1 systemd System and Service Manager
3 CHANGES WITH 255 in spe:
5 Announcements of Future Feature Removals and Incompatible Changes:
7 * Support for split-usr (/usr/ mounted separately during late boot,
8 instead of being mounted by the initrd before switching to the rootfs)
9 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
10 /usr/lib/, …) has been removed. For more details, see:
11 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
13 * We intend to remove cgroup v1 support from a systemd release after
14 the end of 2023. If you run services that make explicit use of
15 cgroup v1 features (i.e. the "legacy hierarchy" with separate
16 hierarchies for each controller), please implement compatibility with
17 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
18 Most of Linux userspace has been ported over already.
20 * Support for System V service scripts is now deprecated and will be
21 removed in a future release. Please make sure to update your software
22 *now* to include a native systemd unit file instead of a legacy
23 System V script to retain compatibility with future systemd releases.
25 * Support for the SystemdOptions EFI variable is deprecated.
26 'bootctl systemd-efi-options' will emit a warning when used. It seems
27 that this feature is little-used and it is better to use alternative
28 approaches like credentials and confexts. The plan is to drop support
29 altogether at a later point, but this might be revisited based on
32 * systemd-run's switch --expand-environment= which currently is disabled
33 by default when combined with --scope, will be changed in a future
34 release to be enabled by default.
36 * "systemctl switch-root" is now restricted to initrd transitions only.
38 Transitions between real systems should be done with
39 "systemctl soft-reboot" instead.
41 * The "ip=off" and "ip=none" kernel command line options interpreted by
42 systemd-network-generator will now result in IPv6RA + link-local
43 addressing being disabled, too. Previously DHCP was turned off, but
44 IPv6RA and IPv6 link-local addressing was left enabled.
46 * The NAMING_BRIDGE_MULTIFUNCTION_SLOT naming scheme has been deprecated
49 * SuspendMode=, HibernateState= and HybridSleepState= in the [Sleep]
50 section of systemd-sleep.conf are now deprecated and have no effect.
51 They did not (and could not) take any value other than the respective
52 default. HybridSleepMode= is also deprecated, and will now always use
53 the 'suspend' disk mode.
57 * The way services are spawned has been overhauled. Previously, a
58 process was forked that shared all of the manager's memory (via
59 copy-on-write) while doing all the required setup (e.g.: mount
60 namespaces, CGroup configuration, etc.) before exec'ing the target
61 executable. This was problematic for various reasons: several glibc
62 APIs were called that are not supposed to be used after a fork but
63 before an exec, copy-on-write meant that if either process (the
64 manager or the child) touched a memory page a copy was triggered, and
65 also the memory footprint of the child process was that of the
66 manager, but with the memory limits of the service. From this version
67 onward, the new process is spawned using CLONE_VM and CLONE_VFORK
68 semantics via posix_spawn(3), and it immediately execs a new internal
69 binary, systemd-executor, that receives the configuration to apply
70 via memfd, and sets up the process before exec'ing the target
71 executable. The systemd-executor binary is pinned by file descriptor
72 by each manager instance (system and users), and the reference is
73 updated on daemon-reexec - it is thus important to reexec all running
74 manager instances when the systemd-executor and/or libsystemd*
75 libraries are updated on the filesystem.
77 * Most of the internal process tracking is being changed to use PIDFDs
78 instead of PIDs when the kernel supports it, to improve robustness
81 * A new option SurviveFinalKillSignal= can be used to configure the
82 unit to be skipped in the final SIGTERM/SIGKILL spree on shutdown.
83 This is part of the required configuration to let a unit's processes
84 survive a soft-reboot operation.
86 * System extension images (sysext) can now set
87 EXTENSION_RELOAD_MANAGER=1 in their extension-release files to
88 automatically reload the service manager (PID 1) when
89 merging/refreshing/unmerging on boot. Generally, while this can be
90 used to ship services in system extension images it's recommended to
91 do that via portable services instead.
93 * The ExtensionImages= and ExtensionDirectories= options now support
94 confexts images/directories.
96 * A new option NFTSet= provides a method for integrating dynamic cgroup
97 IDs into firewall rules with NFT sets. The benefit of using this
98 setting is to be able to use control group as a selector in firewall
99 rules easily and this in turn allows more fine grained filtering.
100 Also, NFT rules for cgroup matching use numeric cgroup IDs, which
101 change every time a service is restarted, making them hard to use in
104 * A new option CoredumpReceive= can be set for service and scope units,
105 together with Delegate=yes, to make systemd-coredump on the host
106 forward core files from processes crashing inside the delegated
107 CGroup subtree to systemd-coredump running in the container. This new
108 option is by default used by systemd-nspawn containers that use the
111 * A new ConditionSecurity=measured-uki option is now available, to ensure
112 a unit can only run when the system has been booted from a measured UKI.
114 * MemoryAvailable= now considers physical memory if there are no CGroup
115 memory limits set anywhere in the tree.
117 * The $USER environment variable is now always set for services, while
118 previously it was only set if User= was specified. A new option
119 SetLoginEnvironment= is now supported to determine whether to also set
120 $HOME, $LOGNAME, and $SHELL.
122 * Socket units now support a new pair of
123 PollLimitBurst=/PollLimitInterval= options to configure a limit on
124 how often polling events on the file descriptors backing this unit
125 will be considered within a time window.
127 * Scope units can now be created using PIDFDs instead of PIDs to select
128 the processes they should include.
130 * Sending SIGRTMIN+18 with 0x500 as sigqueue() value will now cause the
131 manager to dump the list of currently pending jobs.
133 * If the kernel supports MOVE_MOUNT_BENEATH, the systemctl and
134 machinectl bind and mount-image verbs will now cause the new mount to
135 replace the old mount (if any), instead of overmounting it.
137 * Units now have MemoryPeak, MemorySwapPeak, MemorySwapCurrent and
138 MemoryZSwapCurrent properties, which respectively contain the values
139 of the cgroup v2's memory.peak, memory.swap.peak, memory.swap.current
140 and memory.zswap.current properties. This information is also show in
141 "systemctl status" output, if available.
143 TPM2 Support + Disk Encryption & Authentication:
145 * systemd-cryptenroll now allows specifying a PCR bank and explicit hash
146 value in the --tpm2-pcrs= option.
148 * systemd-cryptenroll now allows specifying a TPM2 key handle (nv
149 index) to be used instead of the default SRK via the new
150 --tpm2-seal-key-handle= option.
152 * systemd-cryptenroll now allows TPM2 enrollment using only a TPM2
153 public key (in TPM2B_PUBLIC format) – without access to the TPM2
154 device itself – which enables offline sealing of LUKS images for a
155 specific TPM2 chip, as long as the SRK public key is known. Pass the
156 public to the tool via the new --tpm2-device-key= switch.
158 * systemd-cryptsetup is now installed in /usr/bin/ and is no longer an
159 internal-only executable.
161 * The TPM2 Storage Root Key will now be set up, if not already present,
162 by a new systemd-tpm2-setup.service early boot service. The SRK will
163 be stored in PEM format and TPM2_PUBLIC format (the latter is useful
164 for systemd-cryptenroll --tpm2-device-key=, as mentioned above) for
165 easier access. A new "srk" verb has been added to systemd-analyze to
166 allow extracting it on demand if it is already set up.
168 * The internal systemd-pcrphase executable has been renamed to
171 * The systemd-pcrextend tool gained a new --pcr= switch to override
172 which PCR to measure into.
174 * systemd-pcrextend now exposes a Varlink interface at
175 io.systemd.PCRExtend that can be used to do measurements and event
178 * TPM measurements are now also written to an event log at
179 /run/log/systemd/tpm2-measure.log, using a derivative of the TCG
180 Canonical Event Log format. Previously we'd only log them to the
181 journal, where they however were subject to rotation and similar.
183 * A new component "systemd-pcrlock" has been added that allows managing
184 local TPM2 PCR policies for PCRs 0-7 and similar, which are hard to
185 predict by the OS vendor because of the inherently local nature of
186 what measurements they contain, such as firmware versions of the
187 system and extension cards and suchlike. pcrlock can predict PCR
188 measurements ahead of time based on various inputs, such as the local
189 TPM2 event log, GPT partition tables, PE binaries, UKI kernels, and
190 various other things. It can then pre-calculate a TPM2 policy from
191 this, which it stores in an TPM2 NV index. TPM2 objects (such as disk
192 encryption keys) can be locked against this NV index, so that they
193 are locked against a specific combination of system firmware and
194 state. Alternatives for each component are supported to allowlist
195 multiple kernel versions or boot loader version simultaneously
196 without losing access to the disk encryption keys. The tool can also
197 be used to analyze and validate the local TPM2 event log.
198 systemd-cryptsetup, systemd-cryptenroll, systemd-repart have all been
199 updated to support such policies. There's currently no support for
200 locking the system's root disk against a pcrlock policy, this will be
201 added soon. Moreover, it is currently not possible to combine a
202 pcrlock policy with a signed PCR policy. This component is
203 experimental and its public interface is subject to change.
205 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
207 * bootctl will now show whether the system was booted from a UKI in its
210 * systemd-boot and systemd-stub now use different project keys in their
211 respective SBAT sections, so that they can be revoked individually if
214 * systemd-boot will no longer load unverified Devicetree blobs when UEFI
215 SecureBoot is enabled. For more details see:
216 https://github.com/systemd/systemd/security/advisories/GHSA-6m6p-rjcq-334c
218 * systemd-boot gained new hotkeys to reboot and power off the system
219 from the boot menu ("B" and "O"). If the "auto-poweroff" and
220 "auto-reboot" options in loader.conf are set these entries are also
221 shown as menu items (which is useful on devices lacking a regular
224 * systemd-boot gained a new configuration value "menu-disabled" for the
225 set-timeout option, to allow completely disabling the boot menu,
226 including the hotkey.
228 * systemd-boot will now measure the content of loader.conf in TPM2
231 * systemd-stub will now concatenate the content of all kernel
232 command-line addons before measuring them in TPM2 PCR 12, in a single
233 measurement, instead of measuring them individually.
235 * systemd-stub will now measure and load Devicetree Blob addons, which
236 are searched and loaded following the same model as the existing
237 kernel command-line addons.
239 * systemd-stub will now ignore unauthenticated kernel command line options
240 passed from systemd-boot when running inside Confidential VMs with UEFI
243 * systemd-stub will now load a Devicetree blob even if the firmware did
244 not load any beforehand (e.g.: for ACPI systems).
246 * ukify is no longer considered experimental, and now ships in /usr/bin/.
248 * ukify gained a new verb inspect to describe the sections of a UKI and
249 print the contents of the well-known sections.
251 * ukify gained a new verb genkey to generate a set of key pairs for
252 signing UKIs and their PCR data.
254 * The 90-loaderentry kernel-install hook now supports installing device
257 * kernel-install now supports the --json=, --root=, --image=, and
258 --image-policy= options for the inspect verb.
260 * kernel-install now supports new list and add-all verbs. The former
261 lists all installed kernel images (if those are available in
262 /usr/lib/modules/). The latter will install all the kernels it can
267 * A new option --copy-from= has been added that synthesizes partition
268 definitions from the given image, which are then applied by the
269 systemd-repart algorithm.
271 * A new option --copy-source= has been added, which can be used to specify
272 a directory to which CopyFiles= is considered relative to.
274 * New --make-ddi=confext, --make-ddi=sysext, and --make-ddi=portable
275 options have been added to make it easier to generate these types of
276 DDIs, without having to provide repart.d definitions for them.
278 * The dm-verity salt and UUID will now be derived from the specified
281 * New VerityDataBlockSizeBytes= and VerityHashBlockSizeBytes= can now be
282 configured in repart.d/ configuration files.
284 * A new Subvolumes= setting is now supported in repart.d/ configuration
285 files, to indicate which directories in the target partition should be
288 * A new --tpm2-device-key= option can be used to lock a disk against a
289 specific TPM2 public key. This matches the same switch the
290 systemd-cryptenroll tool now supports (see above).
294 * The journalctl --lines= parameter now accepts +N to show the oldest N
295 entries instead of the newest.
297 * journald now ensures that sealing happens once per epoch, and sets a
298 new compatibility flag to distinguish old journal files that were
299 created before this change, for backward compatibility.
303 * udev will now create symlinks to loopback block devices in the
304 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
305 string field selected during allocation. The systemd-dissect tool and
306 the util-linux losetup command now supports a complementing new switch
307 --loop-ref= for selecting the string. This means a loopback block
308 device may now be allocated under a caller-chosen reference and can
309 subsequently be referenced without first having to look up the block
310 device name the caller ended up with.
312 * udev also creates symlinks to loopback block devices in the
313 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
314 of the inode attached to the loopback block device. This means that
315 attaching a file to a loopback device will implicitly make a handle
316 available to be found via that file's inode information.
318 * udevadm info gained support for JSON output via a new --json= flag, and
319 for filtering output using the same mechanism that udevadm trigger
322 * The predictable network interface naming logic is extended to include
323 the SR-IOV-R "representor" information in network interface names.
324 This feature was intended for v254, but even though the code was
325 merged, the part that actually enabled the feature was forgotten.
326 It is now enabled by default and is part of the new "v255" naming
329 * A new hwdb/rules file has been added that sets the
330 ID_NET_AUTO_LINK_LOCAL_ONLY=1 udev property on all network interfaces
331 that should usually only be configured with link-local addressing
332 (IPv4LL + IPv6LL), i.e. for PC-to-PC cables ("laplink") or
333 Thunderbolt networking. systemd-networkd and NetworkManager (soon)
334 will make use of this information to apply an appropriate network
335 configuration by default.
337 * The ID_NET_DRIVER property on network interfaces is now set
338 relatively early in the udev rule set so that other rules may rely on
339 its use. This is implemented in a new "net-driver" udev built-in.
343 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
344 is now dropped, as it never worked, hence it should not be used by
347 * The 'prefixstable' ipv6 address generation mode now considers the SSID
348 when generating stable addresses, so that a different stable address
349 is used when roaming between wireless networks. If you already use
350 'prefixstable' addresses with wireless networks, the stable address
351 will be changed by the update.
353 * The DHCPv4 client gained a RapidCommit option, true by default, which
354 enables RFC4039 Rapid Commit behavior to obtain a lease in a
355 simplified 2-message exchange instead of the typical 4-message
356 exchange, if also supported by the DHCP server.
358 * The DHCPv4 client gained new InitialCongestionWindow= and
359 InitialAdvertisedReceiveWindow= options for route configurations.
361 * The DHCPv4 client gained a new RequestAddress= option that allows
362 to send a preferred IP address in the initial DHCPDISCOVER message.
364 * The DHCPv4 server and client gained support for IPv6-only mode
367 * The SendHostname= and Hostname= options are now available for the
368 DHCPv6 client, independently of the DHCPv4= option, so that these
369 configuration values can be set independently for each client.
371 * The DHCPv4 and DHCPv6 client state can now be queried via D-Bus,
372 including lease information.
374 * The DHCPv6 client can now be configured to use a custom DUID type.
376 * .network files gained a new IPv4ReversePathFilter= setting in the
377 [Network] section, to control sysctl's rp_filter setting.
379 * .network files gaiend a new HopLimit= setting in the [Route] section,
380 to configure a per-route hop limit.
382 * .network files gained a new TCPRetransmissionTimeoutSec= setting in
383 the [Route] section, to configure a per-route TCP retransmission
386 * A new directive NFTSet= provides a method for integrating network
387 configuration into firewall rules with NFT sets. The benefit of using
388 this setting is that static network configuration or dynamically
389 obtained network addresses can be used in firewall rules with the
390 indirection of NFT set types.
392 * The [IPv6AcceptRA] section supports the following new options:
393 UsePREF64=, UseHopLimit=, UseICMP6RateLimit=, and NFTSet=.
395 * The [IPv6SendRA] section supports the following new options:
396 RetransmitSec=, HopLimit=, HomeAgent=, HomeAgentLifetimeSec=, and
397 HomeAgentPreference=.
399 * A new [IPv6PREF64Prefix] set of options, containing Prefix= and
400 LifetimeSec=, has been introduced to append pref64 options in router
401 advertisements (RFC8781).
403 * The network generator now configures the interfaces with only
404 link-local addressing if "ip=link-local" is specified on the kernel
407 * The prefix of the configuration files generated by the network
408 generator from the kernel command line is now prefixed with '70-',
409 to make them have higher precedence over the default configuration
412 * Added a new -Ddefault-network=BOOL meson option, that causes more
413 .network files to be installed as enabled by default. These configuration
414 files will which match generic setups, e.g. 89-ethernet.network matches
415 all Ethernet interfaces and enables both DHCPv4 and DHCPv6 clients.
417 * If a ID_NET_MANAGED_BY= udev property is set on a network device and
418 it is any other string than "io.systemd.Network" then networkd will
419 not manage this device. This may be used to allow multiple network
420 management services to run in parallel and assign ownership of
421 specific devices explicitly. NetworkManager will soon implement a
426 * systemctl is-failed now checks the system state if no unit is
429 * systemctl will now automatically soft-reboot if a new root file system
430 is found under /run/nextroot/ when a reboot operation is invoked.
434 * Wall messages now work even when utmp support is disabled, using
435 systemd-logind to query the necessary information.
437 * systemd-logind now sends a new PrepareForShutdownWithMetadata D-Bus
438 signal before shutdown/reboot/soft-reboot that includes additional
439 information compared to the PrepareForShutdown signal. Currently the
440 additional information is the type of operation that is about to be
443 Hibernation & Suspend:
445 * The kernel and OS versions will no longer be checked on resume from
448 * Hibernation into swap files backed by btrfs are now
449 supported. (Previously this was supported only for other file
454 * A new systemd-vmspawn tool has been added, that aims to provide for VMs
455 the same interfaces and functionality that systemd-nspawn provides for
456 containers. For now it supports QEMU as a backend, and exposes some of
457 its options to the user. This component is experimental and its public
458 interface is subject to change.
460 * "systemd-analyze plot" has gained tooltips on each unit name with
461 related-unit information in its svg output, such as Before=,
462 Requires=, and similar properties.
464 * A new varlinkctl tool has been added to allow interfacing with
465 Varlink services, and introspection has been added to all such
468 * systemd-sysext and systemd-confext now expose a Varlink service
469 at io.systemd.sysext.
471 * portable services now accept confexts as extensions.
473 * systemd-sysupdate now accepts directories in the MatchPattern= option.
475 * systemd-run will now output the invocation ID of the launched
476 transient unit and its peak memory usage.
478 * systemd-analyze, systemd-tmpfiles, systemd-sysusers, systemd-sysctl,
479 and systemd-binfmt gained a new --tldr option that can be used instead
480 of --cat-config to suppress uninteresting configuration lines, such as
481 comments and whitespace.
483 * resolvectl gained a new "show-server-state" command that shows
484 current statistics of the resolver. This is backed by a new
485 DumpStatistics() Varlink method provided by systemd-resolved.
487 * systemd-timesyncd will now emit a D-Bus signal when the LinkNTPServers
490 * vconsole now supports KEYMAP=@kernel for preserving the kernel keymap
493 * seccomp now supports the LoongArch64 architecture.
495 * seccomp may now be enabled for services running as a non-root User=
496 without NoNewPrivileges=yes.
498 * systemd-id128 now supports a new -P option to show only values. The
499 combination of -P and --app options is also supported.
501 * A new pam_systemd_loadkey.so PAM module is now available, which will
502 automatically fetch the passphrase used by cryptsetup to unlock the
503 root file system and set it as the PAM authtok. This enables, among
504 other things, configuring auto-unlock of the GNOME Keyring / KDE
505 Wallet when autologin is configured.
507 * Many meson options now use the 'feature' type, which means they
508 take enabled/disabled/auto as values.
510 * A new meson option -Dconfigfiledir= can be used to change where
511 configuration files with default values are installed to.
513 * Options and verbs in man pages are now tagged with the version they
514 were first introduced in.
516 * A new component "systemd-storagetm" has been added, which exposes all
517 local block devices as NVMe-TCP devices, fully automatically. It's
518 hooked into a new target unit storage-target-mode.target that is
519 suppsoed to be booted into via
520 rd.systemd.unit=storage-target-mode.target on the kernel command
521 line. This is intended to be used for installers and debugging to
522 quickly get access to the local disk. It's inspired by MacOS "target
525 * A new component "systemd-bsod" has been added, which can show logged
526 error messages full screen, if they have a log level of LOG_EMERG log
529 * The systemd-dissect tool's --with command will now set the
530 $SYSTEMD_DISSECT_DEVICE environment variable to the block device it
531 operates on for the invoked process.
533 * The systemd-mount tool gained a new --tmpfs switch for mounting a new
534 'tmpfs' instance. This is useful since it does so via .mount units
535 and thus can be executed remotely or in containers.
537 * The various tools in systemd that take "verbs" (such as systemctl,
538 loginctl, machinectl, …) now will suggest a close verb name in case
539 the user specified an unrecognized one.
541 * libsystemd now exports a new function sd_id128_get_app_specific()
542 that generates "app-specific" 128bit IDs from any ID. It's similar to
543 sd_id128_get_machine_app_specific() and
544 sd_id128_get_boot_app_specific() but takes the ID to base calculation
545 on as input. This new functionality is also exposed in the
546 "systemd-id128" tool where you can now combine --app= with `show`.
548 * All tools that parse timestamps now can also parse RFC3339 style
549 timestamps that include the "T" and Z" characters.
551 * New documentation has been added:
553 https://systemd.io/FILE_DESCRIPTOR_STORE
554 https://systemd.io/TPM2_PCR_MEASUREMENTS
555 https://systemd.io/MOUNT_REQUIREMENTS
557 * The codebase now recognizes the suffix .confext.raw and .sysext.raw
558 as alternative to the .raw suffix generally accepted for DDIs. It is
559 recommended to name configuration extensions and system extensions
560 with such suffixes, to indicate their purpose in the name.
562 * The sd-device API gained a new function
563 sd_device_enumerator_add_match_property_required() which allows
564 configuring matches on properties that are strictly required. This is
565 different from the existing sd_device_enumerator_add_match_property()
566 matches of which one one needs to apply.
568 * The MAC address the veth side of an nspawn container shall get
569 assigned may now be controlled via the $SYSTEMD_NSPAWN_NETWORK_MAC
570 environment variable.
572 * The libiptc dependency is now implemented via dlopen(), so that tools
573 such as networkd and nspawn no longer have a hard dependency on the
574 shared library when compiled with support for libiptc.
576 * New rpm macros have been added: %systemd_user_daemon_reexec does
577 daemon-reexec for all user managers, and %systemd_postun_with_reload
578 and %systemd_user_postun_with_reload do a reload for system and user
581 * coredumpctl now propagates SIGTERM to the debugger process.
