think about autospawning an entire OS image
13633 when the first SSH or HTTP connection is received. We expect
13634 that similar functionality will also be added to libvirt-lxc
13637 * journalctl will now suppress ANSI color codes when
13638 presenting log data.
13640 * systemctl will no longer show control group information for
13641 a unit if the control group is empty anyway.
13643 * logind can now automatically suspend/hibernate/shutdown the
13646 * /etc/machine-info and hostnamed now also expose the chassis
13647 type of the system. This can be used to determine whether
13648 the local system is a laptop, desktop, handset or
13649 tablet. This information may either be configured by the
13650 user/vendor or is automatically determined from ACPI and DMI
13651 information if possible.
13653 * A number of polkit actions are now bound together with "imply"
13654 rules. This should simplify creating UIs because many actions
13655 will now authenticate similar ones as well.
13657 * Unit files learnt a new condition ConditionACPower= which
13658 may be used to conditionalize a unit depending on whether an
13659 AC power source is connected or not, of whether the system
13660 is running on battery power.
13662 * systemctl gained a new "is-failed" verb that may be used in
13663 shell scripts and suchlike to check whether a specific unit
13664 is in the "failed" state.
13666 * The EnvironmentFile= setting in unit files now supports file
13667 globbing, and can hence be used to easily read a number of
13668 environment files at once.
13670 * systemd will no longer detect and recognize specific
13671 distributions. All distribution-specific #ifdeffery has been
13672 removed, systemd is now fully generic and
13673 distribution-agnostic. Effectively, not too much is lost as
13674 a lot of the code is still accessible via explicit configure
13675 switches. However, support for some distribution specific
13676 legacy configuration file formats has been dropped. We
13677 recommend distributions to simply adopt the configuration
13678 files everybody else uses now and convert the old
13679 configuration from packaging scripts. Most distributions
13680 already did that. If that's not possible or desirable,
13681 distributions are welcome to forward port the specific
13682 pieces of code locally from the git history.
13684 * When logging a message about a unit systemd will now always
13685 log the unit name in the message meta data.
13687 * localectl will now also discover system locale data that is
13688 not stored in locale archives, but directly unpacked.
13690 * logind will no longer unconditionally use framebuffer
13691 devices as seat masters, i.e. as devices that are required
13692 to be existing before a seat is considered preset. Instead,
13693 it will now look for all devices that are tagged as
13694 "seat-master" in udev. By default, framebuffer devices will
13695 be marked as such, but depending on local systems, other
13696 devices might be marked as well. This may be used to
13697 integrate graphics cards using closed source drivers (such
13698 as NVidia ones) more nicely into logind. Note however, that
13699 we recommend using the open source NVidia drivers instead,
13700 and no udev rules for the closed-source drivers will be
13701 shipped from us upstream.
13703 Contributions from: Adam Williamson, Alessandro Crismani, Auke
13704 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
13705 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
13706 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
13707 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
13708 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
13709 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
13710 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
13711 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
13712 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
13713 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
13714 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
13719 * udev gained support for loading additional device properties
13720 from an indexed database that is keyed by vendor/product IDs
13721 and similar device identifiers. For the beginning this
13722 "hwdb" is populated with data from the well-known PCI and
13723 USB database, but also includes PNP, ACPI and OID data. In
13724 the longer run this indexed database shall grow into
13725 becoming the one central database for non-essential
13726 userspace device metadata. Previously, data from the PCI/USB
13727 database was only attached to select devices, since the
13728 lookup was a relatively expensive operation due to O(n) time
13729 complexity (with n being the number of entries in the
13730 database). Since this is now O(1), we decided to add in this
13731 data for all devices where this is available, by
13732 default. Note that the indexed database needs to be rebuilt
13733 when new data files are installed. To achieve this you need
13734 to update your packaging scripts to invoke "udevadm hwdb
13735 --update" after installation of hwdb data files. For
13736 RPM-based distributions we introduced the new
13737 %udev_hwdb_update macro for this purpose.
13739 * The Journal gained support for the "Message Catalog", an
13740 indexed database to link up additional information with
13741 journal entries. For further details please check:
13743 https://www.freedesktop.org/wiki/Software/systemd/catalog
13745 The indexed message catalog database also needs to be
13746 rebuilt after installation of message catalog files. Use
13747 "journalctl --update-catalog" for this. For RPM-based
13748 distributions we introduced the %journal_catalog_update
13749 macro for this purpose.
13751 * The Python Journal bindings gained support for the standard
13752 Python logging framework.
13754 * The Journal API gained new functions for checking whether
13755 the underlying file system of a journal file is capable of
13756 properly reporting file change notifications, or whether
13757 applications that want to reflect journal changes "live"
13758 need to recheck journal files continuously in appropriate
13761 * It is now possible to set the "age" field for tmpfiles
13762 entries to 0, indicating that files matching this entry
13763 shall always be removed when the directories are cleaned up.
13765 * coredumpctl gained a new "gdb" verb which invokes gdb
13766 right-away on the selected coredump.
13768 * There's now support for "hybrid sleep" on kernels that
13769 support this, in addition to "suspend" and "hibernate". Use
13770 "systemctl hybrid-sleep" to make use of this.
13772 * logind's HandleSuspendKey= setting (and related settings)
13773 now gained support for a new "lock" setting to simply
13774 request the screen lock on all local sessions, instead of
13775 actually executing a suspend or hibernation.
13777 * systemd will now mount the EFI variables file system by
13780 * Socket units now gained support for configuration of the
13781 SMACK security label.
13783 * timedatectl will now output the time of the last and next
13784 daylight saving change.
13786 * We dropped support for various legacy and distro-specific
13787 concepts, such as insserv, early-boot SysV services
13788 (i.e. those for non-standard runlevels such as 'b' or 'S')
13789 or ArchLinux /etc/rc.conf support. We recommend the
13790 distributions who still need support this to either continue
13791 to maintain the necessary patches downstream, or find a
13792 different solution. (Talk to us if you have questions!)
13794 * Various systemd components will now bypass polkit checks for
13795 root and otherwise handle properly if polkit is not found to
13796 be around. This should fix most issues for polkit-less
13797 systems. Quite frankly this should have been this way since
13798 day one. It is absolutely our intention to make systemd work
13799 fine on polkit-less systems, and we consider it a bug if
13800 something does not work as it should if polkit is not around.
13802 * For embedded systems it is now possible to build udev and
13803 systemd without blkid and/or kmod support.
13805 * "systemctl switch-root" is now capable of switching root
13806 more than once. I.e. in addition to transitions from the
13807 initrd to the host OS it is now possible to transition to
13808 further OS images from the host. This is useful to implement
13809 offline updating tools.
13811 * Various other additions have been made to the RPM macros
13812 shipped with systemd. Use %udev_rules_update() after
13813 installing new udev rules files. %_udevhwdbdir,
13814 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
13815 %_sysctldir are now available which resolve to the right
13816 directories for packages to place various data files in.
13818 * journalctl gained the new --full switch (in addition to
13819 --all, to disable ellipsation for long messages.
13821 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
13822 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
13823 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
13824 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
13825 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
13826 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
13827 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
13828 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
13829 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
13833 * journalctl gained new --since= and --until= switches to
13834 filter by time. It also now supports nice filtering for
13835 units via --unit=/-u.
13837 * Type=oneshot services may use ExecReload= and do the
13840 * The journal daemon now supports time-based rotation and
13841 vacuuming, in addition to the usual disk-space based
13844 * The journal will now index the available field values for
13845 each field name. This enables clients to show pretty drop
13846 downs of available match values when filtering. The bash
13847 completion of journalctl has been updated
13848 accordingly. journalctl gained a new switch -F to list all
13849 values a certain field takes in the journal database.
13851 * More service events are now written as structured messages
13852 to the journal, and made recognizable via message IDs.
13854 * The timedated, localed and hostnamed mini-services which
13855 previously only provided support for changing time, locale
13856 and hostname settings from graphical DEs such as GNOME now
13857 also have a minimal (but very useful) text-based client
13858 utility each. This is probably the nicest way to changing
13859 these settings from the command line now, especially since
13860 it lists available options and is fully integrated with bash
13863 * There's now a new tool "systemd-coredumpctl" to list and
13864 extract coredumps from the journal.
13866 * We now install a README each in /var/log/ and
13867 /etc/rc.d/init.d explaining where the system logs and init
13868 scripts went. This hopefully should help folks who go to
13869 that dirs and look into the otherwise now empty void and
13870 scratch their heads.
13872 * When user-services are invoked (by systemd --user) the
13873 $MANAGERPID env var is set to the PID of systemd.
13875 * SIGRTMIN+24 when sent to a --user instance will now result
13876 in immediate termination of systemd.
13878 * gatewayd received numerous feature additions such as a
13879 "follow" mode, for live syncing and filtering.
13881 * browse.html now allows filtering and showing detailed
13882 information on specific entries. Keyboard navigation and
13883 mouse screen support has been added.
13885 * gatewayd/journalctl now supports HTML5/JSON
13886 Server-Sent-Events as output.
13888 * The SysV init script compatibility logic will now
13889 heuristically determine whether a script supports the
13890 "reload" verb, and only then make this available as
13891 "systemctl reload".
13893 * "systemctl status --follow" has been removed, use "journalctl
13896 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
13897 have been removed since they are hardly useful to be
13900 * And I'd like to take the opportunity to specifically mention
13901 Zbigniew for his great contributions. Zbigniew, you rock!
13903 Contributions from: Andrew Eikum, Christian Hesse, Colin
13904 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
13905 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
13906 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
13907 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
13908 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
13909 Jędrzejewski-Szmek, Сковорода Никита Андреевич
13913 * If /etc/vconsole.conf is non-existent or empty we will no
13914 longer load any console font or key map at boot by
13915 default. Instead the kernel defaults will be left
13916 intact. This is definitely the right thing to do, as no
13917 configuration should mean no configuration, and hard-coding
13918 font names that are different on all archs is probably a bad
13919 idea. Also, the kernel default key map and font should be
13920 good enough for most cases anyway, and mostly identical to
13921 the userspace fonts/key maps we previously overloaded them
13922 with. If distributions want to continue to default to a
13923 non-kernel font or key map they should ship a default
13924 /etc/vconsole.conf with the appropriate contents.
13926 Contributions from: Colin Walters, Daniel J Walsh, Dave
13927 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
13928 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
13932 * journalctl gained a new --cursor= switch to show entries
13933 starting from the specified location in the journal.
13935 * We now enforce a size limit on journal entry fields exported
13936 with "-o json" in journalctl. Fields larger than 4K will be
13937 assigned null. This can be turned off with --all.