583 Contributions from: 김인수, Abderrahim Kitouni, Adam Goldman,
584 Adam Williamson, Alexandre Peixoto Ferreira, Alex Hudspith,
585 Alvin Alvarado, André Paiusco, Antonio Alvarez Feijoo,
586 Anton Lundin, Arian van Putten, Arseny Maslennikov, Arthur Shau,
587 Balázs Úr, beh_10257, Benjamin Peterson, Bertrand Jacquin,
588 Brian Norris, Charles Lee, Cheng-Chia Tseng, Chris Patterson,
589 Christian Hergert, Christian Hesse, Christian Kirbach,
590 Clayton Craft, commondservice, cunshunxia, Curtis Klein, cvlc12,
591 Daan De Meyer, Daniele Medri, Daniel P. Berrangé, Daniel Rusek,
592 Daniel Thompson, Dan Nicholson, Dan Streetman, David Rheinsberg,
593 David Santamaría Rogado, David Tardon, dependabot[bot],
594 Diego Viola, Dmitry V. Levin, Emanuele Giuseppe Esposito,
595 Emil Renner Berthing, Emil Velikov, Etienne Dechamps, Fabian Vogt,
596 felixdoerre, Felix Dörre, Florian Schmaus, Franck Bui,
597 Frantisek Sumsal, G2-Games, Gioele Barabucci, Hugo Carvalho,
598 huyubiao, Iago López Galeiras, IllusionMan1212, Jade Lovelace,
599 janana, Jan Janssen, Jan Kuparinen, Jan Macku, Jeremy Fleischman,
600 Jin Liu, jjimbo137, Joerg Behrmann, Johannes Segitz, Jordan Rome,
601 Jordan Williams, Julien Malka, Juno Computers, Khem Raj, khm,
602 Kingbom Dou, Kiran Vemula, Krzesimir Nowak, Laszlo Gombos,
603 Lennart Poettering, linuxlion, Luca Boccassi, Lucas Adriano Salles,
604 Lukas, Lukáš Nykrýn, Maanya Goenka, Maarten, Malte Poll,
605 Marc Pervaz Boocha, Martin Beneš, Martin Joerg, Martin Wilck,
606 Mathieu Tortuyaux, Matthias Schiffer, Maxim Mikityanskiy,
607 Max Kellermann, Michael A Cassaniti, Michael Biebl, Michael Kuhn,
608 Michael Vasseur, Michal Koutný, Michal Sekletár, Mike Yuan,
609 Milton D. Miller II, mordner, msizanoen, NAHO, Nandakumar Raghavan,
610 Neil Wilson, Nick Rosbrook, Nils K, NRK, Oğuz Ersen,
611 Omojola Joshua, onenowy, Paul Meyer, Paymon MARANDI, pelaufer,
612 Peter Hutterer, PhylLu, Pierre GRASSER, Piotr Drąg, Priit Laes,
613 Rahil Bhimjiani, Raito Bezarius, Raul Cheleguini, Reto Schneider,
614 Richard Maw, Robby Red, RoepLuke, Roland Hieber, Ronan Pigott,
615 Sam James, Sam Leonard, Sergey A, Susant Sahani, Sven Joachim,
616 Tad Fisher, Takashi Sakamoto, Thorsten Kukuk, Tj, Tomasz Świątek,
617 Topi Miettinen, Valentin David, Valentin Lefebvre,
618 Victor Westerhuis, Vincent Haupert, Vishal Chillara Srinivas,
619 Vito Caputo, Warren, Weblate, Xiaotian Wu, xinpeng wang,
620 Yaron Shahrabani, Yo-Jung Lin, Yu Watanabe,
621 Zbigniew Jędrzejewski-Szmek, zeroskyx,
622 Дамјан Георгиевски, наб
624 — Edinburgh, 2023-12-02
628 Announcements of Future Feature Removals and Incompatible Changes:
630 * The next release (v255) will remove support for split-usr (/usr/
631 mounted separately during late boot, instead of being mounted by the
632 initrd before switching to the rootfs) and unmerged-usr (parallel
633 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
635 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
637 * We intend to remove cgroup v1 support from a systemd release after
638 the end of 2023. If you run services that make explicit use of
639 cgroup v1 features (i.e. the "legacy hierarchy" with separate
640 hierarchies for each controller), please implement compatibility with
641 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
642 Most of Linux userspace has been ported over already.
644 * Support for System V service scripts is now deprecated and will be
645 removed in a future release. Please make sure to update your software
646 *now* to include a native systemd unit file instead of a legacy
647 System V script to retain compatibility with future systemd releases.
649 * Support for the SystemdOptions EFI variable is deprecated.
650 'bootctl systemd-efi-options' will emit a warning when used. It seems
651 that this feature is little-used and it is better to use alternative
652 approaches like credentials and confexts. The plan is to drop support
653 altogether at a later point, but this might be revisited based on
656 * EnvironmentFile= now treats the line following a comment line
657 trailing with escape as a non comment line. For details, see:
658 https://github.com/systemd/systemd/issues/27975
660 * PrivateNetwork=yes and NetworkNamespacePath= now imply
661 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
663 * Behaviour of sandboxing options for the per-user service manager
664 units has changed. They now imply PrivateUsers=yes, which means user
665 namespaces will be implicitly enabled when a sandboxing option is
666 enabled in a user unit. Enabling user namespaces has the drawback
667 that system users will no longer be visible (and processes/files will
668 appear as owned by 'nobody') in the user unit.
670 By definition a sandboxed user unit should run with reduced
671 privileges, so impact should be small. This will remove a great
672 source of confusion that has been reported by users over the years,
673 due to how these options require an extra setting to be manually
674 enabled when used in the per-user service manager, which is not
675 needed in the system service manager. For more details, see:
676 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
678 * systemd-run's switch --expand-environment= which currently is disabled
679 by default when combined with --scope, will be changed in a future
680 release to be enabled by default.
682 Security Relevant Changes:
684 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
685 process capability to invoked session processes of regular users on
686 local seats (as well as to systemd --user), unless configured
687 otherwise via data from JSON user records, or via the PAM module's
688 parameter list. This is useful in order allow desktop tools such as
689 GNOME's Alarm Clock application to set a timer for
690 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
691 per-user service unit file may thus use AmbientCapability= to pass
692 the capability to invoked processes. Note that this capability is
693 relatively narrow in focus (in particular compared to other process
694 capabilities such as CAP_SYS_ADMIN) and we already — by default —
695 permit more impactful operations such as system suspend to local
700 * Memory limits that apply while the unit is activating are now
701 supported. Previously IO and CPU settings were already supported via
702 StartupCPUWeight= and similar. The same logic has been added for the
703 various manager and unit memory settings (DefaultStartupMemoryLow=,
704 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
705 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
707 * The service manager gained support for enqueuing POSIX signals to
708 services that carry an additional integer value, exposing the
709 sigqueue() system call. This is accessible via new D-Bus calls
710 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
711 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
712 via the new --kill-value= option.
714 * systemctl gained a new "list-paths" verb, which shows all currently
715 active .path units, similarly to how "systemctl list-timers" shows
716 active timers, and "systemctl list-sockets" shows active sockets.
718 * systemctl gained a new --when= switch which is honoured by the various
719 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
720 scheduling these operations by time, similar in fashion to how this
721 has been supported by SysV shutdown.
723 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
724 supports the new PR_SET_MDWE prctl() call, it is used instead of the
725 seccomp()-based system call filter to achieve the same effect.
727 * A new set of kernel command line options is now understood:
728 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
729 systemd.tty.columns.<name>= allow configuring the TTY type and
730 dimensions for the tty specified via <name>. When systemd invokes a
731 service on a tty (via TTYName=) it will look for these and configure
732 the TTY accordingly. This is particularly useful in VM environments
733 to propagate host terminal settings into the appropriate TTYs of the
736 * A new RootEphemeral= setting is now understood in service units. It
737 takes a boolean argument. If enabled for services that use RootImage=
738 or RootDirectory= an ephemeral copy of the disk image or directory
739 tree is made when the service is started. It is removed automatically
740 when the service is stopped. That ephemeral copy is made using
741 btrfs/xfs reflinks or btrfs snapshots, if available.
743 * The service activation logic gained new settings RestartSteps= and
744 RestartMaxDelaySec= which allow exponentially-growing restart
745 intervals for Restart=.
747 * The service activation logic gained a new setting RestartMode= which
748 can be set to 'direct' to skip the inactive/failed states when
749 restarting, so that dependent units are not notified until the service
750 converges to a final (successful or failed) state. For example, this
751 means that OnSuccess=/OnFailure= units will not be triggered until the
752 service state has converged.
754 * PID 1 will now automatically load the virtio_console kernel module
755 during early initialization if running in a suitable VM. This is done
756 so that early-boot logging can be written to the console if available.
758 * Similarly, virtio-vsock support is loaded early in suitable VM
759 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
760 to the VMM if configured, thus loading this early is beneficial.
762 * A new verb "fdstore" has been added to systemd-analyze to show the
763 current contents of the file descriptor store of a unit. This is
764 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
767 * The service manager will now set a new $FDSTORE environment variable
768 when invoking processes for services that have the file descriptor
771 * A new service option FileDescriptorStorePreserve= has been added that
772 allows tuning the lifecycle of the per-service file descriptor store.
773 If set to "yes", the entries in the fd store are retained even after
774 the service has been fully stopped.
776 * The "systemctl clean" command may now be used to clear the fdstore of
779 * Unit *.preset files gained a new directive "ignore", in addition to
780 the existing "enable" and "disable". As the name suggests, matching
781 units are left unchanged, i.e. neither enabled nor disabled.
783 * Service units gained a new setting DelegateSubgroup=. It takes the
784 name of a sub-cgroup to place any processes the service manager forks
785 off in. Previously, the service manager would place all service
786 processes directly in the top-level cgroup it created for the
787 service. This usually meant that main process in a service with
788 delegation enabled would first have to create a subgroup and move
789 itself down into it, in order to not conflict with the "no processes
790 in inner cgroups" rule of cgroup v2. With this option, this step is
791 now handled by PID 1.
793 * The service manager will now look for .upholds/ directories,
794 similarly to the existing support for .wants/ and .requires/
795 directories. Symlinks in this directory result in Upholds=
798 The [Install] section of unit files gained support for a new
799 UpheldBy= directive to generate .upholds/ symlinks automatically when
802 * The service manager now supports a new kernel command line option
803 systemd.default_device_timeout_sec=, which may be used to override
804 the default timeout for .device units.
806 * A new "soft-reboot" mechanism has been added to the service manager.
807 A "soft reboot" is similar to a regular reboot, except that it
808 affects userspace only: the service manager shuts down any running
809 services and other units, then optionally switches into a new root
810 file system (mounted to /run/nextroot/), and then passes control to a
811 systemd instance in the new file system which then starts the system
812 up again. The kernel is not rebooted and neither is the hardware,
813 firmware or boot loader. This provides a fast, lightweight mechanism
814 to quickly reset or update userspace, without the latency that a full
815 system reset involves. Moreover, open file descriptors may be passed
816 across the soft reboot into the new system where they will be passed
817 back to the originating services. This allows pinning resources
818 across the reboot, thus minimizing grey-out time further. This new
819 reboot mechanism is accessible via the new "systemctl soft-reboot"
822 * Services using RootDirectory= or RootImage= will now have read-only
823 access to a copy of the host's os-release file under
824 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
825 This was already the case for Portable Services, and the feature has
826 now been extended to all services that do not run off the host's
829 * A new service setting MemoryKSM= has been added to enable kernel
830 same-page merging individually for services.
832 * A new service setting ImportCredentials= has been added that augments
833 LoadCredential= and LoadCredentialEncrypted= and searches for
834 credentials to import from the system, and supports globbing.
836 * A new job mode "restart-dependencies" has been added to the service
837 manager (exposed via systemctl --job-mode=). It is only valid when
838 used with "start" jobs, and has the effect that the "start" job will
839 be propagated as "restart" jobs to currently running units that have
840 a BindsTo= or Requires= dependency on the started unit.
842 * A new verb "whoami" has been added to "systemctl" which determines as
843 part of which unit the command is being invoked. It writes the unit
844 name to standard output. If one or more PIDs are specified reports
845 the unit names the processes referenced by the PIDs belong to.
847 * The system and service credential logic has been improved: there's
848 now a clearly defined place where system provisioning tools running
849 in the initrd can place credentials that will be imported into the
850 system's set of credentials during the initrd → host transition: the
851 /run/credentials/@initrd/ directory. Once the credentials placed
852 there are imported into the system credential set they are deleted
853 from this directory, and the directory itself is deleted afterwards
856 * A new kernel command line option systemd.set_credential_binary= has
857 been added, that is similar to the pre-existing
858 systemd.set_credential= but accepts arbitrary binary credential data,
859 encoded in Base64. Note that the kernel command line is not a
860 recommend way to transfer credentials into a system, since it is
861 world-readable from userspace.
863 * The default machine ID to use may now be configured via the
864 system.machine_id system credential. It will only be used if no
865 machine ID was set yet on the host.
867 * On Linux kernel 6.4 and newer system and service credentials will now
868 be placed in a tmpfs instance that has the "noswap" mount option
869 set. Previously, a "ramfs" instance was used. By switching to tmpfs
870 ACL support and overall size limits can now be enforced, without
871 compromising on security, as the memory is never paged out either
874 * The service manager now can detect when it is running in a
875 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
876 accepted by ConditionSecurity= for units that want to conditionalize
877 themselves on this. systemd-detect-virt gained new 'cvm' and
878 '--list-cvm' switches to respectively perform the detection or list
879 all known flavours of confidential VM, depending on the vendor. The
880 manager will publish a 'ConfidentialVirtualization' D-Bus property,
881 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
882 variable for unit generators. Finally, udev rules can match on a new
883 'cvm' key that will be set when in a confidential VM.
884 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
885 strings and QEMU's fw_cfg protocol will not be used to import
886 credentials and kernel command line parameters by the system manager,
887 systemd-boot and systemd-stub, because the hypervisor is considered
888 untrusted in this particular setting.
892 * The sd-journal API gained a new call sd_journal_get_seqnum() to
893 retrieve the current log record's sequence number and sequence number
894 ID, which allows applications to order records the same way as
895 journal does internally. The sequence number is now also exported in
896 the JSON and "export" output of the journal.
898 * journalctl gained a new switch --truncate-newline. If specified
899 multi-line log records will be truncated at the first newline,
900 i.e. only the first line of each log message will be shown.
902 * systemd-journal-upload gained support for --namespace=, similar to
903 the switch of the same name of journalctl.
907 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
908 may be used to exclude certain files from the effect of CopyFiles=.
910 * systemd-repart's Verity support now implements the Minimize= setting
911 to minimize the size of the resulting partition.
913 * systemd-repart gained a new --offline= switch, which may be used to
914 control whether images shall be built "online" or "offline",
915 i.e. whether to make use of kernel facilities such as loopback block
916 devices and device mapper or not.
918 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
919 partition with some files, it will now default to VFAT rather than
922 * systemd-repart gained a new --architecture= switch. If specified, the
923 per-architecture GPT partition types (i.e. the root and /usr/
924 partitions) configured in the partition drop-in files are
925 automatically adjusted to match the specified CPU architecture, in
926 order to simplify cross-architecture DDI building.
928 * systemd-repart will now default to a minimum size of 300MB for XFS
929 filesystems if no size parameter is specified. This matches what the
930 XFS tools (xfsprogs) can support.
932 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
934 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
935 Instead, pyelftools is now needed, and it will be used to perform the
936 ELF -> PE relocations at build time.
938 * bootctl gained a new switch --print-root-device/-R that prints the
939 block device the root file system is backed by. If specified twice,
940 it returns the whole disk block device (as opposed to partition block
941 device) the root file system is on. It's useful for invocations such
942 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
945 * systemd-stub will now look for the SMBIOS Type 1 field
946 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
947 kernel command line it invokes. This is useful for VMMs such as qemu
948 to pass additional kernel command lines into the system even when
949 booting via full UEFI. The contents of the field are measured into
952 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
953 value "auto". With this value, a kernel will be automatically
954 analyzed, and if it qualifies as UKI, it will be installed as if the
955 setting was to set to "uki", otherwise as "bls".
957 * systemd-stub can now optionally load UEFI PE "add-on" images that may
958 contain additional kernel command line information. These "add-ons"
959 superficially look like a regular UEFI executable, and are expected
960 to be signed via SecureBoot/shim. However, they do not actually
961 contain code, but instead a subset of the PE sections that UKIs
962 support. They are supposed to provide a way to extend UKIs with
963 additional resources in a secure and authenticated way. Currently,
964 only the .cmdline PE section may be used in add-ons, in which case
965 any specified string is appended to the command line embedded into
966 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
967 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
968 the future we expect other sections to be made extensible like this as
971 * ukify has been updated to allow building these UEFI PE "add-on"
972 images, using the new 'addon<EFI-ARCH>.efi.stub'.
974 * ukify now accepts SBAT information to place in the .sbat PE section
975 of UKIs and addons. If a UKI is built the SBAT information from the
976 inner kernel is merged with any SBAT information associated with
977 systemd-stub and the SBAT data specified on the ukify command line.
979 * The kernel-install script has been rewritten in C, and reuses much of
980 the infrastructure of existing tools such as bootctl. It also gained
981 --esp-path= and --boot-path= options to override the path to the ESP,
982 and the $BOOT partition. Options --make-entry-directory= and
983 --entry-token= have been added as well, similar to bootctl's options
986 * A new kernel-install plugin 60-ukify has been added which will
987 combine kernel/initrd locally into a UKI and optionally sign them
988 with a local key. This may be used to switch to UKI mode even on
989 systems where a local kernel or initrd is used. (Typically UKIs are
990 built and signed by the vendor.)
992 * The ukify tool now supports "pesign" in addition to the pre-existing
993 "sbsign" for signing UKIs.
995 * systemd-measure and systemd-stub now look for the .uname PE section
996 that should contain the kernel's "uname -r" string.
998 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
999 "offline", i.e. without access to a TPM (physical or
1002 Memory Pressure & Control:
1004 * The sd-event API gained new calls sd_event_add_memory_pressure(),
1005 sd_event_source_set_memory_pressure_type(),
1006 sd_event_source_set_memory_pressure_period() to create and configure
1007 an event source that is called whenever the OS signals memory
1008 pressure. Another call sd_event_trim_memory() is provided that
1009 compacts the process' memory use by releasing allocated but unused
1010 malloc() memory back to the kernel. Services can also provide their
1011 own custom callback to do memory trimming. This should improve system
1012 behaviour under memory pressure, as on Linux traditionally provided
1013 no mechanism to return process memory back to the kernel if the
1014 kernel was under memory pressure. This makes use of the kernel's PSI
1015 interface. Most long-running services in systemd have been hooked up
1016 with this, and in particular systems with low memory should benefit
1019 * Service units gained new settings MemoryPressureWatch= and
1020 MemoryPressureThresholdSec= to configure the PSI memory pressure
1021 logic individually. If these options are used, the
1022 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
1023 variables will be set for the invoked processes to inform them about
1024 the requested memory pressure behaviour. (This is used by the
1025 aforementioned sd-events API additions, if set.)
1027 * systemd-analyze gained a new "malloc" verb that shows the output
1028 generated by glibc's malloc_info() on services that support it. Right
1029 now, only the service manager has been updated accordingly. This
1030 call requires privileges.
1032 User & Session Management:
1034 * The sd-login API gained a new call sd_session_get_username() to
1035 return the user name of the owner of a login session. It also gained
1036 a new call sd_session_get_start_time() to retrieve the time the login
1037 session started. A new call sd_session_get_leader() has been added to
1038 return the PID of the "leader" process of a session. A new call
1039 sd_uid_get_login_time() returns the time since the specified user has
1040 most recently been continuously logged in with at least one session.
1042 * JSON user records gained a new set of fields capabilityAmbientSet and
1043 capabilityBoundingSet which contain a list of POSIX capabilities to
1044 set for the logged in users in the ambient and bounding sets,
1045 respectively. homectl gained the ability to configure these two sets
1046 for users via --capability-bounding-set=/--capability-ambient-set=.
1048 * pam_systemd learnt two new module options
1049 default-capability-bounding-set= and default-capability-ambient-set=,
1050 which configure the default bounding sets for users as they are
1051 logging in, if the JSON user record doesn't specify this explicitly
1052 (see above). The built-in default for the ambient set now contains
1053 the CAP_WAKE_ALARM, thus allowing regular users who may log in
1054 locally to resume from a system suspend via a timer.
1056 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
1057 call to update the TTY of a session after it has been allocated. This
1058 is useful for SSH sessions which are typically allocated first, and
1059 for which a TTY is added later.
1061 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
1062 combines the various other sd_pid_notify() flavours into one: takes a
1063 format string, an overriding PID, and a set of file descriptors to
1064 send. It also gained a new call sd_pid_notify_barrier() call which is
1065 equivalent to sd_notify_barrier() but allows the originating PID to
1068 * "loginctl list-users" and "loginctl list-sessions" will now show the
1069 state of each logged in user/session in their tabular output. It will
1070 also show the current idle state of sessions.
1074 * systemd-dissect will now show the intended CPU architecture of an
1077 * systemd-dissect will now install itself as mount helper for the "ddi"
1078 pseudo-file system type. This means you may now mount DDIs directly
1079 via /bin/mount or /etc/fstab, making full use of embedded Verity
1080 information and all other DDI features.
1082 Example: mount -t ddi myimage.raw /some/where
1084 * The systemd-dissect tool gained the new switches --attach/--detach to
1085 attach/detach a DDI to a loopback block device without mounting it.
1086 It will automatically derive the right sector size from the image
1087 and set up Verity and similar, but not mount the file systems in it.
1089 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
1090 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
1091 implied. Given that these file systems are typically untrusted, this
1092 should make mounting them automatically have less of a security
1095 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
1096 systemd-tmpfiles, …) now understand a new switch --image-policy= which
1097 takes a string encoding image dissection policy. With this mechanism
1098 automatic discovery and use of specific partition types and the
1099 cryptographic requirements on the partitions (Verity, LUKS, …) can be
1100 restricted, permitting better control of the exposed attack surfaces
1101 when mounting disk images. systemd-gpt-auto-generator will honour such
1102 an image policy too, configurable via the systemd.image_policy= kernel
1103 command line option. Unit files gained the RootImagePolicy=,
1104 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
1105 disk images a service runs off.
1107 * systemd-analyze gained a new verb "image-policy" to validate and
1108 parse image policy strings.
1110 * systemd-dissect gained support for a new --validate switch to
1111 superficially validate DDI structure, and check whether a specific
1112 image policy allows the DDI.
1114 * systemd-dissect gained support for a new --mtree-hash switch to
1115 optionally disable calculating mtree hashes, which can be slow on
1118 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
1119 are now able to operate on directories too, other than images.