13939 * An (optional) journal gateway daemon is now available as
13940 "systemd-journal-gatewayd.service". This service provides
13941 access to the journal via HTTP and JSON. This functionality
13942 will be used to implement live log synchronization in both
13943 pull and push modes, but has various other users too, such
13944 as easy log access for debugging of embedded devices. Right
13945 now it is already useful to retrieve the journal via HTTP:
13947 # systemctl start systemd-journal-gatewayd.service
13948 # wget http://localhost:19531/entries
13950 This will download the journal contents in a
13951 /var/log/messages compatible format. The same as JSON:
13953 # curl -H"Accept: application/json" http://localhost:19531/entries
13955 This service is also accessible via a web browser where a
13956 single static HTML5 app is served that uses the JSON logic
13957 to enable the user to do some basic browsing of the
13958 journal. This will be extended later on. Here's an example
13959 screenshot of this app in its current state:
13961 http://0pointer.de/public/journal-gatewayd
13963 Contributions from: Kay Sievers, Lennart Poettering, Robert
13964 Milasan, Tom Gundersen
13968 * The bash completion logic is now available for journalctl
13971 * We do not mount the "cpuset" controller anymore together with
13972 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
13973 started if no parameters are assigned to it. "cpuset" hence
13974 broke code that assumed it could create "cpu" groups and
13977 * journalctl -f will now subscribe to terminal size changes,
13978 and line break accordingly.
13980 Contributions from: Dave Reisner, Kay Sievers, Lennart
13981 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
13985 * nspawn will now create a symlink /etc/localtime in the
13986 container environment, copying the host's timezone
13987 setting. Previously this has been done via a bind mount, but
13988 since symlinks cannot be bind mounted this has now been
13989 changed to create/update the appropriate symlink.
13991 * journalctl -n's line number argument is now optional, and
13992 will default to 10 if omitted.
13994 * journald will now log the maximum size the journal files may
13995 take up on disk. This is particularly useful if the default
13996 built-in logic of determining this parameter from the file
13997 system size is used. Use "systemctl status
13998 systemd-journald.service" to see this information.
14000 * The multi-seat X wrapper tool has been stripped down. As X
14001 is now capable of enumerating graphics devices via udev in a
14002 seat-aware way the wrapper is not strictly necessary
14003 anymore. A stripped down temporary stop-gap is still shipped
14004 until the upstream display managers have been updated to
14005 fully support the new X logic. Expect this wrapper to be
14006 removed entirely in one of the next releases.
14008 * HandleSleepKey= in logind.conf has been split up into
14009 HandleSuspendKey= and HandleHibernateKey=. The old setting
14010 is not available anymore. X11 and the kernel are
14011 distinguishing between these keys and we should too. This
14012 also means the inhibition lock for these keys has been split
14015 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
14016 Poettering, Lukas Nykryn, Václav Pavlín
14020 * Whenever a unit changes state we will now log this to the
14021 journal and show along the unit's own log output in
14022 "systemctl status".
14024 * ConditionPathIsMountPoint= can now properly detect bind
14025 mount points too. (Previously, a bind mount of one file
14026 system to another place in the same file system could not be
14027 detected as mount, since they shared struct stat's st_dev
14030 * We will now mount the cgroup controllers cpu, cpuacct,
14031 cpuset and the controllers net_cls, net_prio together by
14034 * nspawn containers will now have a virtualized boot
14035 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
14036 over with a randomized ID at container initialization). This
14037 has the effect of making "journalctl -b" do the right thing
14040 * The JSON output journal serialization has been updated not
14041 to generate "endless" list objects anymore, but rather one
14042 JSON object per line. This is more in line how most JSON
14043 parsers expect JSON objects. The new output mode
14044 "json-pretty" has been added to provide similar output, but
14045 neatly aligned for readability by humans.
14047 * We dropped all explicit sync() invocations in the shutdown
14048 code. The kernel does this implicitly anyway in the kernel
14049 reboot() syscall. halt(8)'s -n option is now a compatibility
14052 * We now support virtualized reboot() in containers, as
14053 supported by newer kernels. We will fall back to exit() if
14054 CAP_SYS_REBOOT is not available to the container. Also,
14055 nspawn makes use of this now and will actually reboot the
14056 container if the containerized OS asks for that.
14058 * journalctl will only show local log output by default
14059 now. Use --merge (-m) to show remote log output, too.
14061 * libsystemd-journal gained the new sd_journal_get_usage()
14062 call to determine the current disk usage of all journal
14063 files. This is exposed in the new "journalctl --disk-usage"
14066 * journald gained a new configuration setting SplitMode= in
14067 journald.conf which may be used to control how user journals
14068 are split off. See journald.conf(5) for details.
14070 * A new condition type ConditionFileNotEmpty= has been added.
14072 * tmpfiles' "w" lines now support file globbing, to write
14073 multiple files at once.
14075 * We added Python bindings for the journal submission
14076 APIs. More Python APIs for a number of selected APIs will
14077 likely follow. Note that we intend to add native bindings
14078 only for the Python language, as we consider it common
14079 enough to deserve bindings shipped within systemd. There are
14080 various projects outside of systemd that provide bindings
14081 for languages such as PHP or Lua.
14083 * Many conditions will now resolve specifiers such as %i. In
14084 addition, PathChanged= and related directives of .path units
14085 now support specifiers as well.
14087 * There's now a new RPM macro definition for the system preset
14090 * journald will now warn if it ca not forward a message to the
14091 syslog daemon because its socket is full.
14093 * timedated will no longer write or process /etc/timezone,
14094 except on Debian. As we do not support late mounted /usr
14095 anymore /etc/localtime always being a symlink is now safe,
14096 and hence the information in /etc/timezone is not necessary
14099 * logind will now always reserve one VT for a text getty (VT6
14100 by default). Previously if more than 6 X sessions where
14101 started they took up all the VTs with auto-spawned gettys,
14102 so that no text gettys were available anymore.
14104 * udev will now automatically inform the btrfs kernel logic
14105 about btrfs RAID components showing up. This should make
14106 simple hotplug based btrfs RAID assembly work.
14108 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
14109 (but not for its children which will stay at the kernel
14110 default). This should allow setups with a lot more listening
14113 * systemd will now always pass the configured timezone to the
14114 kernel at boot. timedated will do the same when the timezone
14117 * logind's inhibition logic has been updated. By default,
14118 logind will now handle the lid switch, the power and sleep
14119 keys all the time, even in graphical sessions. If DEs want
14120 to handle these events on their own they should take the new
14121 handle-power-key, handle-sleep-key and handle-lid-switch
14122 inhibitors during their runtime. A simple way to achieve
14123 that is to invoke the DE wrapped in an invocation of:
14125 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
14127 * Access to unit operations is now checked via SELinux taking
14128 the unit file label and client process label into account.
14130 * systemd will now notify the administrator in the journal
14131 when he over-mounts a non-empty directory.
14133 * There are new specifiers that are resolved in unit files,
14134 for the hostname (%H), the machine ID (%m) and the boot ID
14137 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
14138 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
14139 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
14140 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
14141 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
14142 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
14143 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
14147 * Support for reading structured kernel messages from
14148 /dev/kmsg has now been added and is enabled by default.
14150 * Support for reading kernel messages from /proc/kmsg has now
14151 been removed. If you want kernel messages in the journal
14152 make sure to run a recent kernel (>= 3.5) that supports
14153 reading structured messages from /dev/kmsg (see
14154 above). /proc/kmsg is now exclusive property of classic
14155 syslog daemons again.
14157 * The libudev API gained the new
14158 udev_device_new_from_device_id() call.
14160 * The logic for file system namespace (ReadOnlyDirectory=,
14161 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
14162 require pivot_root() anymore. This means fewer temporary
14163 directories are created below /tmp for this feature.
14165 * nspawn containers will now see and receive all submounts
14166 made on the host OS below the root file system of the
14169 * Forward Secure Sealing is now supported for Journal files,
14170 which provide cryptographical sealing of journal files so
14171 that attackers cannot alter log history anymore without this
14172 being detectable. Lennart will soon post a blog story about
14173 this explaining it in more detail.
14175 * There are two new service settings RestartPreventExitStatus=
14176 and SuccessExitStatus= which allow configuration of exit
14177 status (exit code or signal) which will be excepted from the
14178 restart logic, resp. consider successful.
14180 * journalctl gained the new --verify switch that can be used
14181 to check the integrity of the structure of journal files and
14182 (if Forward Secure Sealing is enabled) the contents of
14185 * nspawn containers will now be run with /dev/stdin, /dev/fd/
14186 and similar symlinks pre-created. This makes running shells
14187 as container init process a lot more fun.
14189 * The fstab support can now handle PARTUUID= and PARTLABEL=
14192 * A new ConditionHost= condition has been added to match
14193 against the hostname (with globs) and machine ID. This is
14194 useful for clusters where a single OS image is used to
14195 provision a large number of hosts which shall run slightly
14196 different sets of services.
14198 * Services which hit the restart limit will now be placed in a
14201 Contributions from: Bertram Poettering, Dave Reisner, Huang
14202 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
14203 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
14207 * When running in --user mode systemd will now become a
14208 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
14209 tree a lot more organized.
14211 * A new PartOf= unit dependency type has been introduced that
14212 may be used to group services in a natural way.
14214 * "systemctl enable" may now be used to enable instances of
14217 * journalctl now prints error log levels in red, and
14218 warning/notice log levels in bright white. It also supports
14219 filtering by log level now.
14221 * cgtop gained a new -n switch (similar to top), to configure
14222 the maximum number of iterations to run for. It also gained
14223 -b, to run in batch mode (accepting no input).
14225 * The suffix ".service" may now be omitted on most systemctl
14226 command lines involving service unit names.
14228 * There's a new bus call in logind to lock all sessions, as
14229 well as a loginctl verb for it "lock-sessions".
14231 * libsystemd-logind.so gained a new call sd_journal_perror()
14232 that works similar to libc perror() but logs to the journal
14233 and encodes structured information about the error number.
14235 * /etc/crypttab entries now understand the new keyfile-size=
14238 * shutdown(8) now can send a (configurable) wall message when
14239 a shutdown is cancelled.
14241 * The mount propagation mode for the root file system will now
14242 default to "shared", which is useful to make containers work
14243 nicely out-of-the-box so that they receive new mounts from
14244 the host. This can be undone locally by running "mount
14245 --make-rprivate /" if needed.
14247 * The prefdm.service file has been removed. Distributions
14248 should maintain this unit downstream if they intend to keep
14249 it around. However, we recommend writing normal unit files
14250 for display managers instead.
14252 * Since systemd is a crucial part of the OS we will now
14253 default to a number of compiler switches that improve
14254 security (hardening) such as read-only relocations, stack
14255 protection, and suchlike.