1123 * networkd's GENEVE support as gained a new .network option
1124 InheritInnerProtocol=.
1126 * The [Tunnel] section in .netdev files has gained a new setting
1127 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
1129 * A new global IPv6PrivacyExtensions= setting has been added that
1130 selects the default value of the per-network setting of the same
1133 * The predictable network interface naming logic was extended to
1134 include SR-IOV-R "representor" information in network interface
1135 names. Unfortunately, this feature was not enabled by default and can
1136 only be enabled at compilation time by setting
1137 -Ddefault-net-naming-scheme=v254.
1139 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
1140 the RFC8910 captive portal option.
1144 * udevadm gained the new "verify" verb for validating udev rules files
1147 * udev gained a new tool "iocost" that can be used to configure QoS IO
1148 cost data based on hwdb information onto suitable block devices. Also
1149 see https://github.com/iocost-benchmark/iocost-benchmarks.
1151 TPM2 Support + Disk Encryption & Authentication:
1153 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
1154 ("Storage Root Key") as first step in the TPM2, and then use that
1155 for binding FDE to, if TPM2 support is used. This matches
1156 recommendations of TCG (see
1157 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
1159 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
1160 understand textual identifiers for these PCRs.
1162 * systemd-veritysetup + /etc/veritytab gained support for a series of
1163 new options: hash-offset=, superblock=, format=, data-block-size=,
1164 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
1165 fec-offset=, fec-roots= to configure various aspects of a Verity
1168 * systemd-cryptsetup + /etc/crypttab gained support for a new
1169 veracrypt-pim= option for setting the Personal Iteration Multiplier
1170 of veracrypt volumes.
1172 * systemd-integritysetup + /etc/integritytab gained support for a new
1173 mode= setting for controlling the dm-integrity mode (journal, bitmap,
1174 direct) for the volume.
1176 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
1177 registers, their symbolic names and current values.
1181 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
1182 access right is specified this is equivalent to "x" but only if the
1183 inode in question already has the executable bit set for at least
1184 some user/group. Otherwise the "x" bit will be turned off.
1186 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
1187 with C+ will result in a "merge" copy, i.e. all files of the source
1188 tree are copied into the target tree, even if that tree already
1189 exists, resulting in a combined tree of files already present in the
1190 target tree and those copied in.
1192 * systemd-tmpfiles gained a new --graceful switch. If specified lines
1193 with unknown users/groups will silently be skipped.
1197 * systemd-notify gained two new options --fd= and --fdname= for sending
1198 arbitrary file descriptors to the service manager (while specifying an
1199 explicit name for it).
1201 * systemd-notify gained a new --exec switch, which makes it execute the
1202 specified command line after sending the requested messages. This is
1203 useful for sending out READY=1 first, and then continuing invocation
1204 without changing process ID, so that the tool can be nicely used
1205 within an ExecStart= line of a unit file that uses Type=notify.
1207 sd-event + sd-bus APIs:
1209 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
1210 which may be used to explicitly end a rate-limit state an event
1211 source might be in, resetting all rate limiting counters.
1213 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
1214 sockets, it will now encode the "description" set via
1215 sd_bus_set_description() into the source socket address. It will also
1216 look for this information when accepting a connection. This is useful
1217 to track individual D-Bus connections on a D-Bus broker for debug
1222 * systemd-resolved gained a new resolved.conf setting
1223 StateRetentionSec= which may be used to retain cached DNS records
1224 even after their nominal TTL, and use them in case upstream DNS
1225 servers cannot be reached. This can be used to make name resolution
1226 more resilient in case of network problems.
1228 * resolvectl gained a new verb "show-cache" to show the current cache
1229 contents of systemd-resolved. This verb communicates with the
1230 systemd-resolved daemon and requires privileges.
1234 * Meson >= 0.60.0 is now required to build systemd.
1236 * The default keymap to apply may now be chosen at build-time via the
1237 new -Ddefault-keymap= meson option.
1239 * Most of systemd's long-running services now have a generic handler of
1240 the SIGRTMIN+18 signal handler which executes various operations
1241 depending on the sigqueue() parameter sent along. For example, values
1242 0x100…0x107 allow changing the maximum log level of such
1243 services. 0x200…0x203 allow changing the log target of such
1244 services. 0x300 make the services trim their memory similarly to the
1245 automatic PSI-triggered action, see above. 0x301 make the services
1246 output their malloc_info() data to the logs.
1248 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
1249 files, inspired by systemctl's verbs of the same name which edit unit
1250 files. Similarly, networkctl gained the same verbs for editing
1251 .network, .netdev, .link files.
1253 * A new syscall filter group "@sandbox" has been added that contains
1254 syscalls for sandboxing system calls such as those for seccomp and
1257 * New documentation has been added:
1259 https://systemd.io/COREDUMP
1260 https://systemd.io/MEMORY_PRESSURE
1263 * systemd-firstboot gained a new --reset option. If specified, the
1264 settings in /etc/ it knows how to initialize are reset.
1266 * systemd-sysext is now a multi-call binary and is also installed under
1267 the systemd-confext alias name (via a symlink). When invoked that way
1268 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
1269 powerful, atomic, secure configuration management of sorts, that
1270 locally can merge configuration from multiple confext configuration
1271 images into a single immutable tree.
1273 * The --network-macvlan=, --network-ipvlan=, --network-interface=
1274 switches of systemd-nspawn may now optionally take the intended
1275 network interface inside the container.
1277 * All our programs will now send an sd_notify() message with their exit
1278 status in the EXIT_STATUS= field when exiting, using the usual
1279 protocol, including PID 1. This is useful for VMMs and container
1280 managers to collect an exit status from a system as it shuts down, as
1281 set via "systemctl exit …". This is particularly useful in test cases
1282 and similar, as invocations via a VM can now nicely propagate an exit
1283 status to the host, similar to local processes.
1285 * systemd-run gained a new switch --expand-environment=no to disable
1286 server-side environment variable expansion in specified command
1287 lines. Expansion defaults to enabled for all execution types except
1288 --scope, where it defaults to off (and prints a warning) for backward
1289 compatibility reasons. --scope will be flipped to enabled by default
1290 too in a future release. If you are using --scope and passing a '$'
1291 character in the payload you should start explicitly using
1292 --expand-environment=yes/no according to the use case.
1294 * The systemd-system-update-generator has been updated to also look for
1295 the special flag file /etc/system-update in addition to the existing
1296 support for /system-update to decide whether to enter system update
1299 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
1300 mount options by default.
1302 * systemd-fstab-generator now understands two new kernel command line
1303 options systemd.mount-extra= and systemd.swap-extra=, which configure
1304 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
1305 will be ran on these block devices, like it already happens for
1306 'root='. It also now supports the new fstab.extra and
1307 fstab.extra.initrd credentials that may contain additional /etc/fstab
1308 lines to apply at boot.
1310 * systemd-getty-generator now understands two new credentials
1311 getty.ttys.container and getty.ttys.serial. These credentials may
1312 contain a list of TTY devices – one per line – to instantiate
1313 container-getty@.service and serial-getty@.service on.
1315 * The getty/serial-getty/container-getty units now import the 'agetty.*'
1316 and 'login.*' credentials, which are consumed by the 'login' and
1317 'agetty' programs starting from util-linux v2.40.
1319 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
1320 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
1321 which case the Path= setting is taken relative to the ESP or XBOOTLDR
1322 partitions, rather than the system's root directory /. The relevant
1323 directories are automatically discovered.
1325 * The systemd-ac-power tool gained a new switch --low, which reports
1326 whether the battery charge is considered "low", similar to how the
1327 s2h suspend logic checks this state to decide whether to enter system
1328 suspend or hibernation.
1330 * The /etc/os-release file can now have two new optional fields
1331 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
1334 * When the system hibernates, information about the device and offset
1335 used is now written to a non-volatile EFI variable. On next boot the
1336 system will attempt to resume from the location indicated in this EFI
1337 variable. This should make hibernation a lot more robust, while
1338 requiring no manual configuration of the resume location.
1340 * The $XDG_STATE_HOME environment variable (added in more recent
1341 versions of the XDG basedir specification) is now honoured to
1342 implement the StateDirectory= setting in user services.
1344 * A new component "systemd-battery-check" has been added. It may run
1345 during early boot (usually in the initrd), and checks the battery
1346 charge level of the system. In case the charge level is very low the
1347 user is notified (graphically via Plymouth – if available – as well
1348 as in text form on the console), and the system is turned off after a
1349 10s delay. The feature can be disabled by passing
1350 systemd.battery-check=0 through the kernel command line.
1352 * The 'passwdqc' library is now supported as an alternative to the
1353 'pwquality' library and can be selected at build time.
1355 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
1356 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
1357 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
1358 Arian van Putten, Arthur Shau, A S Alam,
1359 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
1360 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
1361 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
1362 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
1363 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
1364 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
1365 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
1366 David Schroeder, David Tardon, dependabot[bot],
1367 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
1368 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
1369 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
1370 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
1371 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
1372 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
1373 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
1374 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
1375 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
1376 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
1377 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
1378 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
1379 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
1380 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
1381 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
1382 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
1383 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
1384 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
1385 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
1386 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
1387 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
1388 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
1389 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
1390 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
1391 saikat0511, Samanta Navarro, Sam James, Sam Morris,
1392 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
1393 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
1394 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
1395 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
1396 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
1397 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
1398 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
1399 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
1400 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
1401 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
1402 Дамјан Георгиевски, наб
1404 — Edinburgh, 2023-07-28
1408 Announcements of Future Feature Removals and Incompatible Changes:
1410 * We intend to remove cgroup v1 support from systemd release after the
1411 end of 2023. If you run services that make explicit use of cgroup v1
1412 features (i.e. the "legacy hierarchy" with separate hierarchies for
1413 each controller), please implement compatibility with cgroup v2 (i.e.
1414 the "unified hierarchy") sooner rather than later. Most of Linux
1415 userspace has been ported over already.
1417 * We intend to remove support for split-usr (/usr mounted separately
1418 during boot) and unmerged-usr (parallel directories /bin and
1419 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
1420 half of 2023, in the first release that falls into that time window.
1421 For more details, see:
1422 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1424 * We intend to change behaviour w.r.t. units of the per-user service
1425 manager and sandboxing options, so that they work without having to
1426 manually enable PrivateUsers= as well, which is not required for
1427 system units. To make this work, we will implicitly enable user
1428 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
1429 user unit. The drawback is that system users will no longer be visible
1430 (and appear as 'nobody') to the user unit when a sandboxing option is
1431 enabled. By definition a sandboxed user unit should run with reduced
1432 privileges, so impact should be small. This will remove a great source
1433 of confusion that has been reported by users over the years, due to
1434 how these options require an extra setting to be manually enabled when
1435 used in the per-user service manager, as opposed as to the system
1436 service manager. We plan to enable this change in the next release
1437 later this year. For more details, see:
1438 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
1440 Deprecations and incompatible changes:
1442 * systemctl will now warn when invoked without /proc/ mounted
1443 (e.g. when invoked after chroot() into an directory tree without the
1444 API mount points like /proc/ being set up.) Operation in such an
1445 environment is not fully supported.
1447 * The return value of 'systemctl is-active|is-enabled|is-failed' for
1448 unknown units is changed: previously 1 or 3 were returned, but now 4
1449 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
1451 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
1452 systemd-hwdb (added in 2014) should be used instead.
1454 * 'bootctl --json' now outputs a single JSON array, instead of a stream
1455 of newline-separated JSON objects.
1457 * Udev rules in 60-evdev.rules have been changed to load hwdb
1458 properties for all modalias patterns. Previously only the first
1459 matching pattern was used. This could change what properties are
1460 assigned if the user has more and less specific patterns that could
1461 match the same device, but it is expected that the change will have
1462 no effect for most users.
1464 * systemd-networkd-wait-online exits successfully when all interfaces
1465 are ready or unmanaged. Previously, if neither '--any' nor
1466 '--interface=' options were used, at least one interface had to be in
1467 configured state. This change allows the case where systemd-networkd
1468 is enabled, but no interfaces are configured, to be handled
1469 gracefully. It may occur in particular when a different network
1470 manager is also enabled and used.
1472 * Some compatibility helpers were dropped: EmergencyAction= in the user
1473 manager, as well as measuring kernel command line into PCR 8 in
1474 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
1477 * The '-Dupdate-helper-user-timeout=' build-time option has been
1478 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
1479 integer as parameter instead of a string.
1481 * The DDI image dissection logic (which backs RootImage= in service
1482 unit files, the --image= switch in various tools such as
1483 systemd-nspawn, as well as systemd-dissect) will now only mount file
1484 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
1485 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
1486 variable. These file systems are fairly well supported and maintained
1487 in current kernels, while others are usually more niche, exotic or
1488 legacy and thus typically do not receive the same level of security
1491 * The default per-link multicast DNS mode is changed to "yes"
1492 (that was previously "no"). As the default global multicast DNS mode
1493 has been "yes" (but can be changed by the build option), now the
1494 multicast DNS is enabled on all links by default. You can disable the
1495 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
1496 or on an interface by calling "resolvectl mdns INTERFACE no".
1500 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
1501 (UKIs) has been added. This replaces functionality provided by
1502 'dracut --uefi' and extends it with automatic calculation of PE file
1503 offsets, insertion of signed PCR policies generated by
1504 systemd-measure, support for initrd concatenation, signing of the
1505 embedded Linux image and the combined image with sbsign, and
1506 heuristics to autodetect the kernel uname and verify the splash
1509 Changes in systemd and units:
1511 * A new service type Type=notify-reload is defined. When such a unit is
1512 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
1513 service process. The manager will then wait until it receives a
1514 "RELOADING=1" followed by a "READY=1" notification from the unit as
1515 response (via sd_notify()). Otherwise, this type is the same as
1516 Type=notify. A new setting ReloadSignal= may be used to change the
1517 signal to send from the default of SIGHUP.
1519 user@.service, systemd-networkd.service, systemd-udevd.service, and
1520 systemd-logind have been updated to this type.
1522 * Initrd environments which are not on a pure memory file system (e.g.
1523 overlayfs combination as opposed to tmpfs) are now supported. With
1524 this change, during the initrd → host transition ("switch root")
1525 systemd will erase all files of the initrd only when the initrd is
1526 backed by a memory file system such as tmpfs.
1528 * New per-unit MemoryZSwapMax= option has been added to configure
1529 memory.zswap.max cgroup properties (the maximum amount of zswap
1532 * A new LogFilterPatterns= option has been added for units. It may be
1533 used to specify accept/deny regular expressions for log messages
1534 generated by the unit, that shall be enforced by systemd-journald.
1535 Rejected messages are neither stored in the journal nor forwarded.
1536 This option may be used to suppress noisy or uninteresting messages
1539 * The manager has a new
1540 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
1541 query process ownership via a PIDFD, which is more resilient against
1542 PID recycling issues.
1544 * Scope units now support OOMPolicy=. Login session scopes default to
1545 OOMPolicy=continue, allowing login scopes to survive the OOM killer
1546 terminating some processes in the scope.
1548 * systemd-fstab-generator now supports x-systemd.makefs option for
1549 /sysroot/ (in the initrd).
1551 * The maximum rate at which daemon reloads are executed can now be
1552 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
1553 options. (Or the equivalent on the kernel command line:
1554 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
1555 addition, systemd now logs the originating unit and PID when a reload
1556 request is received over D-Bus.
1558 * When enabling a swap device systemd will now reinitialize the device
1559 when the page size of the swap space does not match the page size of
1560 the running kernel. Note that this requires the 'swapon' utility to
1561 provide the '--fixpgsz' option, as implemented by util-linux, and it
1562 is not supported by busybox at the time of writing.
1564 * systemd now executes generator programs in a mount namespace
1565 "sandbox" with most of the file system read-only and write access
1566 restricted to the output directories, and with a temporary /tmp/
1567 mount provided. This provides a safeguard against programming errors
1568 in the generators, but also fixes here-docs in shells, which
1569 previously didn't work in early boot when /tmp/ wasn't available
1570 yet. (This feature has no security implications, because the code is
1571 still privileged and can trivially exit the sandbox.)
1573 * The system manager will now parse a new "vmm.notify_socket"
1574 system credential, which may be supplied to a VM via SMBIOS. If
1575 found, the manager will send a "READY=1" notification on the
1576 specified socket after boot is complete. This allows readiness
1577 notification to be sent from a VM guest to the VM host over a VSOCK
1580 * The sample PAM configuration file for systemd-user@.service now
1581 includes a call to pam_namespace. This puts children of user@.service
1582 in the expected namespace. (Many distributions replace their file
1583 with something custom, so this change has limited effect.)
1585 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
1586 can be used to override the mount units burst late limit for
1587 parsing '/proc/self/mountinfo', which was introduced in v249.
1590 * Drop-ins for init.scope changing control group resource limits are
1591 now applied, while they were previously ignored.
1593 * New build-time configuration options '-Ddefault-timeout-sec=' and
1594 '-Ddefault-user-timeout-sec=' have been added, to let distributions
1595 choose the default timeout for starting/stopping/aborting system and
1596 user units respectively.
1598 * Service units gained a new setting OpenFile= which may be used to
1599 open arbitrary files in the file system (or connect to arbitrary
1600 AF_UNIX sockets in the file system), and pass the open file
1601 descriptor to the invoked process via the usual file descriptor
1602 passing protocol. This is useful to give unprivileged services access
1603 to select files which have restrictive access modes that would
1604 normally not allow this. It's also useful in case RootDirectory= or
1605 RootImage= is used to allow access to files from the host environment
1606 (which is after all not visible from the service if these two options
1611 * The new net naming scheme "v253" has been introduced. In the new
1612 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
1613 a PCI bus. This extends the coverage of predictable interface names
1614 in some embedded systems.
1616 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
1617 a more informative path on some embedded systems.
1619 * Partition block devices will now also get symlinks in
1620 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
1621 block device nodes via the kernel's "diskseq" value. Previously those
1622 symlinks were only created for the main block device.
1624 * A new operator '-=' is supported for SYMLINK variables. This allows
1625 symlinks to be unconfigured even if an earlier rule added them.
1627 * 'udevadm --trigger --settle' now also works for network devices
1628 that are being renamed.
1630 Changes in sd-boot, bootctl, and the Boot Loader Specification:
1632 * systemd-boot now passes its random seed directly to the kernel's RNG
1633 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
1634 means the RNG gets seeded very early in boot before userspace has
1637 * systemd-boot will pass a disk-backed random seed – even when secure
1638 boot is enabled – if it can additionally get a random seed from EFI
1639 itself (via EFI's RNG protocol), or a prior seed in
1640 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
1642 * systemd-boot-system-token.service was renamed to
1643 systemd-boot-random-seed.service and extended to always save a random
1644 seed to ESP on every boot when a compatible boot loader is used. This
1645 allows a refreshed random seed to be used in the boot loader.
1647 * systemd-boot handles various seed inputs using a domain- and
1648 field-separated hashing scheme.
1650 * systemd-boot's 'random-seed-mode' option has been removed. A system
1651 token is now always required to be present for random seeds to be
1654 * systemd-boot now supports being loaded from other locations than the
1655 ESP, for example for direct kernel boot under QEMU or when embedded
1658 * systemd-boot now parses SMBIOS information to detect
1659 virtualization. This information is used to skip some warnings which
1660 are not useful in a VM and to conditionalize other aspects of
1663 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
1664 Secure Boot automated certificate enrollment from the ESP only if it
1665 is considered 'safe' to do so. At the moment 'safe' means running in
1668 * systemd-stub now processes random seeds in the same way as
1669 systemd-boot already does, in case a unified kernel image is being
1670 used from a different bootloader than systemd-boot, or without any
1673 * bootctl will now generate a system token on all EFI systems, even
1674 virtualized ones, and is activated in the case that the system token
1675 is missing from either sd-boot and sd-stub booted systems.
1677 * bootctl now implements two new verbs: 'kernel-identify' prints the
1678 type of a kernel image file, and 'kernel-inspect' provides
1679 information about the embedded command line and kernel version of
1682 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
1683 as for kernel-install.
1685 * The JSON output of "bootctl list" will now contain two more fields:
1686 isDefault and isSelected are boolean fields set to true on the
1687 default and currently booted boot menu entries.
1689 * bootctl gained a new verb "unlink" for removing a boot loader entry
1690 type #1 file from disk in a safe and robust way.
1692 * bootctl also gained a new verb "cleanup" that automatically removes
1693 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
1694 is not referenced anymore by any installed Type #1 boot loader
1695 specification entry. This is particularly useful in environments where
1696 a large number of entries reference the same or partly the same
1697 resources (for example, for snapshot-based setups).
1699 Changes in kernel-install:
1701 * A new "installation layout" can be configured as layout=uki. With
1702 this setting, a Boot Loader Specification Type#1 entry will not be
1703 created. Instead, a new kernel-install plugin 90-uki-copy.install
1704 will copy any .efi files from the staging area into the boot
1705 partition. A plugin to generate the UKI .efi file must be provided
1708 Changes in systemctl:
1710 * 'systemctl reboot' has dropped support for accepting a positional
1711 argument as the argument to the reboot(2) syscall. Please use the
1712 --reboot-argument= option instead.
1714 * 'systemctl disable' will now warn when called on units without
1715 install information. A new --no-warn option has been added that
1716 silences this warning.
1718 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
1719 of the drop-in to edit. (Previously, 'override.conf' was always
1722 * 'systemctl list-dependencies' now respects --type= and --state=.
1724 * 'systemctl kexec' now supports XEN VMM environments.
1726 * 'systemctl edit' will now tell the invoked editor to jump into the
1727 first line with actual unit file data, skipping over synthesized
1730 Changes in systemd-networkd and related tools:
1732 * The [DHCPv4] section in .network file gained new SocketPriority=
1733 setting that assigns the Linux socket priority used by the DHCPv4 raw
1734 socket. This may be used in conjunction with the
1735 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
1736 desired ethernet 802.1Q frame priority for DHCPv4 initial
1737 packets. This cannot be achieved with netfilter mangle tables because
1738 of the raw socket bypass.
1740 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
1741 new QuickAck= boolean setting that enables the TCP quick ACK mode for
1742 the routes configured by the acquired DHCPv4 lease or received router
1743 advertisements (RAs).
1745 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
1746 routes) now accepts three values, for high, medium, and low preference
1747 of the router (which can be set with the RouterPreference=) setting.
1749 * systemd-networkd-wait-online now supports matching via alternative
1752 * The [DHCPv6] section in .network file gained new SendRelease=
1753 setting which enables the DHCPv6 client to send release when
1754 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
1755 It is enabled by default.
1757 * If the Address= setting in [Network] or [Address] sections in .network
1758 specified without its prefix length, then now systemd-networkd assumes
1759 /32 for IPv4 or /128 for IPv6 addresses.
1761 * networkctl shows network and link file dropins in status output.
1763 Changes in systemd-dissect:
1765 * systemd-dissect gained a new option --list, to print the paths of
1766 all files and directories in a DDI.
1768 * systemd-dissect gained a new option --mtree, to generate a file
1769 manifest compatible with BSD mtree(5) of a DDI
1771 * systemd-dissect gained a new option --with, to execute a command with
1772 the specified DDI temporarily mounted and used as working
1773 directory. This is for example useful to convert a DDI to "tar"
1774 simply by running it within a "systemd-dissect --with" invocation.
1776 * systemd-dissect gained a new option --discover, to search for
1777 Discoverable Disk Images (DDIs) in well-known directories of the
1778 system. This will list machine, portable service and system extension
1781 * systemd-dissect now understands 2nd stage initrd images stored as a
1782 Discoverable Disk Image (DDI).
1784 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
1785 disk UUID stored in the GPT header) among the other data it can show.
1787 * systemd-dissect gained a new --in-memory switch to operate on an
1788 in-memory copy of the specified DDI file. This is useful to access a
1789 DDI with write access without persisting any changes. It's also
1790 useful for accessing a DDI without keeping the originating file
1793 * The DDI dissection logic will now automatically detect the intended
1794 sector size of disk images stored in files, based on the GPT
1795 partition table arrangement. Loopback block devices for such DDIs
1796 will then be configured automatically for the right sector size. This
1797 is useful to make dealing with modern 4K sector size DDIs fully
1798 automatic. The systemd-dissect tool will now show the detected sector
1799 size among the other DDI information in its output.
1801 Changes in systemd-repart:
1803 * systemd-repart gained new options --include-partitions= and
1804 --exclude-partitions= to filter operation on partitions by type UUID.
1805 This allows systemd-repart to be used to build images in which the
1806 type of one partition is set based on the contents of another
1807 partition (for example when the boot partition shall include a verity
1808 hash of the root partition).
1810 * systemd-repart also gained a --defer-partitions= option that is
1811 similar to --exclude-partitions=, but the size of the partition is
1812 still taken into account when sizing partitions, but without
1815 * systemd-repart gained a new --sector-size= option to specify what
1816 sector size should be used when an image is created.
1818 * systemd-repart now supports generating erofs file systems via
1819 CopyFiles= (a read-only file system similar to squashfs).
1821 * The Minimize= option was extended to accept "best" (which means the
1822 most minimal image possible, but may require multiple attempts) and
1823 "guess" (which means a reasonably small image).
1825 * The systemd-growfs binary now comes with a regular unit file template
1826 systemd-growfs@.service which can be instantiated directly for any
1827 desired file system. (Previously, the unit was generated dynamically
1828 by various generators, but no regular unit file template was
1831 Changes in journal tools:
1833 * Various systemd tools will append extra fields to log messages when
1834 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
1835 this includes information about D-Bus messages when sd-bus is used,
1836 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
1837 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
1838 Details of what is logged and when are subject to change.
1840 * The systemd-journald-audit.socket can now be disabled via the usual
1841 "systemctl disable" mechanism to stop collection of audit
1842 messages. Please note that it is not enabled statically anymore and
1843 must be handled by the preset/enablement logic in package
1844 installation scripts.
1846 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
1847 be used to curtail disk use by systemd-journal-remote. This is
1848 similar to the options supported by systemd-journald.
1850 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
1853 * When enrolling new keys systemd-cryptenroll now supports unlocking
1854 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
1855 password was strictly required to be specified.
1857 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
1858 (except for tokens with user verification, UV) to identify tokens
1859 before authentication. Multiple FIDO2 tokens can now be enrolled at
1860 the same time, and systemd-cryptsetup will automatically select one
1861 that corresponds to one of the available LUKS key slots.
1863 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
1864 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
1865 bank and number into which the volume key should be measured. This is
1866 automatically enabled for the encrypted root volume discovered and
1867 activated by systemd-gpt-auto-generator.
1869 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
1870 "noexec,nosuid,nodev".
1872 * systemd-gpt-auto-generator will now honour the rootfstype= and
1873 rootflags= kernel command line switches for root file systems it
1874 discovers, to match behaviour in case an explicit root fs is
1875 specified via root=.
1877 * systemd-pcrphase gained new options --machine-id and --file-system=
1878 to measure the machine-id and mount point information into PCR 15.
1879 New service unit files systemd-pcrmachine.service and
1880 systemd-pcrfs@.service have been added that invoke the tool with
1881 these switches during early boot.
1883 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
1884 with a success exit code even if no TPM device is detected.
1886 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
1887 making it harder to brute-force.
1889 Changes in other tools:
1891 * systemd-homed gained support for luksPbkdfForceIterations (the
1892 intended number of iterations for the PBKDF operation on LUKS).
1894 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
1895 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
1896 may now be used to specify additional arguments for mkfs when
1897 systemd-homed formats a file system.
1899 * systemd-hostnamed now exports the contents of
1900 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
1901 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
1902 unprivileged code to access those values.
1904 systemd-hostnamed also exports the SUPPORT_END= field from
1905 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
1906 this to show the status of the installed system.
1908 * systemd-measure gained an --append= option to sign multiple phase
1909 paths with different signing keys. This allows secrets to be
1910 accessible only in certain parts of the boot sequence. Note that
1911 'ukify' provides similar functionality in a more accessible form.
1913 * systemd-timesyncd will now write a structured log message with
1914 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
1915 on a on-disk timestamp, similarly to what it did when reaching
1916 synchronization via NTP.
1918 * systemd-timesyncd will now update the on-disk timestamp file on each
1919 boot at least once, making it more likely that the system time
1920 increases in subsequent boots.
1922 * systemd-vconsole-setup gained support for system/service credentials:
1923 vconsole.keymap/vconsole.keymap_toggle and
1924 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
1925 the similarly-named options in vconsole.conf.
1927 * systemd-localed will now save the XKB keyboard configuration to
1928 /etc/vconsole.conf, and also read it from there with a higher
1929 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
1930 file. Previously, this information was stored in the former file in
1931 converted form, and only in latter file in the original form. Tools
1932 which want to access keyboard configuration can now do so from a
1935 * systemd-resolved gained support for configuring the nameservers and
1936 search domains via kernel command line (nameserver=, domain=) and
1937 credentials (network.dns, network.search_domains).
1939 * systemd-resolved will now synthesize host names for the DNS stub
1940 addresses it supports. Specifically when "_localdnsstub" is resolved,
1941 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
1942 127.0.0.54 is returned.
1944 * systemd-notify will now send a "RELOADING=1" notification when called
1945 with --reloading, and "STOPPING=1" when called with --stopping. This
1946 can be used to implement notifications from units where it's easier
1947 to call a program than to use the sd-daemon library.
1949 * systemd-analyze's 'plot' command can now output its information in
1950 JSON, controlled via the --json= switch. Also, new --table, and
1951 --no-legend options have been added.
1953 * 'machinectl enable' will now automatically enable machines.target
1954 unit in addition to adding the machine unit to the target.
1956 Similarly, 'machinectl start|stop' gained a --now option to enable or
1957 disable the machine unit when starting or stopping it.
1959 * systemd-sysusers will now create /etc/ if it is missing.
1961 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
1962 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
1963 added to provide the new initial value for the new automated battery
1964 estimation functionality. If 'HibernateDelaySec=' is set to any value,
1965 the automated estimate (and thus the automated hibernation on low
1966 battery to avoid data loss) functionality will be disabled.
1968 * Default tmpfiles.d/ configuration will now automatically create
1969 credentials storage directory '/etc/credstore/' with the appropriate,
1970 secure permissions. If '/run/credstore/' exists, its permissions will
1971 be fixed too in case they are not correct.
1973 Changes in libsystemd and shared code:
1975 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
1976 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
1978 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
1979 128-bit ID in files such as /etc/machine-id has an invalid
1980 format. They also accept NULL as output parameter in more places,
1981 which is useful when the caller only wants to validate the inputs and
1982 does not need the output value.
1984 * sd-login gained new functions sd_pidfd_get_session(),
1985 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
1986 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
1987 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
1988 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
1989 but accept a PIDFD instead of a PID.
1991 * sd-path (and systemd-path) now export four new paths:
1992 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
1993 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
1994 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
1995 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
1997 * sd_notify() now supports AF_VSOCK as transport for notification
1998 messages (in addition to the existing AF_UNIX support). This is
1999 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
2001 * Detection of chroot() environments now works if /proc/ is not
2002 mounted. This affects systemd-detect-virt --chroot, but also means
2003 that systemd tools will silently skip various operations in such an
2006 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
2007 virtualization is now detected.
2009 Changes in the build system:
2011 * Standalone variants of systemd-repart and systemd-shutdown may now be
2012 built (if -Dstandalone=true).
2014 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
2015 example, allow scripts to conditionalize execution on AC power
2018 * The libp11kit library is now loaded through dlopen(3).
2020 Changes in the documentation:
2022 * Specifications that are not closely tied to systemd have moved to
2023 https://uapi-group.org/specifications/: the Boot Loader Specification
2024 and the Discoverable Partitions Specification.
2026 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
2027 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
2028 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
2029 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
2030 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
2031 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
2032 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
2033 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
2034 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
2035 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
2036 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
2037 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
2038 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
2039 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
2040 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
2041 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
2042 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
2043 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
2044 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
2045 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
2046 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
2047 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
2048 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
2049 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
2050 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
2051 reuben olinsky, Richard E. van der Luit, Richard Phibel,
2052 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
2053 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
2054 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
2055 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
2056 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
2057 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
2058 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
2059 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
2062 — Warsaw, 2023-02-15
2066 Announcements of Future Feature Removals:
2068 * We intend to remove cgroup v1 support from systemd release after the
2069 end of 2023. If you run services that make explicit use of cgroup v1
2070 features (i.e. the "legacy hierarchy" with separate hierarchies for
2071 each controller), please implement compatibility with cgroup v2 (i.e.
2072 the "unified hierarchy") sooner rather than later. Most of Linux
2073 userspace has been ported over already.
2075 * We intend to remove support for split-usr (/usr mounted separately
2076 during boot) and unmerged-usr (parallel directories /bin and
2077 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
2078 half of 2023, in the first release that falls into that time window.
2079 For more details, see:
2080 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
2082 Compatibility Breaks:
2084 * ConditionKernelVersion= checks that use the '=' or '!=' operators
2085 will now do simple string comparisons (instead of version comparisons
2086 à la stverscmp()). Version comparisons are still done for the
2087 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
2088 specified, a shell-style glob match is now done. This creates a minor
2089 incompatibility compared to older systemd versions when the '*', '?',
2090 '[', ']' characters are used, as these will now match as shell globs
2091 instead of literally. Given that kernel version strings typically do
2092 not include these characters we expect little breakage through this
2095 * The service manager will now read the SELinux label used for SELinux
2096 access checks from the unit file at the time it loads the file.
2097 Previously, the label would be read at the moment of the access
2098 check, which was problematic since at that time the unit file might
2099 already have been updated or removed.
2103 * systemd-measure is a new tool for calculating and signing expected
2104 TPM2 PCR values for a given unified kernel image (UKI) booted via
2105 sd-stub. The public key used for the signature and the signed
2106 expected PCR information can be embedded inside the UKI. This
2107 information can be extracted from the UKI by external tools and code
2108 in the image itself and is made available to userspace in the booted
2111 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
2112 updated to make use of this information if available in the booted
2113 kernel: when locking an encrypted volume/credential to the TPM
2114 systemd-cryptenroll/systemd-creds will use the public key to bind the
2115 volume/credential to any kernel that carries PCR information signed
2116 by the same key pair. When unlocking such volumes/credentials
2117 systemd-cryptsetup/systemd-creds will use the signature embedded in
2118 the booted UKI to gain access.
2120 Binding TPM-based disk encryption to public keys/signatures of PCR
2121 values — instead of literal PCR values — addresses the inherent
2122 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
2123 disks remain accessible even if the UKI is updated, without any TPM
2124 specific preparation during the OS update — as long as each UKI
2125 carries the necessary PCR signature information.
2127 Net effect: if you boot a properly prepared kernel, TPM-bound disk
2128 encryption now defaults to be locked to kernels which carry PCR
2129 signatures from the same key pair. Example: if a hypothetical distro
2130 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
2131 by default – bound to only FooOS kernels, and encrypted volumes bound
2132 to the TPM cannot be unlocked on kernels from other sources. (But do
2133 note this behaviour requires preparation/enabling in the UKI, and of
2134 course users can always enroll non-TPM ways to unlock the volume.)
2136 * systemd-pcrphase is a new tool that is invoked at six places during
2137 system runtime, and measures additional words into TPM2 PCR 11, to
2138 mark milestones of the boot process. This allows binding access to
2139 specific TPM2-encrypted secrets to specific phases of the boot
2140 process. (Example: LUKS2 disk encryption key only accessible in the
2141 initrd, but not later.)
2143 Changes in systemd itself, i.e. the manager and units
2145 * The cpu controller is delegated to user manager units by default, and
2146 CPUWeight= settings are applied to the top-level user slice units
2147 (app.slice, background.slice, session.slice). This provides a degree
2148 of resource isolation between different user services competing for
2151 * Systemd can optionally do a full preset in the "first boot" condition
2152 (instead of just enable-only). This behaviour is controlled by the
2153 compile-time option -Dfirst-boot-full-preset. Right now it defaults
2154 to 'false', but the plan is to switch it to 'true' for the subsequent
2157 * Drop-ins are now allowed for transient units too.
2159 * Systemd will set the taint flag 'support-ended' if it detects that
2160 the OS image is past its end-of-support date. This date is declared
2161 in a new /etc/os-release field SUPPORT_END= described below.
2163 * Two new settings ConditionCredential= and AssertCredential= can be
2164 used to skip or fail units if a certain system credential is not
2167 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
2169 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
2170 specify the SMACK security label to use when not specified in a unit
2173 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
2174 specify the default timeout when waiting for device units to
2177 * C.UTF-8 is used as the default locale if nothing else has been
2180 * [Condition|Assert]Firmware= have been extended to support certain
2181 SMBIOS fields. For example
2183 ConditionFirmware=smbios-field(board_name = "Custom Board")
2185 conditionalizes the unit to run only when
2186 /sys/class/dmi/id/board_name contains "Custom Board" (without the
2189 * ConditionFirstBoot= now correctly evaluates as true only during the
2190 boot phase of the first boot. A unit executed later, after booting
2191 has completed, will no longer evaluate this condition as true.
2193 * Socket units will now create sockets in the SELinuxContext= of the
2194 associated service unit, if any.
2196 * Boot phase transitions (start initrd → exit initrd → boot complete →
2197 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
2198 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
2199 unsealed only in the initrd.
2201 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
2202 also be provided to ExecStartPre= processes.
2204 * Various units are now correctly ordered against
2205 initrd-switch-root.target where previously a conflict without
2206 ordering was configured. A stop job for those units would be queued,
2207 but without the ordering it could be executed only after
2208 initrd-switch-root.service, leading to units not being restarted in
2209 the host system as expected.
2211 * In order to fully support the IPMI watchdog driver, which has not yet
2212 been ported to the new common watchdog device interface,
2213 /dev/watchdog0 will be tried first and systemd will silently fallback
2214 to /dev/watchdog if it is not found.
2216 * New watchdog-related D-Bus properties are now published by systemd:
2217 WatchdogDevice, WatchdogLastPingTimestamp,
2218 WatchdogLastPingTimestampMonotonic.
2220 * At shutdown, API virtual files systems (proc, sys, etc.) will be
2223 * At shutdown, systemd will now log about processes blocking unmounting
2226 * A new meson build option 'clock-valid-range-usec-max' was added to
2227 allow disabling system time correction if RTC returns a timestamp far
2230 * Propagated restart jobs will no longer be discarded while a unit is
2233 * PID 1 will now import system credentials from SMBIOS Type 11 fields
2234 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
2235 simple, fast and generic path for supplying credentials to a VM,
2236 without involving external tools such as cloud-init/ignition.
2238 * The CPUWeight= setting of unit files now accepts a new special value
2239 "idle", which configures "idle" level scheduling for the unit.
2241 * Service processes that are activated due to a .timer or .path unit
2242 triggering will now receive information about this via environment
2243 variables. Note that this is information is lossy, as activation
2244 might be coalesced and only one of the activating triggers will be
2245 reported. This is hence more suited for debugging or tracing rather
2246 than for behaviour decisions.
2248 * The riscv_flush_icache(2) system call has been added to the list of
2249 system calls allowed by default when SystemCallFilter= is used.
2251 * The selinux context derived from the target executable, instead of
2252 'init_t' used for the manager itself, is now used when creating
2253 listening sockets for units that specify SELinuxContextFromNet=yes.
2255 Changes in sd-boot, bootctl, and the Boot Loader Specification:
2257 * The Boot Loader Specification has been cleaned up and clarified.
2258 Various corner cases in version string comparisons have been fixed
2259 (e.g. comparisons for empty strings). Boot counting is now part of
2260 the main specification.
2262 * New PCRs measurements are performed during boot: PCR 11 for the
2263 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
2264 took place this is now reported to userspace via the new
2265 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
2267 * As before, systemd-stub will measure kernel parameters and system
2268 credentials into PCR 12. It will now report this fact via the
2269 StubPcrKernelParameters EFI variable to userspace.
2271 * The UEFI monotonic boot counter is now included in the updated random
2272 seed file maintained by sd-boot, providing some additional entropy.
2274 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
2275 of arranging the image manually and jumping to the kernel entry
2276 point. sd-stub also installs a temporary UEFI SecurityOverride to
2277 allow the (unsigned) nested image to be booted. This is safe because
2278 the outer (signed) stub+kernel binary must have been verified before
2279 the stub was executed.
2281 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
2282 is now supported by sd-boot.
2284 * bootctl gained a bunch of new options: --all-architectures to install
2285 binaries for all supported EFI architectures, --root= and --image=
2286 options to operate on a directory or disk image, and
2287 --install-source= to specify the source for binaries to install,
2288 --efi-boot-option-description= to control the name of the boot entry.
2290 * The sd-boot stub exports a StubFeatures flag, which is used by
2291 bootctl to show features supported by the stub that was used to boot.
2293 * The PE section offsets that are used by tools that assemble unified
2294 kernel images have historically been hard-coded. This may lead to
2295 overlapping PE sections which may break on boot. The UKI will now try
2296 to detect and warn about this.
2298 Any tools that assemble UKIs must update to calculate these offsets
2299 dynamically. Future sd-stub versions may use offsets that will not
2300 work with the currently used set of hard-coded offsets!
2302 * sd-stub now accepts (and passes to the initrd and then to the full
2303 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
2304 signatures of expected PCR values, to allow sealing secrets via the
2305 TPM2 against pre-calculated PCR measurements.
2307 Changes in the hardware database:
2309 * 'systemd-hwdb query' now supports the --root= option.
2311 Changes in systemctl:
2313 * systemctl now supports --state= and --type= options for the 'show'
2316 * systemctl gained a new verb 'list-automounts' to list automount
2319 * systemctl gained support for a new --image= switch to be able to
2320 operate on the specified disk image (similar to the existing --root=
2321 which operates relative to some directory).
2323 Changes in systemd-networkd:
2325 * networkd can set Linux NetLabel labels for integration with the
2326 network control in security modules via a new NetLabel= option.
2328 * The RapidCommit= is (re-)introduced to enable faster configuration
2329 via DHCPv6 (RFC 3315).
2331 * networkd gained a new option TCPCongestionControlAlgorithm= that
2332 allows setting a per-route TCP algorithm.
2334 * networkd gained a new option KeepFileDescriptor= to allow keeping a
2335 reference (file descriptor) open on TUN/TAP interfaces, which is
2336 useful to avoid link flaps while the underlying service providing the
2337 interface is being serviced.
2339 * RouteTable= now also accepts route table names.
2341 Changes in systemd-nspawn:
2343 * The --bind= and --overlay= options now support relative paths.
2345 * The --bind= option now supports a 'rootidmap' value, which will
2346 use id-mapped mounts to map the root user inside the container to the
2347 owner of the mounted directory on the host.
2349 Changes in systemd-resolved:
2351 * systemd-resolved now persists DNSOverTLS in its state file too. This
2352 fixes a problem when used in combination with NetworkManager, which
2353 sends the setting only once, causing it to be lost if resolved was
2354 restarted at any point.
2356 * systemd-resolved now exposes a Varlink socket at
2357 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
2358 root. Processed DNS requests in a JSON format will be published to
2359 any clients connected to this socket.
2361 resolvectl gained a 'monitor' verb to make use of this.
2363 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
2364 instead of returning SERVFAIL, as per RFC:
2365 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
2367 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
2368 is still supported.)
2370 Changes in libsystemd and other libraries:
2372 * libsystemd now exports sd_bus_error_setfv() (a convenience function
2373 for setting bus errors), sd_id128_string_equal (a convenience
2374 function for 128-bit ID string comparisons), and
2375 sd_bus_message_read_strv_extend() (a function to incrementally read
2378 * libsystemd now exports sd_device_get_child_first()/_next() as a
2379 high-level interface for enumerating child devices. It also supports
2380 sd_device_new_child() for opening a child device given a device
2383 * libsystemd now exports sd_device_monitor_set()/get_description()
2384 which allow setting a custom description that will be used in log
2385 messages by sd_device_monitor*.
2387 * Private shared libraries (libsystemd-shared-nnn.so,
2388 libsystemd-core-nnn.so) are now installed into arch-specific
2389 directories to allow multi-arch installs.
2391 * A new sd-gpt.h header is now published, listing GUIDs from the
2392 Discoverable Partitions specification. For more details see:
2393 https://systemd.io/DISCOVERABLE_PARTITIONS/
2395 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
2396 database given an explicit path to the file.
2398 * The signal number argument to sd_event_add_signal() now can now be
2399 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
2400 be automatically invoked to block the specified signal. This is
2401 useful to simplify invocations as the caller doesn't have to do this
2404 * A new convenience call sd_event_set_signal_exit() has been added to
2405 sd-event to set up signal handling so that the event loop
2406 automatically terminates cleanly on SIGTERM/SIGINT.
2408 Changes in other components:
2410 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
2411 can now be provided via the credential mechanism.
2413 * systemd-analyze gained a new verb 'compare-versions' that implements
2414 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
2415 'dpkg --compare-versions').
2417 * 'systemd-analyze dump' is extended to accept glob patterns for unit
2418 names to limit the output to matching units.
2420 * tmpfiles.d/ lines can read file contents to write from a credential.