14257 * The TimeoutSec= setting for services is now split into
14258 TimeoutStartSec= and TimeoutStopSec= to allow configuration
14259 of individual time outs for the start and the stop phase of
14262 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
14263 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
14264 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
14265 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
14266 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
14267 Gundersen, Zbigniew Jędrzejewski-Szmek
14271 * The journal and id128 C APIs are now fully documented as man
14274 * Extra safety checks have been added when transitioning from
14275 the initial RAM disk to the main system to avoid accidental
14278 * /etc/crypttab entries now understand the new keyfile-offset=
14281 * systemctl -t can now be used to filter by unit load state.
14283 * The journal C API gained the new sd_journal_wait() call to
14284 make writing synchronous journal clients easier.
14286 * journalctl gained the new -D switch to show journals from a
14287 specific directory.
14289 * journalctl now displays a special marker between log
14290 messages of two different boots.
14292 * The journal is now explicitly flushed to /var via a service
14293 systemd-journal-flush.service, rather than implicitly simply
14294 by seeing /var/log/journal to be writable.
14296 * journalctl (and the journal C APIs) can now match for much
14297 more complex expressions, with alternatives and
14300 * When transitioning from the initial RAM disk to the main
14301 system we will now kill all processes in a killing spree to
14302 ensure no processes stay around by accident.
14304 * Three new specifiers may be used in unit files: %u, %h, %s
14305 resolve to the user name, user home directory resp. user
14306 shell. This is useful for running systemd user instances.
14308 * We now automatically rotate journal files if their data
14309 object hash table gets a fill level > 75%. We also size the
14310 hash table based on the configured maximum file size. This
14311 together should lower hash collisions drastically and thus
14312 speed things up a bit.
14314 * journalctl gained the new "--header" switch to introspect
14315 header data of journal files.
14317 * A new setting SystemCallFilters= has been added to services which may
14318 be used to apply deny lists or allow lists to system calls. This is
14319 based on SECCOMP Mode 2 of Linux 3.5.
14321 * nspawn gained a new --link-journal= switch (and quicker: -j)
14322 to link the container journal with the host. This makes it
14323 very easy to centralize log viewing on the host for all
14324 guests while still keeping the journal files separated.
14326 * Many bugfixes and optimizations
14328 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
14329 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
14330 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
14335 * Several tools now understand kernel command line arguments,
14336 which are only read when run in an initial RAM disk. They
14337 usually follow closely their normal counterparts, but are
14340 * There's a new tool to analyze the readahead files that are
14341 automatically generated at boot. Use:
14343 /usr/lib/systemd/systemd-readahead analyze /.readahead
14345 * We now provide an early debug shell on tty9 if this enabled. Use:
14347 systemctl enable debug-shell.service
14349 * All plymouth related units have been moved into the Plymouth
14350 package. Please make sure to upgrade your Plymouth version
14353 * systemd-tmpfiles now supports getting passed the basename of
14354 a configuration file only, in which case it will look for it
14355 in all appropriate directories automatically.
14357 * udevadm info now takes a /dev or /sys path as argument, and
14358 does the right thing. Example:
14360 udevadm info /dev/sda
14361 udevadm info /sys/class/block/sda
14363 * systemctl now prints a warning if a unit is stopped but a
14364 unit that might trigger it continues to run. Example: a
14365 service is stopped but the socket that activates it is left
14368 * "systemctl status" will now mention if the log output was
14369 shortened due to rotation since a service has been started.
14371 * The journal API now exposes functions to determine the
14372 "cutoff" times due to rotation.
14374 * journald now understands SIGUSR1 and SIGUSR2 for triggering
14375 immediately flushing of runtime logs to /var if possible,
14376 resp. for triggering immediate rotation of the journal
14379 * It is now considered an error if a service is attempted to
14380 be stopped that is not loaded.
14382 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
14384 * systemd-analyze now supports Python 3
14386 * tmpfiles now supports cleaning up directories via aging
14387 where the first level dirs are always kept around but
14388 directories beneath it automatically aged. This is enabled
14389 by prefixing the age field with '~'.
14391 * Seat objects now expose CanGraphical, CanTTY properties
14392 which is required to deal with very fast bootups where the
14393 display manager might be running before the graphics drivers
14394 completed initialization.
14396 * Seat objects now expose a State property.
14398 * We now include RPM macros for service enabling/disabling
14399 based on the preset logic. We recommend RPM based
14400 distributions to make use of these macros if possible. This
14401 makes it simpler to reuse RPM spec files across
14404 * We now make sure that the collected systemd unit name is
14405 always valid when services log to the journal via
14408 * There's a new man page kernel-command-line(7) detailing all
14409 command line options we understand.
14411 * The fstab generator may now be disabled at boot by passing
14412 fstab=0 on the kernel command line.
14414 * A new kernel command line option modules-load= is now understood
14415 to load a specific kernel module statically, early at boot.
14417 * Unit names specified on the systemctl command line are now
14418 automatically escaped as needed. Also, if file system or
14419 device paths are specified they are automatically turned
14420 into the appropriate mount or device unit names. Example:
14422 systemctl status /home
14423 systemctl status /dev/sda
14425 * The SysVConsole= configuration option has been removed from
14426 system.conf parsing.
14428 * The SysV search path is no longer exported on the D-Bus
14431 * The Names= option has been removed from unit file parsing.
14433 * There's a new man page bootup(7) detailing the boot process.
14435 * Every unit and every generator we ship with systemd now
14436 comes with full documentation. The self-explanatory boot is
14439 * A couple of services gained "systemd-" prefixes in their
14440 name if they wrap systemd code, rather than only external
14441 code. Among them fsck@.service which is now
14442 systemd-fsck@.service.
14444 * The HaveWatchdog property has been removed from the D-Bus
14447 * systemd.confirm_spawn= on the kernel command line should now
14450 * There's a new man page crypttab(5) which details all options
14451 we actually understand.
14453 * systemd-nspawn gained a new --capability= switch to pass
14454 additional capabilities to the container.
14456 * timedated will now read known NTP implementation unit names
14457 from /usr/lib/systemd/ntp-units.d/*.list,
14458 systemd-timedated-ntp.target has been removed.
14460 * journalctl gained a new switch "-b" that lists log data of
14461 the current boot only.
14463 * The notify socket is in the abstract namespace again, in
14464 order to support daemons which chroot() at start-up.
14466 * There is a new Storage= configuration option for journald
14467 which allows configuration of where log data should go. This
14468 also provides a way to disable journal logging entirely, so
14469 that data collected is only forwarded to the console, the
14470 kernel log buffer or another syslog implementation.
14472 * Many bugfixes and optimizations
14474 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
14475 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
14476 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
14477 Shawn Landden, Tom Gundersen
14481 * "systemctl help <unit>" now shows the man page if one is
14484 * Several new man pages have been added.
14486 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
14487 MaxLevelConsole= can now be specified in
14488 journald.conf. These options allow reducing the amount of
14489 data stored on disk or forwarded by the log level.
14491 * TimerSlackNSec= can now be specified in system.conf for
14492 PID1. This allows system-wide power savings.
14494 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
14495 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
14500 * logind is now capable of (optionally) handling power and
14501 sleep keys as well as the lid switch.
14503 * journalctl now understands the syntax "journalctl
14504 /usr/bin/avahi-daemon" to get all log output of a specific
14507 * CapabilityBoundingSet= in system.conf now also influences
14508 the capability bound set of usermode helpers of the kernel.
14510 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
14511 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
14512 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
14513 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
14517 * Note that we skipped 139 releases here in order to set the
14518 new version to something that is greater than both udev's
14519 and systemd's most recent version number.
14521 * udev: all udev sources are merged into the systemd source tree now.
14522 All future udev development will happen in the systemd tree. It
14523 is still fully supported to use the udev daemon and tools without
14524 systemd running, like in initramfs or other init systems. Building
14525 udev though, will require the *build* of the systemd tree, but
14526 udev can be properly *run* without systemd.
14528 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
14529 should be used to create dead device nodes as workarounds for broken
14532 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
14533 no longer supported. udev_monitor_new_from_netlink() needs to be
14534 used to subscribe to events.
14536 * udev: when udevd is started by systemd, processes which are left
14537 behind by forking them off of udev rules, are unconditionally cleaned
14538 up and killed now after the event handling has finished. Services or
14539 daemons must be started as systemd services. Services can be
14540 pulled-in by udev to get started, but they can no longer be directly
14541 forked by udev rules.
14543 * udev: the daemon binary is called systemd-udevd now and installed
14544 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
14545 to adapt to that, create symlink, or rename the binary after building
14548 * libudev no longer provides these symbols:
14549 udev_monitor_from_socket()
14550 udev_queue_get_failed_list_entry()
14551 udev_get_{dev,sys,run}_path()
14552 The versions number was bumped and symbol versioning introduced.
14554 * systemd-loginctl and systemd-journalctl have been renamed
14555 to loginctl and journalctl to match systemctl.
14557 * The config files: /etc/systemd/systemd-logind.conf and
14558 /etc/systemd/systemd-journald.conf have been renamed to
14559 logind.conf and journald.conf. Package updates should rename
14560 the files to the new names on upgrade.
14562 * For almost all files the license is now LGPL2.1+, changed
14563 from the previous GPL2.0+. Exceptions are some minor stuff
14564 of udev (which will be changed to LGPL2.1 eventually, too),
14565 and the MIT licensed sd-daemon.[ch] library that is suitable
14566 to be used as drop-in files.
14568 * systemd and logind now handle system sleep states, in
14569 particular suspending and hibernating.
14571 * logind now implements a sleep/shutdown/idle inhibiting logic
14572 suitable for a variety of uses. Soonishly Lennart will blog
14573 about this in more detail.
14575 * var-run.mount and var-lock.mount are no longer provided
14576 (which previously bind mounted these directories to their new
14577 places). Distributions which have not converted these
14578 directories to symlinks should consider stealing these files
14579 from git history and add them downstream.
14581 * We introduced the Documentation= field for units and added
14582 this to all our shipped units. This is useful to make it
14583 easier to explore the boot and the purpose of the various
14586 * All smaller setup units (such as
14587 systemd-vconsole-setup.service) now detect properly if they
14588 are run in a container and are skipped when
14589 appropriate. This guarantees an entirely noise-free boot in
14590 Linux container environments such as systemd-nspawn.
14592 * A framework for implementing offline system updates is now
14593 integrated, for details see:
14594 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
14596 * A new service type Type=idle is available now which helps us
14597 avoiding ugly interleaving of getty output and boot status
14600 * There's now a system-wide CapabilityBoundingSet= option to
14601 globally reduce the set of capabilities for the
14602 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
14603 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
14604 even CAP_NET_ADMIN system-wide for secure systems.