2421 The new modifier char '^' is used to specify that the argument is a
2422 credential name. This mechanism is used to automatically populate
2423 /etc/motd, /etc/issue, and /etc/hosts from credentials.
2425 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
2426 an inode if the specification is prefixed with ':' and the inode
2429 * Default tmpfiles.d/ configuration now carries a line to automatically
2430 use an 'ssh.authorized_keys.root' credential if provided to set up
2431 the SSH authorized_keys file for the root user.
2433 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
2436 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
2437 in base64. This is useful to write arbitrary binary data into files.
2439 * The pkgconfig and rpm macros files now export the directory for user
2440 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
2442 * Detection of Apple Virtualization and detection of Parallels and
2443 KubeVirt virtualization on non-x86 archs have been added.
2445 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
2446 user when their system will become unsupported.
2448 * When performing suspend-then-hibernate, the system will estimate the
2449 discharge rate and use that to set the delay until hibernation and
2450 hibernate immediately instead of suspending when running from a
2451 battery and the capacity is below 5%.
2453 * systemd-sysctl gained a --strict option to fail when a sysctl
2454 setting is unknown to the kernel.
2456 * machinectl supports --force for the 'copy-to' and 'copy-from'
2459 * coredumpctl gained the --root and --image options to look for journal
2460 files under the specified root directory, image, or block device.
2462 * 'journalctl -o' and similar commands now implement a new output mode
2463 "short-delta". It is similar to "short-monotonic", but also shows the
2464 time delta between subsequent messages.
2466 * journalctl now respects the --quiet flag when verifying consistency
2469 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
2470 will indicate whether a message was logged in the 'initrd' phase or
2471 in the 'system' phase of the boot process.
2473 * Journal files gained a new compatibility flag
2474 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
2475 to the storage format that allow reducing size on disk. As with other
2476 compatibility flags, older journalctl versions will not be able to
2477 read journal files using this new format. The environment variable
2478 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
2479 disable this functionality. It is enabled by default.
2481 * systemd-run's --working-directory= switch now works when used in
2482 combination with --scope.
2484 * portablectl gained a --force flag to skip certain sanity checks. This
2485 is implemented using new flags accepted by systemd-portabled for the
2486 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
2487 flag now means that the attach/detach checks whether the units are
2488 already present and running will be skipped. Similarly,
2489 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
2490 image name matches the name declared inside of the image will be
2491 skipped. Callers must be sure to do those checks themselves if
2494 * systemd-portabled will now use the original filename to check
2495 extension-release.NAME for correctness, in case it is passed a
2498 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
2501 * sysext's extension-release files now support '_any' as a special
2502 value for the ID= field, to allow distribution-independent extensions
2503 (e.g.: fully statically compiled binaries, scripts). It also gained
2504 support for a new ARCHITECTURE= field that may be used to explicitly
2505 restrict an image to hosts of a specific architecture.
2507 * systemd-repart now supports creating squashfs partitions. This
2508 requires mksquashfs from squashfs-tools.
2510 * systemd-repart gained a --split flag to also generate split
2511 artifacts, i.e. a separate file for each partition. This is useful in
2512 conjunction with systemd-sysupdate or other tools, or to generate
2513 split dm-verity artifacts.
2515 * systemd-repart is now able to generate dm-verity partitions, including
2518 * systemd-repart can now set a partition UUID to zero, allowing it to
2519 be filled in later, such as when using verity partitions.
2521 * systemd-repart now supports drop-ins for its configuration files.
2523 * Package metadata logged by systemd-coredump in the system journal is
2526 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
2528 * systemd-oomd now automatically links against libatomic, if available.
2530 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
2533 * scope units now also provide oom-kill status.
2535 * systemd-pstore will now try to load only the efi_pstore kernel module
2536 before running, ensuring that pstore can be used.
2538 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
2539 session after a preconfigure timeout.
2541 * systemd-homed will now wait up to 30 seconds for workers to terminate,
2542 rather than indefinitely.
2544 * homectl gained a new '--luks-sector-size=' flag that allows users to
2545 select the preferred LUKS sector size. Must be a power of 2 between 512
2546 and 4096. systemd-userdbd records gained a corresponding field.
2548 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
2549 variable when generating the 'sp_lstchg' field, to ensure an image
2550 build can be reproducible.
2552 * 'udevadm wait' will now listen to kernel uevents too when called with
2555 * When naming network devices udev will now consult the Devicetree
2556 "alias" fields for the device.
2558 * systemd-udev will now create infiniband/by-path and
2559 infiniband/by-ibdev links for Infiniband verbs devices.
2561 * systemd-udev-trigger.service will now also prioritize input devices.
2563 * ConditionACPower= and systemd-ac-power will now assume the system is
2564 running on AC power if no battery can be found.
2566 * All features and tools using the TPM2 will now communicate with it
2567 using a bind key. Beforehand, the tpm2 support used encrypted sessions
2568 by creating a primary key that was used to encrypt traffic. This
2569 creates a problem as the key created for encrypting the traffic could
2570 be faked by an active interposer on the bus. In cases when a pin is
2571 used, a bind key will be used. The pin is used as the auth value for
2572 the seal key, aka the disk encryption key, and that auth value will be
2573 used in the session establishment. An attacker would need the pin
2574 value to create the secure session and thus an active interposer
2575 without the pin cannot interpose on TPM2 traffic.
2577 * systemd-growfs no longer requires udev to run.
2579 * systemd-backlight now will better support systems with multiple
2582 * systemd-cryptsetup's keyfile-timeout= option now also works when a
2583 device is used as a keyfile.
2585 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
2586 unlocking key from a key file (instead of prompting the user). Note
2587 that this is the key for unlocking the volume in order to be able to
2588 enroll a new key, but it is not the key that is enrolled.
2590 * systemd-dissect gained a new --umount switch that will safely and
2591 synchronously unmount all partitions of an image previously mounted
2592 with 'systemd-dissect --mount'.
2594 * When using gcrypt, all systemd tools and services will now configure
2595 it to prefer the OS random number generator if present.
2597 * All example code shipped with documentation has been relicensed from CC0
2600 * Unit tests will no longer fail when running on a system without
2603 Experimental features:
2605 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
2606 and bpftool >= 7.0).
2608 * sd-boot can automatically enroll SecureBoot keys from files found on
2609 the ESP. This enrollment can be either automatic ('force' mode) or
2610 controlled by the user ('manual' mode). It is sufficient to place the
2611 SecureBoot keys in the right place in the ESP and they will be picked
2612 up by sd-boot and shown in the boot menu.
2614 * The mkosi config in systemd gained support for automatically
2615 compiling a kernel with the configuration appropriate for testing
2616 systemd. This may be useful when developing or testing systemd in
2617 tandem with the kernel.
2619 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
2620 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
2621 Alexander Graf, Alexander Shopov, Alexander Wilson,
2622 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
2623 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
2624 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
2625 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
2626 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
2627 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
2628 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
2629 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
2630 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
2631 David Jaša, David Rheinsberg, David Seifert, David Tardon,
2632 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
2633 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
2634 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
2635 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
2636 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
2637 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
2638 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
2639 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
2640 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
2641 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
2642 JeroenHD, jiangchuangang, João Loureiro,
2643 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
2644 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
2645 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
2646 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
2647 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
2648 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
2649 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
2650 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
2651 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
2652 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
2653 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
2654 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
2655 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
2656 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
2657 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
2658 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
2659 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
2660 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
2661 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
2662 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
2663 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
2664 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
2665 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
2666 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
2668 – The Great Beyond, 2022-10-31 👻
2672 Backwards-incompatible changes:
2674 * The minimum kernel version required has been bumped from 3.13 to 4.15,
2675 and CLOCK_BOOTTIME is now assumed to always exist.
2677 * C11 with GNU extensions (aka "gnu11") is now used to build our
2678 components. Public API headers are still restricted to ISO C89.
2680 * In v250, a systemd-networkd feature that automatically configures
2681 routes to addresses specified in AllowedIPs= was added and enabled by
2682 default. However, this causes network connectivity issues in many
2683 existing setups. Hence, it has been disabled by default since
2684 systemd-stable 250.3. The feature can still be used by explicitly
2685 configuring RouteTable= setting in .netdev files.
2687 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
2688 when a Condition*= check does not succeed, restoring the JobRemoved
2689 signal to the behaviour it had before v250.
2691 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
2692 GetImageMetadataWithExtensions() have been fixed to provide an extra
2693 return parameter, containing the actual extension release metadata.
2694 The current implementation was judged to be broken and unusable, and
2695 thus the usual procedure of adding a new set of methods was skipped,
2696 and backward compatibility broken instead on the assumption that
2697 nobody can be affected given the current state of this interface.
2699 * All kernels supported by systemd mix bytes returned by RDRAND (or
2700 similar) into the entropy pool at early boot. This means that on
2701 those systems, even if /dev/urandom is not yet initialized, it still
2702 returns bytes that are of at least RDRAND quality. For that reason,
2703 we no longer have reason to invoke RDRAND from systemd itself, which
2704 has historically been a source of bugs. Furthermore, kernels ≥5.6
2705 provide the getrandom(GRND_INSECURE) interface for returning random
2706 bytes before the entropy pool is initialized without warning into
2707 kmsg, which is what we attempt to use if available. systemd's direct
2708 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
2709 running an older kernel may experience kmsg warnings that were not
2710 seen with 250. For newer kernels, non-x86 systems, or older x86
2711 systems, there should be no visible changes.
2713 * sd-boot will now measure the kernel command line into TPM PCR 12
2714 rather than PCR 8. This improves usefulness of the measurements on
2715 systems where sd-boot is chainloaded from Grub. Grub measures all
2716 commands its executes into PCR 8, which makes it very hard to use
2717 reasonably, hence separate ourselves from that and use PCR 12
2718 instead, which is what certain Ubuntu editions already do. To retain
2719 compatibility with systems running older systemd systems a new meson
2720 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
2721 If enabled, the measurement is done twice: into the new-style PCR 12
2722 *and* the old-style PCR 8. It's strongly advised to migrate all users
2723 to PCR 12 for this purpose in the long run, as we intend to remove
2724 this compatibility feature in two years' time.
2726 * busctl capture now writes output in the newer pcapng format instead
2729 * A udev rule that imported hwdb matches for USB devices with lowercase
2730 hexadecimal vendor/product ID digits was added in systemd 250. This
2731 has been reverted, since uppercase hexadecimal digits are supposed to
2732 be used, and we already had a rule with the appropriate match.
2734 Users might need to adjust their local hwdb entries.
2736 * arch_prctl(2) has been moved to the @default set in the syscall filters
2737 (as exposed via the SystemCallFilter= setting in service unit files).
2738 It is apparently used by the linker now.
2740 * The tmpfiles entries that create the /run/systemd/netif directory and
2741 its subdirectories were moved from tmpfiles.d/systemd.conf to
2742 tmpfiles.d/systemd-network.conf.
2744 Users might need to adjust their files that override tmpfiles.d/systemd.conf
2745 to account for this change.
2747 * The requirement for Portable Services images to contain a well-formed
2748 os-release file (i.e.: contain at least an ID field) is now enforced.
2749 This applies to base images and extensions, and also to systemd-sysext.
2751 Changes in the Boot Loader Specification, kernel-install and sd-boot:
2753 * kernel-install's and bootctl's Boot Loader Specification Type #1
2754 entry generation logic has been reworked. The user may now pick
2755 explicitly by which "token" string to name the installation's boot
2756 entries, via the new /etc/kernel/entry-token file or the new
2757 --entry-token= switch to bootctl. By default — as before — the
2758 entries are named after the local machine ID. However, in "golden
2759 image" environments, where the machine ID shall be initialized on
2760 first boot (as opposed to at installation time before first boot) the
2761 machine ID will not be available at build time. In this case the
2762 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
2763 file) may be used to override the "token" for the entries, for
2764 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
2765 make the OS images independent of any machine ID, and ensure that the
2766 images will not carry any identifiable information before first boot,
2767 but on the other hand means that multiple parallel installations of
2768 the very same image on the same disk cannot be supported.
2770 Summary: if you are building golden images that shall acquire
2771 identity information exclusively on first boot, make sure to both
2772 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
2773 value of the IMAGE_ID= or ID= field of /etc/os-release or another
2774 suitable identifier before deploying the image.
2776 * The Boot Loader Specification has been extended with
2777 /loader/entries.srel file located in the EFI System Partition (ESP)
2778 that disambiguates the format of the entries in the /loader/entries/
2779 directory (in order to discern them from incompatible uses of this
2780 directory by other projects). For entries that follow the
2781 Specification, the string "type1" is stored in this file.
2783 bootctl will now write this file automatically when installing the
2784 systemd-boot boot loader.
2786 * kernel-install supports a new initrd_generator= setting in
2787 /etc/kernel/install.conf, that is exported as
2788 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
2789 allows choosing different initrd generators.
2791 * kernel-install will now create a "staging area" (an initially-empty
2792 directory to gather files for a Boot Loader Specification Type #1
2793 entry). The path to this directory is exported as
2794 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
2795 drop files there instead of writing them directly to the final
2796 location. kernel-install will move them when all files have been
2797 prepared successfully.
2799 * New option sort-key= has been added to the Boot Loader Specification
2800 to override the sorting order of the entries in the boot menu. It is
2801 read by sd-boot and bootctl, and will be written by kernel-install,
2802 with the default value of IMAGE_ID= or ID= fields from
2803 os-release. Together, this means that on multiboot installations,
2804 entries should be grouped and sorted in a predictable way.
2806 * The sort order of boot entries has been updated: entries which have
2807 the new field sort-key= are sorted by it first, and all entries
2808 without it are ordered later. After that, entries are sorted by
2809 version so that newest entries are towards the beginning of the list.
2811 * The kernel-install tool gained a new 'inspect' verb which shows the
2812 paths and other settings used.
2814 * sd-boot can now optionally beep when the menu is shown and menu
2815 entries are selected, which can be useful on machines without a
2816 working display. (Controllable via a loader.conf setting.)
2818 * The --make-machine-id-directory= switch to bootctl has been replaced
2819 by --make-entry-directory=, given that the entry directory is not
2820 necessarily named after the machine ID, but after some other suitable
2821 ID as selected via --entry-token= described above. The old name of
2822 the option is still understood to maximize compatibility.
2824 * 'bootctl list' gained support for a new --json= switch to output boot
2825 menu entries in JSON format.
2827 * 'bootctl is-installed' now supports the --graceful, and various verbs
2828 omit output with the new option --quiet.
2830 Changes in systemd-homed:
2832 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
2833 of activated home directories it manages (if the kernel and selected
2834 file systems support it). So far it mapped three UID ranges: the
2835 range from 0…60000, the user's own UID, and the range 60514…65534,
2836 leaving everything else unmapped (in other words, the 16-bit UID range
2837 is mapped almost fully, with the exception of the UID subrange used
2838 for systemd-homed users, with one exception: the user's own UID).
2839 Unmapped UIDs may not be used for file ownership in the home
2840 directory — any chown() attempts with them will fail. With this
2841 release a fourth range is added to these mappings:
2842 524288…1879048191. This range is the UID range intended for container
2845 https://systemd.io/UIDS-GIDS
2847 This range may be used for container managers that place container OS
2848 trees in the home directory (which is a questionable approach, for
2849 quota, permission, SUID handling and network file system
2850 compatibility reasons, but nonetheless apparently commonplace). Note
2851 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
2852 UID assignments from the range are not managed or mapped by
2853 `systemd-homed`, and must be managed with other mechanisms, in the
2854 context of the local system.
2856 Typically, a better approach to user namespacing in relevant
2857 container managers would be to leave container OS trees on disk at
2858 UID offset 0, but then map them to a dynamically allocated runtime
2859 UID range via another UID mount map at container invocation
2860 time. That way user namespace UID ranges become strictly a runtime
2861 concept, and do not leak into persistent file systems, persistent
2862 user databases or persistent configuration, thus greatly simplifying
2863 handling, and improving compatibility with home directories intended
2864 to be portable like the ones managed by systemd-homed.
2866 Changes in shared libraries:
2868 * A new libsystemd-core-<version>.so private shared library is
2869 installed under /usr/lib/systemd/system, mirroring the existing
2870 libsystemd-shared-<version>.so library. This allows the total
2871 installation size to be reduced by binary code reuse.
2873 * The <version> tag used in the name of libsystemd-shared.so and
2874 libsystemd-core.so can be configured via the meson option
2875 'shared-lib-tag'. Distributions may build subsequent versions of the
2876 systemd package with unique tags (e.g. the full package version),
2877 thus allowing multiple installations of those shared libraries to be
2878 available at the same time. This is intended to fix an issue where
2879 programs that link to those libraries would fail to execute because
2880 they were installed earlier or later than the appropriate version of
2883 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
2884 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
2885 format instead of as a simple series of hex characters.
2887 * The sd-device API gained two new calls sd_device_new_from_devname()
2888 and sd_device_new_from_path() which permit allocating an sd_device
2889 object from a device node name or file system path.
2891 * sd-device also gained a new call sd_device_open() which will open the
2892 device node associated with a device for which an sd_device object
2893 has been allocated. The call is supposed to address races around
2894 device nodes being removed/recycled due to hotplug events, or media
2895 change events: the call checks internally whether the major/minor of
2896 the device node and the "diskseq" (in case of block devices) match
2897 with the metadata loaded in the sd_device object, thus ensuring that
2898 the device once opened really matches the provided sd_device object.
2900 Changes in PID1, systemctl, and systemd-oomd:
2902 * A new set of service monitor environment variables will be passed to
2903 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
2904 handler unit as OnFailure=/OnSuccess=. The variables are:
2905 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
2906 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
2907 handler needs to watch multiple units, use a templated handler.
2909 * A new ExtensionDirectories= setting in service unit files allows
2910 system extensions to be loaded from a directory. (It is similar to
2911 ExtensionImages=, but takes paths to directories, instead of
2914 'portablectl attach --extension=' now also accepts directory paths.
2916 * The user.delegate and user.invocation_id extended attributes on
2917 cgroups are used in addition to trusted.delegate and
2918 trusted.invocation_id. The latter pair requires privileges to set,
2919 but the former doesn't and can be also set by the unprivileged user
2922 (Only supported on kernels ≥5.6.)
2924 * Units that were killed by systemd-oomd will now have a service result
2925 of 'oom-kill'. The number of times a service was killed is tallied
2926 in the 'user.oomd_ooms' extended attribute.
2928 The OOMPolicy= unit file setting is now also honoured by
2931 * In unit files the new %y/%Y specifiers can be used to refer to
2932 normalized unit file path, which is particularly useful for symlinked
2935 The new %q specifier resolves to the pretty hostname
2936 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
2938 The new %d specifier resolves to the credentials directory of a
2939 service (same as $CREDENTIALS_DIRECTORY).
2941 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
2942 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
2943 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
2944 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
2945 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
2946 MountFlags= service settings now also work in unprivileged user
2947 services, i.e. those run by the user's --user service manager, as long
2948 as user namespaces are enabled on the system.
2950 * Services with Restart=always and a failing ExecCondition= will no
2951 longer be restarted, to bring ExecCondition= behaviour in line with
2952 Condition*= settings.
2954 * LoadCredential= now accepts a directory as the argument; all files
2955 from the directory will be loaded as credentials.
2957 * A new D-Bus property ControlGroupId is now exposed on service units,
2958 that encapsulates the service's numeric cgroup ID that newer kernels
2959 assign to each cgroup.
2961 * PID 1 gained support for configuring the "pre-timeout" of watchdog
2962 devices and the associated governor, via the new
2963 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
2964 options in /etc/systemd/system.conf.
2966 * systemctl's --timestamp= option gained a new choice "unix", to show
2967 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
2969 * A new "taint" flag named "old-kernel" is introduced which is set when
2970 the kernel systemd runs on is older then the current baseline version
2971 (see above). The flag is shown in "systemctl status" output.
2973 * Two additional taint flags "short-uid-range" and "short-gid-range"
2974 have been added as well, which are set when systemd notices it is run
2975 within a userns namespace that does not define the full 0…65535 UID
2978 * A new "unmerged-usr" taint flag has been added that is set whenever
2979 running on systems where /bin/ + /sbin/ are *not* symlinks to their
2980 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
2983 * Generators invoked by PID 1 will now have a couple of useful
2984 environment variables set describing the execution context a
2985 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
2986 system service manager, or from the per-user service
2987 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
2988 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
2989 systemd considers the current boot to be a "first"
2990 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
2991 detected and which type of hypervisor/container
2992 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
2993 kernel is built for.
2995 * PID 1 will now automatically pick up system credentials from qemu's
2996 fw_cfg interface, thus allowing passing arbitrary data into VM
2997 systems similar to how this is already supported for passing them
2998 into `systemd-nspawn` containers. Credentials may now also be passed
2999 in via the new kernel command line option `systemd.set_credential=`
3000 (note that kernel command line options are world-readable during
3001 runtime, and only useful for credentials that require no
3002 confidentiality). The credentials that can be passed to unified
3003 kernels that use the `systemd-stub` UEFI stub are now similarly
3004 picked up automatically. Automatic importing of system credentials
3005 this way can be turned off via the new
3006 `systemd.import_credentials=no` kernel command line option.
3008 * LoadCredential= will now automatically look for credentials in the
3009 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
3010 the argument is not an absolute path. Similarly,
3011 LoadCredentialEncrypted= will check the same directories plus
3012 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
3013 /usr/lib/credstore.encrypted/. The idea is to use those directories
3014 as the system-wide location for credentials that services should pick
3017 * System and service credentials are described in great detail in a new
3020 https://systemd.io/CREDENTIALS
3022 Changes in systemd-journald:
3024 * The journal JSON export format has been added to listed of stable
3025 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
3027 * journalctl --list-boots now supports JSON output and the --reverse option.
3029 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
3030 updated, BUILDING_IMAGES is new:
3032 https://systemd.io/JOURNAL_EXPORT_FORMATS
3033 https://systemd.io/BUILDING_IMAGES
3037 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
3038 calculators, etc.), the other AV production devices (DJ tables,
3039 keypads, etc.) that should accessible to the seat owner user by
3042 * udevadm trigger gained a new --prioritized-subsystem= option to
3043 process certain subsystems (and all their parent devices) earlier.
3045 systemd-udev-trigger.service now uses this new option to trigger
3046 block and TPM devices first, hopefully making the boot a bit faster.
3048 * udevadm trigger now implements --type=all, --initialized-match,
3049 --initialized-nomatch to trigger both subsystems and devices, only
3050 already-initialized devices, and only devices which haven't been
3051 initialized yet, respectively.
3053 * udevadm gained a new "wait" command for safely waiting for a specific
3054 device to show up in the udev device database. This is useful in
3055 scripts that asynchronously allocate a block device (e.g. through
3056 repartitioning, or allocating a loopback device or similar) and need
3057 to synchronize on the creation to complete.
3059 * udevadm gained a new "lock" command for locking one or more block
3060 devices while formatting it or writing a partition table to it. It is
3061 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
3062 usable in scripts dealing with block devices.
3064 * udevadm info will show a couple of additional device fields in its
3065 output, and will not apply a limited set of coloring to line types.
3067 * udevadm info --tree will now show a tree of objects (i.e. devices and
3068 suchlike) in the /sys/ hierarchy.
3070 * Block devices will now get a new set of device symlinks in
3071 /dev/disk/by-diskseq/<nr>, which may be used to reference block
3072 device nodes via the kernel's "diskseq" value. Note that this does
3073 not guarantee that opening a device by a symlink like this will
3074 guarantee that the opened device actually matches the specified
3075 diskseq value. To be safe against races, the actual diskseq value of
3076 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
3077 the one in the symlink path.
3079 * .link files gained support for setting MDI/MID-X on a link.
3081 * .link files gained support for [Match] Firmware= setting to match on
3082 the device firmware description string. By mistake, it was previously
3083 only supported in .network files.
3085 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
3086 and [SR-IOV] section to configure SR-IOV virtual functions.
3088 Changes in systemd-networkd:
3090 * The default scope for unicast routes configured through [Route]
3091 section is changed to "link", to make the behavior consistent with
3092 "ip route" command. The manual configuration of [Route] Scope= is
3095 * A new unit systemd-networkd-wait-online@<interface>.service has been
3096 added that can be used to wait for a specific network interface to be
3099 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
3100 that configures the eponymous kernel attribute on the bridge.
3102 * .netdev files now can be used to create virtual WLAN devices, and
3103 configure various settings on them, via the [WLAN] section.
3105 * .link/.network files gained support for [Match] Kind= setting to match
3106 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
3108 This value is also shown by 'networkctl status'.
3110 * The Local= setting in .netdev files for various virtual network
3111 devices gained support for specifying, in addition to the network
3112 address, the name of a local interface which must have the specified
3115 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
3116 files, to configure tunnels in external mode (a.k.a. collect metadata
3119 * [Network] L2TP= setting was removed. Please use interface specifier in
3120 Local= setting in .netdev files of corresponding L2TP interface.
3122 * New [DHCPServer] BootServerName=, BootServerAddress=, and
3123 BootFilename= settings can be used to configure the server address,
3124 server name, and file name sent in the DHCP packet (e.g. to configure
3127 Changes in systemd-resolved:
3129 * systemd-resolved is started earlier (in sysinit.target), so it
3130 available earlier and will also be started in the initrd if installed
3133 Changes in disk encryption:
3135 * systemd-cryptenroll can now control whether to require the user to
3136 enter a PIN when using TPM-based unlocking of a volume via the new
3137 --tpm2-with-pin= option.
3139 Option tpm2-pin= can be used in /etc/crypttab.
3141 * When unlocking devices via TPM, TPM2 parameter encryption is now
3142 used, to ensure that communication between CPU and discrete TPM chips
3143 cannot be eavesdropped to acquire disk encryption keys.
3145 * A new switch --fido2-credential-algorithm= has been added to
3146 systemd-cryptenroll allowing selection of the credential algorithm to
3147 use when binding encryption to FIDO2 tokens.
3149 Changes in systemd-hostnamed:
3151 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
3152 to override the values gleaned from the hwdb.
3154 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
3155 /sys/class/dmi/id) to override the chassis that is reported by
3158 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
3159 for reading the hardware serial number, as reportd by DMI. It also
3160 exposes a new method D-Bus property FirmwareVersion that encode the
3161 firmware version of the system.
3163 Changes in other components:
3165 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
3166 handling with the values that were configured during systemd build
3167 (if /etc/locale.conf has not been created through some other
3168 mechanism). This means that /etc/locale.conf should always have
3169 reasonable contents and we avoid a potential mismatch in defaults.
3171 * The userdbctl tool will now show UID range information as part of the
3172 list of known users.
3174 * A new build-time configuration setting default-user-shell= can be
3175 used to set the default shell for user records and nspawn shell
3176 invocations (instead of the default /bin/bash).
3178 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
3179 information dynamically at runtime via IPC.
3181 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
3182 whether a functioning TPM2 infrastructure is available, i.e. if
3183 firmware, kernel driver and systemd all have TPM2 support enabled and
3186 * The systemd-creds tool gained support for generating encrypted
3187 credentials that are using an empty encryption key. While this
3188 provides no integrity nor confidentiality it's useful to implement
3189 codeflows that work the same on TPM-ful and TPM2-less systems. The
3190 service manager will only accept credentials "encrypted" that way if
3191 a TPM2 device cannot be detected, to ensure that credentials
3192 "encrypted" like that cannot be used to trick TPM2 systems.
3194 * When deciding whether to colorize output, all systemd programs now
3195 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
3198 * Meson's new install_tag feature is now in use for several components,
3199 allowing to build and install select binaries only: pam, nss, devel
3200 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
3201 $ meson build systemd-boot
3202 $ meson install --tags systemd-boot --no-rebuild
3203 https://mesonbuild.com/Installing.html#installation-tags
3205 * A new build configuration option has been added, to allow selecting the
3206 default compression algorithm used by systemd-journald and systemd-coredump.
3207 This allows to build-in support for decompressing all supported formats,
3208 but choose a specific one for compression. E.g.:
3209 $ meson -Ddefault-compression=xz
3211 Experimental features:
3213 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
3214 loader.conf that implements booting Microsoft Windows from the
3215 sd-boot in a way that first reboots the system, to reset the TPM
3216 PCRs. This improves compatibility with BitLocker's TPM use, as the
3217 PCRs will only record the Windows boot process, and not sd-boot
3218 itself, thus retaining the PCR measurements not involving sd-boot.
3219 Note that this feature is experimental for now, and is likely going
3220 to be generalized and renamed in a future release, without retaining
3221 compatibility with the current implementation.
3223 * A new systemd-sysupdate component has been added that automatically
3224 discovers, downloads, and installs A/B-style updates for the host
3225 installation itself, or container images, portable service images,
3226 and other assets. See the new systemd-sysupdate man page for updates.
3228 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
3229 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
3230 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
3231 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
3232 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
3233 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
3234 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
3235 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
3236 David, David Bond, Davide Cavalca, David Tardon, davijosw,
3237 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
3238 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
3239 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
3240 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
3241 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
3242 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
3243 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
3244 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
3245 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
3246 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
3247 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
3248 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
3249 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
3250 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
3251 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
3252 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
3253 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
3254 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
3255 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
3256 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
3257 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
3258 Simon Ellmann, Sonali Srivastava, Stefan Seering,
3259 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
3260 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
3261 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
3262 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
3263 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
3264 Zbigniew Jędrzejewski-Szmek, наб
3266 — Edinburgh, 2022-05-21
3270 * Support for encrypted and authenticated credentials has been added.
3271 This extends the credential logic introduced with v247 to support
3272 non-interactive symmetric encryption and authentication, based on a
3273 key that is stored on the /var/ file system or in the TPM2 chip (if
3274 available), or the combination of both (by default if a TPM2 chip
3275 exists the combination is used, otherwise the /var/ key only). The
3276 credentials are automatically decrypted at the moment a service is
3277 started, and are made accessible to the service itself in unencrypted
3278 form. A new tool 'systemd-creds' encrypts credentials for this
3279 purpose, and two new service file settings LoadCredentialEncrypted=
3280 and SetCredentialEncrypted= configure such credentials.
3282 This feature is useful to store sensitive material such as SSL
3283 certificates, passwords and similar securely at rest and only decrypt
3284 them when needed, and in a way that is tied to the local OS
3285 installation or hardware.
3287 * systemd-gpt-auto-generator can now automatically set up discoverable
3288 LUKS2 encrypted swap partitions.
3290 * The GPT Discoverable Partitions Specification has been substantially
3291 extended with support for root and /usr/ partitions for the majority
3292 of architectures systemd supports. This includes platforms that do
3293 not natively support UEFI, because even though GPT is specified under
3294 UEFI umbrella, it is useful on other systems too. Specifically,
3295 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
3296 Portable Services use the concept without requiring UEFI.
3298 * The GPT Discoverable Partitions Specifications has been extended with
3299 a new set of partitions that may carry PKCS#7 signatures for Verity
3300 partitions, encoded in a simple JSON format. This implements a simple
3301 mechanism for building disk images that are fully authenticated and
3302 can be tested against a set of cryptographic certificates. This is
3303 now implemented for the various systemd tools that can operate with
3304 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
3305 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
3306 The PKCS#7 signatures are passed to the kernel (where they are
3307 checked against certificates from the kernel keyring), or can be
3308 verified against certificates provided in userspace (via a simple
3309 drop-in file mechanism).
3311 * systemd-dissect's inspection logic will now report for which uses a
3312 disk image is intended. Specifically, it will display whether an
3313 image is suitable for booting on UEFI or in a container (using
3314 systemd-nspawn's --image= switch), whether it can be used as portable
3315 service, or attached as system extension.
3317 * The system-extension.d/ drop-in files now support a new field
3318 SYSEXT_SCOPE= that may encode which purpose a system extension image
3319 is for: one of "initrd", "system" or "portable". This is useful to
3320 make images more self-descriptive, and to ensure system extensions
3321 cannot be attached in the wrong contexts.
3323 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
3324 be used in portable service images to indicate which unit prefixes
3327 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
3328 now is able to decode images for non-native architectures as well.
3329 This allows systemd-nspawn to boot images of non-native architectures
3330 if the corresponding user mode emulator is installed and
3331 systemd-binfmtd is running.
3333 * systemd-logind gained new settings HandlePowerKeyLongPress=,
3334 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
3335 HandleHibernateKeyLongPress= which may be used to configure actions
3336 when the relevant keys are pressed for more than 5s. This is useful
3337 on devices that only have hardware for a subset of these keys. By
3338 default, if the reboot key is pressed long the poweroff operation is
3339 now triggered, and when the suspend key is pressed long the hibernate
3340 operation is triggered. Long pressing the other two keys currently
3341 does not trigger any operation by default.
3343 * When showing unit status updates on the console during boot and
3344 shutdown, and a service is slow to start so that the cylon animation
3345 is shown, the most recent sd_notify() STATUS= text is now shown as
3346 well. Services may use this to make the boot/shutdown output easier
3347 to understand, and to indicate what precisely a service that is slow
3348 to start or stop is waiting for. In particular, the per-user service
3349 manager instance now reports what it is doing and which service it is
3350 waiting for this way to the system service manager.
3352 * The service manager will now re-execute on reception of the
3353 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
3354 only when running as PID 1. There was no signal to request this when
3355 running as per-user service manager, i.e. as any other PID than 1.
3356 SIGRTMIN+25 works for both system and user managers.
3358 * The hardware watchdog logic in PID 1 gained support for operating
3359 with the default timeout configured in the hardware, instead of
3360 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
3361 request this behavior.
3363 * A new kernel command line option systemd.watchdog_sec= is now
3364 understood which may be used to override the hardware watchdog
3365 time-out for the boot.
3367 * A new setting DefaultOOMScoreAdjust= is now supported in
3368 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
3369 to set the default process OOM score adjustment value for processes
3370 started by the service manager. For per-user service managers this
3371 now defaults to 100, but for per-system service managers is left as
3372 is. This means that by default now services forked off the user
3373 service manager are more likely to be killed by the OOM killer than
3374 system services or the managers themselves.
3376 * A new per-service setting RestrictFileSystems= as been added that
3377 restricts the file systems a service has access to by their type.
3378 This is based on the new BPF LSM of the Linux kernel. It provides an
3379 effective way to make certain API file systems unavailable to
3380 services (and thus minimizing attack surface). A new command
3381 "systemd-analyze filesystems" has been added that lists all known
3382 file system types (and how they are grouped together under useful
3385 * Services now support a new setting RestrictNetworkInterfaces= for
3386 restricting access to specific network interfaces.
3388 * Service unit files gained new settings StartupAllowedCPUs= and
3389 StartupAllowedMemoryNodes=. These are similar to their counterparts
3390 without the "Startup" prefix and apply during the boot process
3391 only. This is useful to improve boot-time behavior of the system and
3392 assign resources differently during boot than during regular
3393 runtime. This is similar to the preexisting StartupCPUWeight=
3396 * Related to this: the various StartupXYZ= settings
3397 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
3398 during shutdown. The settings not prefixed with "Startup" hence apply
3399 during regular runtime, and those that are prefixed like that apply
3400 during boot and shutdown.
3402 * A new per-unit set of conditions/asserts
3403 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
3404 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
3405 pressure is above the configured threshold, using the kernel PSI
3406 feature. For more details see systemd.unit(5) and
3407 https://docs.kernel.org/accounting/psi.html
3409 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
3410 ProtectKernelLogs=yes can now be used.
3412 * The default maximum numbers of inodes have been raised from 64k to 1M
3413 for /dev/, and from 400k to 1M for /tmp/.
3415 * The per-user service manager learnt support for communicating with
3416 systemd-oomd to acquire OOM kill information.
3418 * A new service setting ExecSearchPath= has been added that allows
3419 changing the search path for executables for services. It affects
3420 where we look for the binaries specified in ExecStart= and similar,
3421 and the specified directories are also added the $PATH environment
3422 variable passed to invoked processes.
3424 * A new setting RuntimeRandomizedExtraSec= has been added for service
3425 and scope units that allows extending the runtime time-out as
3426 configured by RuntimeMaxSec= with a randomized amount.
3428 * The syntax of the service unit settings RuntimeDirectory=,
3429 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
3430 if the specified value is now suffixed with a colon, followed by
3431 another filename, the latter will be created as symbolic link to the
3432 specified directory. This allows creating these service directories
3433 together with alias symlinks to make them available under multiple
3436 * Service unit files gained two new settings TTYRows=/TTYColumns= for
3437 configuring rows/columns of the TTY device passed to
3438 stdin/stdout/stderr of the service. This is useful to propagate TTY
3439 dimensions to a virtual machine.
3441 * A new service unit file setting ExitType= has been added that
3442 specifies when to assume a service has exited. By default systemd
3443 only watches the main process of a service. By setting
3444 ExitType=cgroup it can be told to wait for the last process in a
3447 * Automount unit files gained a new setting ExtraOptions= that can be
3448 used to configure additional mount options to pass to the kernel when
3449 mounting the autofs instance.
3451 * "Urlification" (generation of ESC sequences that generate clickable
3452 hyperlinks in modern terminals) may now be turned off altogether
3455 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
3456 settings that default to 200 and 2 s respectively. The ratelimit
3457 ensures that a path unit cannot cause PID1 to busy-loop when it is
3458 trying to trigger a service that is skipped because of a Condition*=
3459 not being satisfied. This matches the configuration and behaviour of
3462 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
3463 as a plug-in for cryptsetup. This means the plain cryptsetup command
3464 may now be used to unlock volumes set up this way.
3466 * The TPM2 logic in cryptsetup will now automatically detect systems
3467 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
3468 updates the SHA1 banks. In such a case PCR policies will be
3469 automatically bound to the latter, not the former. This makes the PCR
3470 policies reliable, but of course do not provide the same level of
3471 trust as SHA256 banks.
3473 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
3474 RSA primary keys in addition to ECC, improving compatibility with
3475 TPM2 chips that do not support ECC. RSA keys are much slower to use
3476 than ECC, and hence are only used if ECC is not available.
3478 * /etc/crypttab gained support for a new token-timeout= setting for
3479 encrypted volumes that allows configuration of the maximum time to
3480 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
3481 the logic will query the user for a regular passphrase/recovery key
3484 * Support for activating dm-integrity volumes at boot via a new file
3485 /etc/integritytab and the tool systemd-integritysetup have been
3486 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
3487 with dm-integrity instead of dm-crypt/dm-verity.
3489 * The systemd-veritysetup-generator now understands a new usrhash=
3490 kernel command line option for specifying the Verity root hash for
3491 the partition backing the /usr/ file system. A matching set of
3492 systemd.verity_usr_* kernel command line options has been added as
3493 well. These all work similar to the corresponding options for the
3496 * The sd-device API gained a new API call sd_device_get_diskseq() to
3497 return the DISKSEQ property of a device structure. The "disk
3498 sequence" concept is a new feature recently introduced to the Linux
3499 kernel that allows detecting reuse cycles of block devices, i.e. can
3500 be used to recognize when loopback block devices are reused for a
3501 different purpose or CD-ROM drives get their media changed.
3503 * A new unit systemd-boot-update.service has been added. If enabled
3504 (the default) and the sd-boot loader is detected to be installed, it
3505 is automatically updated to the newest version when out of date. This
3506 is useful to ensure the boot loader remains up-to-date, and updates
3507 automatically propagate from the OS tree in /usr/.
3509 * sd-boot will now build with SBAT by default in order to facilitate
3510 working with recent versions of Shim that require it to be present.
3512 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
3513 This is used to robustly generate boot entry titles for Windows.
3515 * A new generic target unit factory-reset.target has been added. It is
3516 hooked into systemd-logind similar in fashion to
3517 reboot/poweroff/suspend/hibernate, and is supposed to be used to
3518 initiate a factory reset operation. What precisely this operation
3519 entails is up for the implementer to decide, the primary goal of the
3520 new unit is provide a framework where to plug in the implementation
3521 and how to trigger it.
3523 * A new meson build-time option 'clock-valid-range-usec-max' has been
3524 added which takes a time in µs and defaults to 15 years. If the RTC
3525 time is noticed to be more than the specified time ahead of the
3526 built-in epoch of systemd (which by default is the release timestamp
3527 of systemd) it is assumed that the RTC is not working correctly, and
3528 the RTC is reset to the epoch. (It already is reset to the epoch when
3529 noticed to be before it.) This should increase the chance that time
3530 doesn't accidentally jump too far ahead due to faulty hardware or
3533 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
3534 which may be used to automatically save the current system time to
3535 disk in regular intervals. This is useful to maintain a roughly
3536 monotonic clock even without RTC hardware and with some robustness
3537 against abnormal system shutdown.
3539 * systemd-analyze verify gained support for a pair of new --image= +
3540 --root= switches for verifying units below a specific root
3541 directory/image instead of on the host.
3543 * systemd-analyze verify gained support for verifying unit files under
3544 an explicitly specified unit name, independently of what the filename
3547 * systemd-analyze verify gained a new switch --recursive-errors= which
3548 controls whether to only fail on errors found in the specified units
3549 or recursively any dependent units.
3551 * systemd-analyze security now supports a new --offline mode for
3552 analyzing unit files stored on disk instead of loaded units. It may
3553 be combined with --root=/--image to analyze unit files under a root
3554 directory or disk image. It also learnt a new --threshold= parameter
3555 for specifying an exposure level threshold: if the exposure level
3556 exceeds the specified value the call will fail. It also gained a new
3557 --security-policy= switch for configuring security policies to
3558 enforce on the units. A policy is a JSON file that lists which tests
3559 shall be weighted how much to determine the overall exposure
3560 level. Altogether these new features are useful for fully automatic
3561 analysis and enforcement of security policies on unit files.
3563 * systemd-analyze security gain a new --json= switch for JSON output.
3565 * systemd-analyze learnt a new --quiet switch for reducing
3566 non-essential output. It's honored by the "dot", "syscall-filter",
3567 "filesystems" commands.
3569 * systemd-analyze security gained a --profile= option that can be used
3570 to take into account a portable profile when analyzing portable
3571 services, since a lot of the security-related settings are enabled
3574 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
3575 files, binaries and executables and prints metadata information,
3576 including the build-id and other info described on:
3577 https://systemd.io/COREDUMP_PACKAGE_METADATA/
3579 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
3580 section, for automatically propagating DNS settings from other
3583 * The static lease DHCP server logic in systemd-networkd may now serve
3584 IP addresses outside of the configured IP pool range for the server.
3586 * CAN support in systemd-networkd gained four new settings Loopback=,
3587 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
3588 control modes. It gained a number of further settings for tweaking
3591 * The [CAN] section in .network file gained new TimeQuantaNSec=,
3592 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
3593 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
3594 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
3595 DataSyncJumpWidth= settings to control bit-timing processed by the
3598 * DHCPv4 client support in systemd-networkd learnt a new Label= option
3599 for configuring the address label to apply to configure IPv4
3602 * The [IPv6AcceptRA] section of .network files gained support for a new
3603 UseMTU= setting that may be used to control whether to apply the
3604 announced MTU settings to the local interface.
3606 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
3607 setting to control whether the DHCPv4 client request and process the
3610 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
3611 [DHCPPrefixDelegation], as now the prefix delegation is also supported
3612 with DHCPv4 protocol by enabling the Use6RD= setting.
3614 * The [DHCPPrefixDelegation] section in .network file gained a new
3615 setting UplinkInterface= to specify the upstream interface.
3617 * The [DHCPv6] section in .network file gained a new setting
3618 UseDelegatedPrefix= to control whether the delegated prefixes will be
3619 propagated to the downstream interfaces.
3621 * The [IPv6AcceptRA] section of .network files now understands two new
3622 settings UseGateway=/UseRoutePrefix= for explicitly configuring
3623 whether to use the relevant fields from the IPv6 Router Advertisement
3626 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
3627 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
3628 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
3629 [IPv6AcceptRA] section to control when the DHCPv6 client is started
3630 and how the delegated prefixes are handled by the DHCPv6 client.
3632 * The IPv6Token= section in the [Network] section is deprecated, and
3633 the [IPv6AcceptRA] section gained the Token= setting for its
3634 replacement. The [IPv6Prefix] section also gained the Token= setting.
3635 The Token= setting gained 'eui64' mode to explicitly configure an
3636 address with the EUI64 algorithm based on the interface MAC address.
3637 The 'prefixstable' mode can now optionally take a secret key. The
3638 Token= setting in the [DHCPPrefixDelegation] section now supports all
3639 algorithms supported by the same settings in the other sections.
3641 * The [RoutingPolicyRule] section of .network file gained a new
3642 SuppressInterfaceGroup= setting.
3644 * The IgnoreCarrierLoss= setting in the [Network] section of .network
3645 files now allows a duration to be specified, controlling how long to
3646 wait before reacting to carrier loss.
3648 * The [DHCPServer] section of .network file gained a new Router=
3649 setting to specify the router address.
3651 * The [CAKE] section of .network files gained various new settings
3652 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
3653 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
3654 and UseRawPacketSize= for configuring CAKE.
3656 * systemd-networkd now ships with new default .network files:
3657 80-container-vb.network which matches host-side network bridge device
3658 created by systemd-nspawn's --network-bridge or --network-zone
3659 switch, and 80-6rd-tunnel.network which matches automatically created
3660 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
3662 * systemd-networkd's handling of Endpoint= resolution for WireGuard
3663 interfaces has been improved.
3665 * systemd-networkd will now automatically configure routes to addresses
3666 specified in AllowedIPs=. This feature can be controlled via
3667 RouteTable= and RouteMetric= settings in [WireGuard] or
3668 [WireGuardPeer] sections.
3670 * systemd-networkd will now once again automatically generate persistent
3671 MAC addresses for batadv and bridge interfaces. Users can disable this
3672 by using MACAddress=none in .netdev files.
3674 * systemd-networkd and systemd-udevd now support IP over InfiniBand
3675 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
3676 systemd.netdev files gained the [IPoIB] section.
3678 * systemd-networkd and systemd-udevd now support net.ifname-policy=
3679 option on the kernel command-line. This is implemented through the
3680 systemd-network-generator service that automatically generates
3681 appropriate .link, .network, and .netdev files.
3683 * The various systemd-udevd "ethtool" buffer settings now understand
3684 the special value "max" to configure the buffers to the maximum the
3687 * systemd-udevd's .link files may now configure a large variety of
3688 NIC coalescing settings, plus more hardware offload settings.
3690 * .link files gained a new WakeOnLanPassword= setting in the [Link]
3691 section that allows to specify a WoL "SecureOn" password on hardware
3694 * systemd-nspawn's --setenv= switch now supports an additional syntax:
3695 if only a variable name is specified (i.e. without being suffixed by
3696 a '=' character and a value) the current value of the environment
3697 variable is propagated to the container. e.g. --setenv=FOO will
3698 lookup the current value of $FOO in the environment, and pass it down
3699 to the container. Similar behavior has been added to homectl's,
3700 machinectl's and systemd-run's --setenv= switch.
3702 * systemd-nspawn gained a new switch --suppress-sync= which may be used
3703 to optionally suppress the effect of the sync()/fsync()/fdatasync()
3704 system calls for the container payload. This is useful for build
3705 system environments where safety against abnormal system shutdown is
3706 not essential as all build artifacts can be regenerated any time, but
3707 the performance win is beneficial.
3709 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
3710 same value that PID 1 uses for most forked off processes.
3712 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
3713 uidmap/nouidmap options as last parameter. If "uidmap" is used the
3714 bind mounts are created with UID mapping taking place that ensures
3715 the host's file ownerships are mapped 1:1 to container file
3716 ownerships, even if user namespacing is used. This way
3717 files/directories bound into containers will no longer show up as
3718 owned by the nobody user as they typically did if no special care was
3719 taken to shift them manually.
3721 * When discovering Windows installations sd-boot will now attempt to
3722 show the Windows version.
3724 * The color scheme to use in sd-boot may now be configured at
3727 * sd-boot gained the ability to change screen resolution during
3728 boot-time, by hitting the "r" key. This will cycle through available
3729 resolutions and save the last selection.
3731 * sd-boot learnt a new hotkey "f". When pressed the system will enter
3732 firmware setup. This is useful in environments where it is difficult
3733 to hit the right keys early enough to enter the firmware, and works
3734 on any firmware regardless which key it natively uses.
3736 * sd-boot gained support for automatically booting into the menu item
3737 selected on the last boot (using the "@saved" identifier for menu
3740 * sd-boot gained support for automatically loading all EFI drivers
3741 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
3742 Partition (ESP). These drivers are loaded before the menu entries are
3743 loaded. This is useful e.g. to load additional file system drivers
3744 for the XBOOTLDR partition.
3746 * systemd-boot will now paint the input cursor on its own instead of
3747 relying on the firmware to do so, increasing compatibility with broken
3748 firmware that doesn't make the cursor reasonably visible.
3750 * sd-boot now embeds a .osrel PE section like we expect from Boot
3751 Loader Specification Type #2 Unified Kernels. This means sd-boot
3752 itself may be used in place of a Type #2 Unified Kernel. This is
3753 useful for debugging purposes as it allows chain-loading one a
3754 (development) sd-boot instance from another.
3756 * sd-boot now supports a new "devicetree" field in Boot Loader
3757 Specification Type #1 entries: if configured the specified device
3758 tree file is installed before the kernel is invoked. This is useful
3759 for installing/applying new devicetree files without updating the
3762 * Similarly, sd-stub now can read devicetree data from a PE section
3763 ".dtb" and apply it before invoking the kernel.
3765 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
3766 gained the ability to pick up credentials and sysext files, wrap them
3767 in a cpio archive, and pass as an additional initrd to the invoked
3768 Linux kernel, in effect placing those files in the /.extra/ directory
3769 of the initrd environment. This is useful to implement trusted initrd
3770 environments which are fully authenticated but still can be extended
3771 (via sysexts) and parameterized (via encrypted/authenticated
3772 credentials, see above).
3774 Credentials can be located next to the kernel image file (credentials
3775 specific to a single boot entry), or in one of the shared directories
3776 (credentials applicable to multiple boot entries).
3778 * sd-stub now comes with a full man page, that explains its feature set
3779 and how to combine a kernel image, an initrd and the stub to build a
3780 complete EFI unified kernel image, implementing Boot Loader
3781 Specification Type #2.
3783 * sd-stub may now provide the initrd to the executed kernel via the
3784 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
3785 non-x86 architectures.
3787 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
3788 may be used to set the boot menu time-out of the boot loader (for all
3789 or just the subsequent boot).
3791 * bootctl and kernel-install will now read variables
3792 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
3793 /etc/kernel/install.conf. When set, it specifies the layout to use
3794 for installation directories on the boot partition, so that tools
3795 don't need to guess it based on the already-existing directories. The
3796 only value that is defined natively is "bls", corresponding to the
3798 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
3799 kernel-install that implement a different layout can declare other
3800 values for this variable.
3802 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
3803 assumption that if the user installed sd-boot to the ESP, they intend
3804 to use the entry layout understood by sd-boot. It'll also write
3805 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
3806 (and it wasn't specified in the config file yet). Similarly,
3807 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
3808 wasn't specified in the config file yet). Effectively, those changes
3809 mean that the machine-id used for boot loader entry installation is
3810 "frozen" upon first use and becomes independent of the actual
3813 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
3814 images created for distribution ("golden images") are built with no
3815 machine-id, so that a unique machine-id can be created on the first
3816 boot. But those images may contain boot loader entries with the
3817 machine-id used during build included in paths. Using a "frozen"
3818 value allows unambiguously identifying entries that match the
3819 specific installation, while still permitting parallel installations
3822 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
3823 kernel-install to guess the installation layout. This fixes the
3824 problem where a (possibly empty) directory in the boot partition is
3825 created from a different layout causing kernel-install plugins to
3826 assume the wrong layout. A particular example of how this may happen
3827 is the grub2 package in Fedora which includes directories under /boot
3828 directly in its file list. Various other packages pull in grub2 as a
3829 dependency, so it may be installed even if unused, breaking
3830 installations that use the bls layout.
3832 * bootctl and systemd-bless-boot can now be linked statically.
3834 * systemd-sysext now optionally doesn't insist on extension-release.d/
3835 files being placed in the image under the image's file name. If the
3836 file system xattr user.extension-release.strict is set on the
3837 extension release file, it is accepted regardless of its name. This
3838 relaxes security restrictions a bit, as system extension may be
3839 attached under a wrong name this way.
3841 * udevadm's test-builtin command learnt a new --action= switch for
3842 testing the built-in with the specified action (in place of the
3845 * udevadm info gained new switches --property=/--value for showing only
3846 specific udev properties/values instead of all.
3848 * A new hwdb database has been added that contains matches for various
3849 types of signal analyzers (protocol analyzers, logic analyzers,
3850 oscilloscopes, multimeters, bench power supplies, etc.) that should
3851 be accessible to regular users.
3853 * A new hwdb database entry has been added that carries information
3854 about types of cameras (regular or infrared), and in which direction
3855 they point (front or back).
3857 * A new rule to allow console users access to rfkill by default has been
3860 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
3861 now also owned by the system group "sgx".
3863 * A new build-time meson option "extra-net-naming-schemes=" has been
3864 added to define additional naming schemes for udev's network
3865 interface naming logic. This is useful for enterprise distributions
3866 and similar which want to pin the schemes of certain distribution
3867 releases under a specific name and previously had to patch the
3868 sources to introduce new named schemes.
3870 * The predictable naming logic for network interfaces has been extended
3871 to generate stable names from Xen netfront device information.
3873 * hostnamed's chassis property can now be sourced from chassis-type
3874 field encoded in devicetree (in addition to the existing DMI
3877 * systemd-cgls now optionally displays cgroup IDs and extended
3878 attributes for each cgroup. (Controllable via the new --xattr= +
3879 --cgroup-id= switches.)
3881 * coredumpctl gained a new --all switch for operating on all
3882 Journal files instead of just the local ones.
3884 * systemd-coredump will now use libdw/libelf via dlopen() rather than
3885 directly linking, allowing users to easily opt-out of backtrace/metadata
3886 analysis of core files, and reduce image sizes when this is not needed.
3888 * systemd-coredump will now analyze core files with libdw/libelf in a
3889 forked, sandboxed process.
3891 * systemd-homed will now try to unmount an activate home area in
3892 regular intervals once the user logged out fully. Previously this was
3893 attempted exactly once but if the home directory was busy for some
3894 reason it was not tried again.
3896 * systemd-homed's LUKS2 home area backend will now create a BSD file
3897 system lock on the image file while the home area is active
3898 (i.e. mounted). If a home area is found to be locked, logins are
3899 politely refused. This should improve behavior when using home areas
3900 images that are accessible via the network from multiple clients, and
3901 reduce the chance of accidental file system corruption in that case.
3903 * Optionally, systemd-homed will now drop the kernel buffer cache once
3904 a user has fully logged out, configurable via the new --drop-caches=
3907 * systemd-homed now makes use of UID mapped mounts for the home areas.
3908 If the kernel and used file system support it, files are now
3909 internally owned by the "nobody" user (i.e. the user typically used
3910 for indicating "this ownership is not mapped"), and dynamically
3911 mapped to the UID used locally on the system via the UID mapping
3912 mount logic of recent kernels. This makes migrating home areas
3913 between different systems cheaper because recursively chown()ing file
3914 system trees is no longer necessary.
3916 * systemd-homed's CIFS backend now optionally supports CIFS service
3917 names with a directory suffix, in order to place home directories in
3918 a subdirectory of a CIFS share, instead of the top-level directory.
3920 * systemd-homed's CIFS backend gained support for specifying additional
3921 mount options in the JSON user record (cifsExtraMountOptions field,
3922 and --cifs-extra-mount-options= homectl switch). This is for example
3923 useful for configuring mount options such as "noserverino" that some
3924 SMB3 services require (use that to run a homed home directory from a
3925 FritzBox SMB3 share this way).
3927 * systemd-homed will now default to btrfs' zstd compression for home
3928 areas. This is inspired by Fedora's recent decision to switch to zstd
3931 * Additional mount options to use when mounting the file system of
3932 LUKS2 volumes in systemd-homed has been added. Via the
3933 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
3934 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
3935 systemd-homed or via the luksExtraMountOptions user record JSON
3936 property. (Exposed via homectl --luks-extra-mount-options)
3938 * homectl's resize command now takes the special size specifications
3939 "min" and "max" to shrink/grow the home area to the minimum/maximum
3940 size possible, taking disk usage/space constraints and file system
3941 limitations into account. Resizing is now generally graceful: the
3942 logic will try to get as close to the specified size as possible, but
3943 not consider it a failure if the request couldn't be fulfilled
3946 * systemd-homed gained the ability to automatically shrink home areas
3947 on logout to their minimal size and grow them again on next
3948 login. This ensures that while inactive, a home area only takes up
3949 the minimal space necessary, but once activated, it provides
3950 sufficient space for the user's needs. This behavior is only
3951 supported if btrfs is used as file system inside the home area
3952 (because only for btrfs online growing/shrinking is implemented in
3953 the kernel). This behavior is now enabled by default, but may be
3954 controlled via the new --auto-resize-mode= setting of homectl.
3956 * systemd-homed gained support for automatically re-balancing free disk
3957 space among active home areas, in case the LUKS2 backends are used,
3958 and no explicit disk size was requested. This way disk space is
3959 automatically managed and home areas resized in regular intervals and
3960 manual resizing when disk space becomes scarce should not be
3961 necessary anymore. This behavior is only supported if btrfs is used
3962 within the home areas (as only then online shrinking and growing is
3963 supported), and may be configured via the new rebalanceWeight JSON
3964 user record field (as exposed via the new --rebalance-weight= homectl
3965 setting). Re-balancing is mostly automatic, but can also be requested
3966 explicitly via "homectl rebalance", which is synchronous, and thus
3967 may be used to wait until the rebalance run is complete.
3969 * userdbctl gained a --json= switch for configured the JSON formatting
3970 to use when outputting user or group records.
3972 * userdbctl gained a new --multiplexer= switch for explicitly
3973 configuring whether to use the systemd-userdbd server side user
3974 record resolution logic.
3976 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
3977 for chaining up another command to execute after completing the
3978 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
3979 configuration of a single command to invoke, this maybe used to
3980 invoke multiple: first userdbctl's own implementation, and then any
3981 other also configured in the command line.
3983 * The sd-event API gained a new function sd_event_add_inotify_fd() that
3984 is similar to sd_event_add_inotify() but accepts a file descriptor
3985 instead of a path in the file system for referencing the inode to
3988 * The sd-event API gained a new function
3989 sd_event_source_set_ratelimit_expire_callback() that may be used to
3990 define a callback function that is called whenever an event source
3991 leaves the rate limiting phase.
3993 * New documentation has been added explaining which steps are necessary
3994 to port systemd to a new architecture:
3996 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
3998 * The x-systemd.makefs option in /etc/fstab now explicitly supports
3999 ext2, ext3, and f2fs file systems.
4001 * Mount units and units generated from /etc/fstab entries with 'noauto'
4002 are now ordered the same as other units. Effectively, they will be
4003 started earlier (if something actually pulled them in) and stopped
4004 later, similarly to normal mount units that are part of
4005 fs-local.target. This change should be invisible to users, but
4006 should prevent those units from being stopped too early during
4009 * The systemd-getty-generator now honors a new kernel command line
4010 argument systemd.getty_auto= and a new environment variable
4011 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
4012 example useful to turn off gettys inside of containers or similar
4015 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
4016 (in addition to 127.0.0.53, as before). If DNS requests are sent to
4017 this address they are propagated in "bypass" mode only, i.e. are
4018 almost not processed locally, but mostly forwarded as-is to the
4019 current upstream DNS servers. This provides a stable DNS server
4020 address that proxies all requests dynamically to the right upstream
4021 DNS servers even if these dynamically change. This stub does not do
4022 mDNS/LLMNR resolution. However, it will translate look-ups to
4023 DNS-over-TLS if necessary. This new stub is particularly useful in
4024 container/VM environments, or for tethering setups: use DNAT to
4025 redirect traffic to any IP address to this stub.
4027 * systemd-importd now honors new environment variables
4028 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
4029 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
4030 generation, btrfs quota setup and disk synchronization.
4032 * systemd-importd and systemd-resolved can now be optionally built with
4033 OpenSSL instead of libgcrypt.
4035 * systemd-repart no longer requires OpenSSL.
4037 * systemd-sysusers will no longer create the redundant 'nobody' group
4038 by default, as the 'nobody' user is already created with an
4039 appropriate primary group.
4041 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
4043 * systemctl show-environment gained support for --output=json.
4045 * pam_systemd will now first try to use the X11 abstract socket, and
4046 fallback to the socket file in /tmp/.X11-unix/ only if that does not
4049 * systemd-journald will no longer go back to volatile storage
4050 regardless of configuration when its unit is restarted.
4052 * Initial support for the LoongArch architecture has been added (system
4053 call lists, GPT partition table UUIDs, etc).
4055 * systemd-journald's own logging messages are now also logged to the
4056 journal itself when systemd-journald logs to /dev/kmsg.
4058 * systemd-journald now re-enables COW for archived journal files on
4059 filesystems that support COW. One benefit of this change is that
4060 archived journal files will now get compressed on btrfs filesystems
4061 that have compression enabled.
4063 * systemd-journald now deduplicates fields in a single log message
4064 before adding it to the journal. In archived journal files, it will
4065 also punch holes for unused parts and truncate the file as
4066 appropriate, leading to reductions in disk usage.
4068 * journalctl --verify was extended with more informative error
4071 * More of sd-journal's functions are now resistant against journal file
4074 * The shutdown command learnt a new option --show, to display the
4077 * A LICENSES/ directory is now included in the git tree. It contains a
4078 README.md file that explains the licenses used by source files in
4079 this repository. It also contains the text of all applicable
4080 licenses as they appear on spdx.org.
4082 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
4083 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
4084 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
4085 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
4086 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
4087 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
4088 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
4089 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
4090 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
4091 Christian Brauner, Christian Göttsche, Christian Wehrli,
4092 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
4093 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
4094 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
4095 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
4096 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
4097 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
4098 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
4099 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
4100 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
4101 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
4102 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
4103 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
4104 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
4105 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
4106 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
4107 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
4108 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
4109 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
4110 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
4111 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
4112 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
4113 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
4114 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
4115 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
4116 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
4117 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
4118 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
4119 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
4120 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
4121 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
4122 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
4123 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
4124 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
4125 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
4126 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
4127 Дамјан Георгиевски, наб
4129 — Warsaw, 2021-12-23
4133 * When operating on disk images via the --image= switch of various
4134 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
4135 no 'root=' parameter on the kernel command line, and multiple
4136 suitable root or /usr/ partitions exist in the image, then a simple
4137 comparison inspired by strverscmp() is done on the GPT partition
4138 label, and the newest partition is picked. This permits a simple and
4139 generic whole-file-system A/B update logic where new operating system
4140 versions are dropped into partitions whose label is then updated with
4141 a matching version identifier.
4143 * systemd-sysusers now supports querying the passwords to set for the
4144 users it creates via the "credentials" logic introduced in v247: the
4145 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
4146 credentials are consulted for the password to use (either in UNIX
4147 hashed form, or literally). By default these credentials are inherited
4148 down from PID1 (which in turn imports it from a container manager if
4149 there is one). This permits easy configuration of user passwords
4150 during first boot. Example:
4152 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
4154 Note that systemd-sysusers operates in purely additive mode: it
4155 executes no operation if the declared users already exist, and hence
4156 doesn't set any passwords as effect of the command line above if the
4157 specified root user exists already in the image. (Note that
4158 --volatile=yes ensures it doesn't, though.)
4160 * systemd-firstboot now also supports querying various system
4161 parameters via the credential subsystems. Thus, as above this may be
4162 used to initialize important system parameters on first boot of
4163 previously unprovisioned images (i.e. images with a mostly empty
4166 * PID 1 may now show both the unit name and the unit description
4167 strings in its status output during boot. This may be configured with
4168 StatusUnitFormat=combined in system.conf or
4169 systemd.status-unit-format=combined on the kernel command line.
4171 * The systemd-machine-id-setup tool now supports a --image= switch for
4172 provisioning a machine ID file into an OS disk image, similar to how
4173 --root= operates on an OS file tree. This matches the existing switch
4174 of the same name for systemd-tmpfiles, systemd-firstboot, and
4175 systemd-sysusers tools.
4177 * Similarly, systemd-repart gained support for the --image= switch too.
4178 In combination with the existing --size= option, this makes the tool
4179 particularly useful for easily growing disk images in a single
4180 invocation, following the declarative rules included in the image
4183 * systemd-repart's partition configuration files gained support for a
4184 new switch MakeDirectories= which may be used to create arbitrary
4185 directories inside file systems that are created, before registering
4186 them in the partition table. This is useful in particular for root
4187 partitions to create mount point directories for other partitions
4188 included in the image. For example, a disk image that contains a
4189 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
4190 create /home/ and /var/ as empty directories in the root file system
4191 on its creation, so that the resulting image can be mounted
4192 immediately, even in read-only mode.
4194 * systemd-repart's CopyBlocks= setting gained support for the special
4195 value "auto". If used, a suitable matching partition on the booted OS
4196 is found as source to copy blocks from. This is useful when
4197 implementing replicating installers, that are booted from one medium
4198 and then stream their own root partition onto the target medium.
4200 * systemd-repart's partition configuration files gained support for a
4201 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
4202 GPT partition flags for the created partitions: this is useful for
4203 marking newly created partitions as read-only, or as not being
4204 subject for automatic mounting from creation on.
4206 * The /etc/os-release file has been extended with two new (optional)
4207 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
4208 information for OS images that are updated comprehensively and
4209 atomically as one image. Two new specifiers %M, %A now resolve to
4210 these two fields in the various configuration options that resolve
4213 * portablectl gained a new switch --extension= for enabling portable
4214 service images with extensions that follow the extension image
4215 concept introduced with v248, and thus allows layering multiple
4216 images when setting up the root filesystem of the service.
4218 * systemd-coredump will now extract ELF build-id information from
4219 processes dumping core and include it in the coredump report.
4220 Moreover, it will look for ELF .note.package sections with
4221 distribution packaging meta-information about the crashing process.
4222 This is useful to directly embed the rpm or deb (or any other)
4223 package name and version in ELF files, making it easy to match
4224 coredump reports with the specific package for which the software was
4225 compiled. This is particularly useful on environments with ELF files
4226 from multiple vendors, different distributions and versions, as is
4227 common today in our containerized and sand-boxed world. For further
4230 https://systemd.io/COREDUMP_PACKAGE_METADATA
4232 * A new udev hardware database has been added for FireWire devices
4235 * The "net_id" built-in of udev has been updated with three
4236 backwards-incompatible changes:
4238 - PCI hotplug slot names on s390 systems are now parsed as
4239 hexadecimal numbers. They were incorrectly parsed as decimal
4240 previously, or ignored if the name was not a valid decimal
4243 - PCI onboard indices up to 65535 are allowed. Previously, numbers
4244 above 16383 were rejected. This primarily impacts s390 systems,
4245 where values up to 65535 are used.
4247 - Invalid characters in interface names are replaced with "_".
4249 The new version of the net naming scheme is "v249". The previous
4250 scheme can be selected via the "net.naming-scheme=v247" kernel
4251 command line parameter.
4253 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
4254 NULL bus object, for which they will return false. Or in other words,
4255 an unallocated bus connection is neither ready nor open.
4257 * The sd-device API acquired a new API function
4258 sd_device_get_usec_initialized() that returns the monotonic time when
4259 the udev device first appeared in the database.
4261 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
4262 sd_device_get_trigger_uuid(). The former is similar to
4263 sd_device_trigger() but returns a randomly generated UUID that is
4264 associated with the synthetic uevent generated by the call. This UUID
4265 may be read from the sd_device object a monitor eventually receives,
4266 via the sd_device_get_trigger_uuid(). This interface requires kernel
4267 4.13 or above to work, and allows tracking a synthetic uevent through
4268 the entire device management stack. The "udevadm trigger --settle"
4269 logic has been updated to make use of this concept if available to
4270 wait precisely for the uevents it generates. "udevadm trigger" also
4271 gained a new parameter --uuid that prints the UUID for each generated
4274 * sd-device also gained new APIs sd_device_new_from_ifname() and
4275 sd_device_new_from_ifindex() for allocating an sd-device object for
4276 the specified network interface. The former accepts an interface name
4277 (either a primary or an alternative name), the latter an interface
4280 * The native Journal protocol has been documented. Clients may talk
4281 this as alternative to the classic BSD syslog protocol for locally
4282 delivering log records to the Journal. The protocol has been stable
4283 for a long time and in fact been implemented already in a variety
4284 of alternative client libraries. This documentation makes the support
4287 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
4289 * A new BPFProgram= setting has been added to service files. It may be
4290 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
4291 file, or a bind mount or symlink to one. This may be used to upload
4292 and manage BPF programs externally and then hook arbitrary systemd
4295 * The "home.arpa" domain that has been officially declared as the
4296 choice for domain for local home networks per RFC 8375 has been added
4297 to the default NTA list of resolved, since DNSSEC is generally not
4298 available on private domains.
4300 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
4302 * A new ManageForeignRoutingPolicyRules= setting has been added to
4303 .network files which may be used to exclude foreign-created routing
4304 policy rules from systemd-networkd management.
4306 * systemd-network-wait-online gained two new switches -4 and -6 that
4307 may be used to tweak whether to wait for only IPv4 or only IPv6
4310 * .network files gained a new RequiredFamilyForOnline= setting to
4311 fine-tune whether to require an IPv4 or IPv6 address in order to
4312 consider an interface "online".
4314 * networkctl will now show an over-all "online" state in the per-link
4317 * In .network files a new OutgoingInterface= setting has been added to
4318 specify the output interface in bridge FDB setups.
4320 * In .network files the Multipath group ID may now be configured for
4321 [NextHop] entries, via the new Group= setting.
4323 * The DHCP server logic configured in .network files gained a new
4324 setting RelayTarget= that turns the server into a DHCP server relay.
4325 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
4326 to further tweak the DHCP relay behaviour.
4328 * The DHCP server logic also gained a new ServerAddress= setting in
4329 .network files that explicitly specifies the server IP address to
4330 use. If not specified, the address is determined automatically, as
4333 * The DHCP server logic in systemd-networkd gained support for static
4334 DHCP leases, configurable via the [DHCPServerStaticLease]
4335 section. This allows explicitly mapping specific MAC addresses to
4336 fixed IP addresses and vice versa.
4338 * The RestrictAddressFamilies= setting in service files now supports a
4339 new special value "none". If specified sockets of all address
4340 families will be made unavailable to services configured that way.
4342 * systemd-fstab-generator and systemd-repart have been updated to
4343 support booting from disks that carry only a /usr/ partition but no
4344 root partition yet, and where systemd-repart can add it in on the
4345 first boot. This is useful for implementing systems that ship with a
4346 single /usr/ file system, and whose root file system shall be set up
4347 and formatted on a LUKS-encrypted volume whose key is generated
4348 locally (and possibly enrolled in the TPM) during the first boot.
4350 * The [Address] section of .network files now accepts a new
4351 RouteMetric= setting that configures the routing metric to use for
4352 the prefix route created as effect of the address configuration.
4353 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
4354 gained matching settings for their prefix routes. (The option of the
4355 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
4356 it conceptually belongs there; the old option is still understood for
4359 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
4362 * A new udev property ID_NET_DHCP_BROADCAST on network interface
4363 devices is now honoured by systemd-networkd, controlling whether to
4364 issue DHCP offers via broadcasting. This is used to ensure that s390
4365 layer 3 network interfaces work out-of-the-box with systemd-networkd.
4367 * nss-myhostname and systemd-resolved will now synthesize address
4368 records for a new special hostname "_outbound". The name will always
4369 resolve to the local IP addresses most likely used for outbound
4370 connections towards the default routes. On multi-homed hosts this is
4371 useful to have a stable handle referring to "the" local IP address
4372 that matters most, to the point where this is defined.
4374 * The Discoverable Partition Specification has been updated with a new
4375 GPT partition flag "grow-file-system" defined for its partition
4376 types. Whenever partitions with this flag set are automatically
4377 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
4378 of systemd-nspawn or other tools; and as opposed to explicit mounting
4379 via /etc/fstab), the file system within the partition is
4380 automatically grown to the full size of the partition. If the file
4381 system size already matches the partition size this flag has no
4382 effect. Previously, this functionality has been available via the
4383 explicit x-systemd.growfs mount option, and this new flag extends
4384 this to automatically discovered mounts. A new GrowFileSystem=
4385 setting has been added to systemd-repart drop-in files that allows
4386 configuring this partition flag. This new flag defaults to on for
4387 partitions automatically created by systemd-repart, except if they
4388 are marked read-only. See the specification for further details:
4390 https://systemd.io/DISCOVERABLE_PARTITIONS
4392 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
4393 section. If enabled (which is the default), and an NTP server address
4394 is acquired through a DHCP lease on this interface an explicit route
4395 to this address is created on this interface to ensure that NTP
4396 traffic to the NTP server acquired on an interface is also routed
4397 through that interface. The pre-existing RoutesToDNS= setting that
4398 implements the same for DNS servers is now enabled by default.
4400 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
4401 been added that may be used to restrict the network interfaces
4402 sockets created by the service may be bound to. This is implemented
4405 * A new ConditionFirmware= setting has been added to unit files to
4406 conditionalize on certain firmware features. At the moment it may
4407 check whether running on a UEFI system, a device.tree system, or if
4408 the system is compatible with some specified device-tree feature.
4410 * A new ConditionOSRelease= setting has been added to unit files to
4411 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
4412 operators may be used to check if some field has some specific value
4413 or do an alphanumerical comparison. Equality comparisons are useful
4414 for fields like ID, but relative comparisons for fields like
4415 VERSION_ID or IMAGE_VERSION.
4417 * hostnamed gained a new Describe() D-Bus method that returns a JSON
4418 serialization of the host data it exposes. This is exposed via
4419 "hostnamectl --json=" to acquire a host identity description in JSON.
4420 It's our intention to add a similar features to most services and
4421 objects systemd manages, in order to simplify integration with
4422 program code that can consume JSON.
4424 * Similarly, networkd gained a Describe() method on its Manager and
4425 Link bus objects. This is exposed via "networkctl --json=".
4427 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
4428 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
4429 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
4430 that is used both to get the value (when no argument is given), and
4431 to set the value (when an argument is specified). The old names
4432 continue to be supported for compatibility.
4434 * systemd-detect-virt and ConditionVirtualization= are now able to
4435 correctly identify Amazon EC2 environments.
4437 * The LogLevelMax= setting of unit files now applies not only to log
4438 messages generated *by* the service, but also to log messages
4439 generated *about* the service by PID 1. To suppress logs concerning a
4440 specific service comprehensively, set this option to a high log
4443 * bootctl gained support for a new --make-machine-id-directory= switch
4444 that allows precise control on whether to create the top-level
4445 per-machine directory in the boot partition that typically contains
4446 Type 1 boot loader entries.
4448 * During build SBAT data to include in the systemd-boot EFI PE binaries
4449 may be specified now.
4451 * /etc/crypttab learnt a new option "headless". If specified any
4452 requests to query the user interactively for passwords or PINs will
4453 be skipped. This is useful on systems that are headless, i.e. where
4454 an interactive user is generally not present.
4456 * /etc/crypttab also learnt a new option "password-echo=" that allows
4457 configuring whether the encryption password prompt shall echo the
4458 typed password and if so, do so literally or via asterisks. (The
4459 default is the same behaviour as before: provide echo feedback via
4462 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
4463 systemd-homed has been updated to allow explicit configuration of the
4464 "user presence" and "user verification" checks, as well as whether a
4465 PIN is required for authentication, via the new switches
4466 --fido2-with-user-presence=, --fido2-with-user-verification=,
4467 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
4468 features are available, and may be enabled or disabled depends on the
4471 * systemd-nspawn's --private-user= switch now accepts the special value
4472 "identity" which configures a user namespacing environment with an
4473 identity mapping of 65535 UIDs. This means the container UID 0 is
4474 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
4475 this doesn't appear to be useful, however it does reduce the attack
4476 surface a bit, since the resulting container will possess process
4477 capabilities only within its namespace and not on the host.
4479 * systemd-nspawn's --private-user-chown switch has been replaced by a
4480 more generic --private-user-ownership= switch that accepts one of
4481 three values: "chown" is equivalent to the old --private-user-chown,
4482 and "off" is equivalent to the absence of the old switch. The value
4483 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
4484 of files and directories of the underlying image to the chosen UID
4485 range for the container. "auto" is equivalent to "map" if UID mapping
4486 mount are supported, otherwise it is equivalent to "chown". The short
4487 -U switch systemd-nspawn now implies --private-user-ownership=auto
4488 instead of the old --private-user-chown. Effectively this means: if
4489 the backing file system supports UID mapping mounts the feature is
4490 now used by default if -U is used. Generally, it's a good idea to use
4491 UID mapping mounts instead of recursive chown()ing, since it allows
4492 running containers off immutable images (since no modifications of
4493 the images need to take place), and share images between multiple
4494 instances. Moreover, the recursive chown()ing operation is slow and
4495 can be avoided. Conceptually it's also a good thing if transient UID
4496 range uses do not leak into persistent file ownership anymore. TLDR:
4497 finally, the last major drawback of user namespacing has been
4498 removed, and -U should always be used (unless you use btrfs, where
4499 UID mapped mounts do not exist; or your container actually needs
4500 privileges on the host).
4502 * nss-systemd now synthesizes user and group shadow records in addition
4503 to the main user and group records. Thus, hashed passwords managed by
4504 systemd-homed are now accessible via the shadow database.
4506 * The userdb logic (and thus nss-systemd, and so on) now read
4507 additional user/group definitions in JSON format from the drop-in
4508 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
4509 /usr/lib/userdb/. This is a simple and powerful mechanism for making
4510 additional users available to the system, with full integration into
4511 NSS including the shadow databases. Since the full JSON user/group
4512 record format is supported this may also be used to define users with
4513 resource management settings and other runtime settings that
4514 pam_systemd and systemd-logind enforce at login.
4516 * The userdbctl tool gained two new switches --with-dropin= and
4517 --with-varlink= which can be used to fine-tune the sources used for
4518 user database lookups.
4520 * systemd-nspawn gained a new switch --bind-user= for binding a host
4521 user account into the container. This does three things: the user's
4522 home directory is bind mounted from the host into the container,
4523 below the /run/userdb/home/ hierarchy. A free UID is picked in the
4524 container, and a user namespacing UID mapping to the host user's UID
4525 installed. And finally, a minimal JSON user and group record (along
4526 with its hashed password) is dropped into /run/host/userdb/. These
4527 records are picked up automatically by the userdb drop-in logic
4528 describe above, and allow the user to login with the same password as
4529 on the host. Effectively this means: if host and container run new
4530 enough systemd versions making a host user available to the container
4531 is trivially simple.
4533 * systemd-journal-gatewayd now supports the switches --user, --system,
4534 --merge, --file= that are equivalent to the same switches of
4535 journalctl, and permit exposing only the specified subset of the
4538 * The OnFailure= dependency between units is now augmented with a
4539 implicit reverse dependency OnFailureOf= (this new dependency cannot
4540 be configured directly it's only created as effect of an OnFailure=
4541 dependency in the reverse order — it's visible in "systemctl show"
4542 however). Similar, Slice= now has an reverse dependency SliceOf=,
4543 that is also not configurable directly, but useful to determine all
4544 units that are members of a slice.
4546 * A pair of new dependency types between units PropagatesStopTo= +
4547 StopPropagatedFrom= has been added, that allows propagation of unit
4548 stop events between two units. It operates similar to the existing
4549 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
4551 * A new dependency type OnSuccess= has been added (plus the reverse
4552 dependency OnSuccessOf=, which cannot be configured directly, but
4553 exists only as effect of the reverse OnSuccess=). It is similar to
4554 OnFailure=, but triggers in the opposite case: when a service exits
4555 cleanly. This allows "chaining up" of services where one or more
4556 services are started once another service has successfully completed.
4558 * A new dependency type Upholds= has been added (plus the reverse
4559 dependency UpheldBy=, which cannot be configured directly, but exists
4560 only as effect of Upholds=). This dependency type is a stronger form
4561 of Wants=: if a unit has an UpHolds= dependency on some other unit
4562 and the former is active then the latter is started whenever it is
4563 found inactive (and no job is queued for it). This is an alternative
4564 to Restart= inside service units, but less configurable, and the
4565 request to uphold a unit is not encoded in the unit itself but in
4566 another unit that intends to uphold it.
4568 * The systemd-ask-password tool now also supports reading passwords
4569 from the credentials subsystem, via the new --credential= switch.
4571 * The systemd-ask-password tool learnt a new switch --emoji= which may
4572 be used to explicit control whether the lock and key emoji (🔐) is
4573 shown in the password prompt on suitable TTYs.
4575 * The --echo switch of systemd-ask-password now optionally takes a
4576 parameter that controls character echo. It may either show asterisks
4577 (default, as before), turn echo off entirely, or echo the typed
4578 characters literally.
4580 * The systemd-ask-password tool also gained a new -n switch for
4581 suppressing output of a trailing newline character when writing the
4582 acquired password to standard output, similar to /bin/echo's -n
4585 * New documentation has been added that describes the organization of
4586 the systemd source code tree:
4588 https://systemd.io/ARCHITECTURE
4590 * Units using ConditionNeedsUpdate= will no longer be activated in
4593 * It is now possible to list a template unit in the WantedBy= or
4594 RequiredBy= settings of the [Install] section of another template
4595 unit, which will be instantiated using the same instance name.
4597 * A new MemoryAvailable property is available for units. If the unit,
4598 or the slices it is part of, have a memory limit set via MemoryMax=/
4599 MemoryHigh=, MemoryAvailable will indicate how much more memory the
4600 unit can claim before hitting the limits.
4602 * systemd-coredump will now try to stay below the cgroup memory limit
4603 placed on itself or one of the slices it runs under, if the storage
4604 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
4605 since files written on such filesystems count toward the cgroup memory
4606 limit. If there is not enough available memory in such cases to store
4607 the core file uncompressed, systemd-coredump will skip to compressed
4608 storage directly (if enabled) and it will avoid analyzing the core file
4609 to print backtrace and metadata in the journal.
4611 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
4612 of a path matches the configured expectations, and remove it if not.
4614 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
4615 specify which of the several available filesystem timestamps (access
4616 time, birth time, change time, modification time) to look at when
4617 deciding whether a path has aged enough to be cleaned.
4619 * A new IPv6StableSecretAddress= setting has been added to .network
4620 files, which takes an IPv6 address to use as secret for IPv6 address
4623 * The [DHCPServer] logic in .network files gained support for a new
4624 UplinkInterface= setting that permits configuration of the uplink
4625 interface name to propagate DHCP lease information from.
4627 * The WakeOnLan= setting in .link files now accepts a list of flags
4628 instead of a single one, to configure multiple wake-on-LAN policies.
4630 * User-space defined tracepoints (USDT) have been added to udev at
4631 strategic locations. This is useful for tracing udev behaviour and
4632 performance with bpftrace and similar tools.
4634 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
4635 setting a network timeout time.
4637 * If a system service is running in a new mount namespace (RootDirectory=
4638 and friends), all file systems will be mounted with MS_NOSUID by
4639 default, unless the system is running with SELinux enabled.
4641 * When enumerating time zones the timedatectl tool will now consult the
4642 'tzdata.zi' file shipped by the IANA time zone database package, in
4643 addition to 'zone1970.tab', as before. This makes sure time zone
4644 aliases are now correctly supported. Some distributions so far did
4645 not install this additional file, most do however. If you
4646 distribution does not install it yet, it might make sense to change
4649 * Intel HID rfkill event is no longer masked, since it's the only
4650 source of rfkill event on newer HP laptops. To have both backward and
4651 forward compatibility, userspace daemon needs to debounce duplicated
4652 events in a short time window.
4654 Contributions from: Aakash Singh, adrian5, Albert Brox,
4655 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
4656 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
4657 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
4658 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
4659 Christian Hesse, Daniel Schaefer, Dan Streetman,
4660 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
4661 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
4662 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
4663 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
4664 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
4665 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
4666 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
4667 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
4668 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
4669 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
4670 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
4671 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
4672 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
4673 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
4674 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
4675 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
4676 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
4677 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
4678 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
4679 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
4680 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
4681 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
4682 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
4683 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
4684 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
4686 — Edinburgh, 2021-07-07
4690 * A concept of system extension images is introduced. Such images may
4691 be used to extend the /usr/ and /opt/ directory hierarchies at
4692 runtime with additional files (even if the file system is read-only).
4693 When a system extension image is activated, its /usr/ and /opt/
4694 hierarchies and os-release information are combined via overlayfs
4695 with the file system hierarchy of the host OS.
4697 A new systemd-sysext tool can be used to merge, unmerge, list, and
4698 refresh system extension hierarchies. See
4699 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
4701 The systemd-sysext.service automatically merges installed system
4702 extensions during boot (before basic.target, but not in very early
4703 boot, since various file systems have to be mounted first).
4705 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
4706 supported system extension level.
4708 * A new ExtensionImages= unit setting can be used to apply the same
4709 system extension image concept from systemd-sysext to the namespaced
4710 file hierarchy of specific services, following the same rules and
4713 * Support for a new special "root=tmpfs" kernel command-line option has
4714 been added. When specified, a tmpfs is mounted on /, and mount.usr=
4715 should be used to point to the operating system implementation.
4717 * A new configuration file /etc/veritytab may be used to configure
4718 dm-verity integrity protection for block devices. Each line is in the
4719 format "volume-name data-device hash-device roothash options",
4720 similar to /etc/crypttab.
4722 * A new kernel command-line option systemd.verity.root_options= may be
4723 used to configure dm-verity behaviour for the root device.
4725 * The key file specified in /etc/crypttab (the third field) may now
4726 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
4727 acquired by connecting to that socket and reading from it. This
4728 allows the implementation of a service to provide key information
4729 dynamically, at the moment when it is needed.
4731 * When the hostname is set explicitly to "localhost", systemd-hostnamed
4732 will respect this. Previously such a setting would be mostly silently
4733 ignored. The goal is to honour configuration as specified by the
4736 * The fallback hostname that will be used by the system manager and
4737 systemd-hostnamed can now be configured in two new ways: by setting
4738 DEFAULT_HOSTNAME= in os-release(5), or by setting
4739 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
4740 also be configured during compilation. The environment variable is
4741 intended for testing and local overrides, the os-release(5) field is
4742 intended to allow customization by different variants of a
4743 distribution that share the same compiled packages.
4745 * The environment block of the manager itself may be configured through
4746 a new ManagerEnvironment= setting in system.conf or user.conf. This
4747 complements existing ways to set the environment block (the kernel
4748 command line for the system manager, the inherited environment and
4749 user@.service unit file settings for the user manager).
4751 * systemd-hostnamed now exports the default hostname and the source of
4752 the configured hostname ("static", "transient", or "default") as
4755 * systemd-hostnamed now exports the "HardwareVendor" and
4756 "HardwareModel" D-Bus properties, which are supposed to contain a
4757 pair of cleaned up, human readable strings describing the system's
4758 vendor and model. It's typically sourced from the firmware's DMI
4759 tables, but may be augmented from a new hwdb database. hostnamectl
4760 shows this in the status output.
4762 * Support has been added to systemd-cryptsetup for extracting the
4763 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
4764 metadata header. This allows the information how to open the
4765 encrypted device to be embedded directly in the device and obviates
4766 the need for configuration in an external file.
4768 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
4769 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
4770 pre-existing support for PKCS#11 security tokens).
4772 * systemd-repart may enroll encrypted partitions using TPM2
4773 hardware. This may be useful for example to create an encrypted /var
4774 partition bound to the machine on first boot.
4776 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
4777 and PKCS#11 security tokens to LUKS volumes, list and destroy
4780 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
4782 It also supports enrolling "recovery keys" and regular passphrases.
4784 * The libfido2 dependency is now based on dlopen(), so that the library
4785 is used at runtime when installed, but is not a hard runtime
4788 * systemd-cryptsetup gained support for two new options in
4789 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
4790 request synchronous processing of encryption/decryption IO.
4792 * The manager may be configured at compile time to use the fexecve()
4793 instead of the execve() system call when spawning processes. Using
4794 fexecve() closes a window between checking the security context of an
4795 executable and spawning it, but unfortunately the kernel displays
4796 stale information in the process' "comm" field, which impacts ps
4799 * The configuration option -Dcompat-gateway-hostname has been dropped.
4800 "_gateway" is now the only supported name.
4802 * The ConditionSecurity=tpm2 unit file setting may be used to check if
4803 the system has at least one TPM2 (tpmrm class) device.
4805 * A new ConditionCPUFeature= has been added that may be used to
4806 conditionalize units based on CPU features. For example,
4807 ConditionCPUFeature=rdrand will condition a unit so that it is only
4808 run when the system CPU supports the RDRAND opcode.
4810 * The existing ConditionControlGroupController= setting has been
4811 extended with two new values "v1" and "v2". "v2" means that the
4812 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
4813 hierarchy or the hybrid hierarchy are used.
4815 * A new PrivateIPC= setting on a unit file allows executed processes to
4816 be moved into a private IPC namespace, with separate System V IPC
4817 identifiers and POSIX message queues.
4819 A new IPCNamespacePath= allows the unit to be joined to an existing
4822 * The tables of system calls in seccomp filters are now automatically
4823 generated from kernel lists exported on
4824 https://fedora.juszkiewicz.com.pl/syscalls.html.
4826 The following architectures should now have complete lists:
4827 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
4828 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
4830 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
4831 on /run/ if it is not already a mount point. A writable /run/ has
4832 always been a requirement for a functioning system, but this was not
4833 guaranteed when using a read-only image.
4835 Users can always specify BindPaths= or InaccessiblePaths= as
4836 overrides, and they will take precedence. If the host's root mount
4837 point is used, there is no change in behaviour.
4839 * New bind mounts and file system image mounts may be injected into the
4840 mount namespace of a service (without restarting it). This is exposed
4841 respectively as 'systemctl bind <unit> <path>…' and
4842 'systemctl mount-image <unit> <image>…'.
4844 * The StandardOutput= and StandardError= settings can now specify files
4845 to be truncated for output (as "truncate:<path>").
4847 * The ExecPaths= and NoExecPaths= settings may be used to specify
4848 noexec for parts of the file system.