14606 * There are now system-wide DefaultLimitXXX= options to
14607 globally change the defaults of the various resource limits
14608 for all units started by PID 1.
14610 * Harald Hoyer's systemd test suite has been integrated into
14611 systemd which allows easy testing of systemd builds in qemu
14612 and nspawn. (This is really awesome! Ask us for details!)
14614 * The fstab parser is now implemented as generator, not inside
14617 * systemctl will now warn you if .mount units generated from
14618 /etc/fstab are out of date due to changes in fstab that
14619 have not been read by systemd yet.
14621 * systemd is now suitable for usage in initrds. Dracut has
14622 already been updated to make use of this. With this in place
14623 initrds get a slight bit faster but primarily are much
14624 easier to introspect and debug since "systemctl status" in
14625 the host system can be used to introspect initrd services,
14626 and the journal from the initrd is kept around too.
14628 * systemd-delta has been added, a tool to explore differences
14629 between user/admin configuration and vendor defaults.
14631 * PrivateTmp= now affects both /tmp and /var/tmp.
14633 * Boot time status messages are now much prettier and feature
14634 proper english language. Booting up systemd has never been
14637 * Read-ahead pack files now include the inode number of all
14638 files to pre-cache. When the inode changes the pre-caching
14639 is not attempted. This should be nicer to deal with updated
14640 packages which might result in changes of read-ahead
14643 * We now temporaritly lower the kernel's read_ahead_kb variable
14644 when collecting read-ahead data to ensure the kernel's
14645 built-in read-ahead does not add noise to our measurements
14646 of necessary blocks to pre-cache.
14648 * There's now RequiresMountsFor= to add automatic dependencies
14649 for all mounts necessary for a specific file system path.
14651 * MountAuto= and SwapAuto= have been removed from
14652 system.conf. Mounting file systems at boot has to take place
14655 * nspawn now learned a new switch --uuid= to set the machine
14656 ID on the command line.
14658 * nspawn now learned the -b switch to automatically search
14659 for an init system.
14661 * vt102 is now the default TERM for serial TTYs, upgraded from
14664 * systemd-logind now works on VT-less systems.
14666 * The build tree has been reorganized. The individual
14667 components now have directories of their own.
14669 * A new condition type ConditionPathIsReadWrite= is now available.
14671 * nspawn learned the new -C switch to create cgroups for the
14672 container in other hierarchies.
14674 * We now have support for hardware watchdogs, configurable in
14677 * The scheduled shutdown logic now has a public API.
14679 * We now mount /tmp as tmpfs by default, but this can be
14680 masked and /etc/fstab can override it.
14682 * Since udisks does not make use of /media anymore we are not
14683 mounting a tmpfs on it anymore.
14685 * journalctl gained a new --local switch to only interleave
14686 locally generated journal files.
14688 * We can now load the IMA policy at boot automatically.
14690 * The GTK tools have been split off into a systemd-ui.
14692 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
14693 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
14694 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
14695 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
14696 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
14697 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
14698 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
14699 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
14700 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
14705 * This is mostly a bugfix release
14707 * Support optional initialization of the machine ID from the
14708 KVM or container configured UUID.
14710 * Support immediate reboots with "systemctl reboot -ff"
14712 * Show /etc/os-release data in systemd-analyze output
14714 * Many bugfixes for the journal, including endianness fixes and
14715 ensuring that disk space enforcement works
14717 * sd-login.h is C++ compatible again
14719 * Extend the /etc/os-release format on request of the Debian
14722 * We now refuse non-UTF8 strings used in various configuration
14723 and unit files. This is done to ensure we do not pass invalid
14724 data over D-Bus or expose it elsewhere.
14726 * Register Mimo USB Screens as suitable for automatic seat
14729 * Read SELinux client context from journal clients in a race
14732 * Reorder configuration file lookup order. /etc now always
14733 overrides /run in order to allow the administrator to always
14734 and unconditionally override vendor-supplied or
14735 automatically generated data.
14737 * The various user visible bits of the journal now have man
14738 pages. We still lack man pages for the journal API calls
14741 * We now ship all man pages in HTML format again in the
14744 Contributions from: Dave Reisner, Dirk Eibach, Frederic
14745 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
14746 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
14751 * This is mostly a bugfix release
14753 * systems lacking /etc/os-release are no longer supported.
14755 * Various functionality updates to libsystemd-login.so
14757 * Track class of PAM logins to distinguish greeters from
14758 normal user logins.
14760 Contributions from: Kay Sievers, Lennart Poettering, Michael
14765 * This is an important bugfix release for v41.
14767 * Building man pages is now optional which should be useful
14768 for those building systemd from git but unwilling to install
14771 * Watchdog support for supervising services is now usable. In
14772 a future release support for hardware watchdogs
14773 (i.e. /dev/watchdog) will be added building on this.
14775 * Service start rate limiting is now configurable and can be
14776 turned off per service. When a start rate limit is hit a
14777 reboot can automatically be triggered.
14779 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
14781 Contributions from: Benjamin Franzke, Bill Nottingham,
14782 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
14783 Schmidt, Michał Górny, Piotr Drąg
14787 * The systemd binary is installed /usr/lib/systemd/systemd now;
14788 An existing /sbin/init symlink needs to be adapted with the
14791 * The code that loads kernel modules has been ported to invoke
14792 libkmod directly, instead of modprobe. This means we do not
14793 support systems with module-init-tools anymore.
14795 * Watchdog support is now already useful, but still not
14798 * A new kernel command line option systemd.setenv= is
14799 understood to set system wide environment variables
14800 dynamically at boot.
14802 * We now limit the set of capabilities of systemd-journald.
14804 * We now set SIGPIPE to ignore by default, since it only is
14805 useful in shell pipelines, and has little use in general
14806 code. This can be disabled with IgnoreSIPIPE=no in unit
14809 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
14810 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
14815 * This is mostly a bugfix release
14817 * We now expose the reason why a service failed in the
14818 "Result" D-Bus property.
14820 * Rudimentary service watchdog support (will be completed over
14821 the next few releases.)
14823 * When systemd forks off in order execute some service we will
14824 now immediately changes its argv[0] to reflect which process
14825 it will execute. This is useful to minimize the time window
14826 with a generic argv[0], which makes bootcharts more useful
14828 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
14829 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
14830 Mike Kazantsev, Ray Strode
14834 * This is mostly a test release, but incorporates many
14837 * New systemd-cgtop tool to show control groups by their
14840 * Linking against libacl for ACLs is optional again. If
14841 disabled, support tracking device access for active logins
14842 goes becomes unavailable, and so does access to the user
14843 journals by the respective users.
14845 * If a group "adm" exists, journal files are automatically
14846 owned by them, thus allow members of this group full access
14847 to the system journal as well as all user journals.
14849 * The journal now stores the SELinux context of the logging
14850 client for all entries.
14852 * Add C++ inclusion guards to all public headers
14854 * New output mode "cat" in the journal to print only text
14855 messages, without any meta data like date or time.
14857 * Include tiny X server wrapper as a temporary stop-gap to
14858 teach XOrg udev display enumeration. This is used by display
14859 managers such as gdm, and will go away as soon as XOrg
14860 learned native udev hotplugging for display devices.
14862 * Add new systemd-cat tool for executing arbitrary programs
14863 with STDERR/STDOUT connected to the journal. Can also act as
14864 BSD logger replacement, and does so by default.
14866 * Optionally store all locally generated coredumps in the
14867 journal along with meta data.
14869 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
14870 writing short strings to files (for usage for /sys), and for
14871 creating symlinks, character and block device nodes.
14873 * New unit file option ControlGroupPersistent= to make cgroups
14874 persistent, following the mechanisms outlined in
14875 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
14877 * Support multiple local RTCs in a sane way
14879 * No longer monopolize IO when replaying readahead data on
14880 rotating disks, since we might starve non-file-system IO to
14881 death, since fanotify() will not see accesses done by blkid,
14884 * Do not show kernel threads in systemd-cgls anymore, unless
14885 requested with new -k switch.
14887 Contributions from: Dan Horák, Kay Sievers, Lennart
14888 Poettering, Michal Schmidt
14892 * This is mostly a test release, but incorporates many
14895 * The git repository moved to:
14896 git://anongit.freedesktop.org/systemd/systemd
14897 ssh://git.freedesktop.org/git/systemd/systemd
14899 * First release with the journal
14900 http://0pointer.de/blog/projects/the-journal.html
14902 * The journal replaces both systemd-kmsg-syslogd and
14903 systemd-stdout-bridge.
14905 * New sd_pid_get_unit() API call in libsystemd-logind
14907 * Many systemadm clean-ups
14909 * Introduce remote-fs-pre.target which is ordered before all
14910 remote mounts and may be used to start services before all
14913 * Added Mageia support
14915 * Add bash completion for systemd-loginctl
14917 * Actively monitor PID file creation for daemons which exit in
14918 the parent process before having finished writing the PID
14919 file in the daemon process. Daemons which do this need to be
14920 fixed (i.e. PID file creation must have finished before the
14921 parent exits), but we now react a bit more gracefully to them.
14923 * Add colourful boot output, mimicking the well-known output
14924 of existing distributions.
14926 * New option PassCredentials= for socket units, for
14927 compatibility with a recent kernel ABI breakage.
14929 * /etc/rc.local is now hooked in via a generator binary, and
14930 thus will no longer act as synchronization point during
14933 * systemctl list-unit-files now supports --root=.
14935 * systemd-tmpfiles now understands two new commands: z, Z for
14936 relabelling files according to the SELinux database. This is
14937 useful to apply SELinux labels to specific files in /sys,
14938 among other things.
14940 * Output of SysV services is now forwarded to both the console
14941 and the journal by default, not only just the console.
14943 * New man pages for all APIs from libsystemd-login.
14945 * The build tree got reorganized and the build system is a
14946 lot more modular allowing embedded setups to specifically
14947 select the components of systemd they are interested in.
14949 * Support for Linux systems lacking the kernel VT subsystem is
14952 * configure's --with-rootdir= got renamed to
14953 --with-rootprefix= to follow the naming used by udev and
14956 * Unless specified otherwise we will now install to /usr instead
14957 of /usr/local by default.
14959 * Processes with '@' in argv[0][0] are now excluded from the
14960 final shut-down killing spree, following the logic explained
14962 https://systemd.io/ROOT_STORAGE_DAEMONS/
14964 * All processes remaining in a service cgroup when we enter
14965 the START or START_PRE states are now killed with
14966 SIGKILL. That means it is no longer possible to spawn
14967 background processes from ExecStart= lines (which was never
14968 supported anyway, and bad style).
14970 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
14971 reloading of units together.
14973 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
14974 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
14975 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
14976 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
14977 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek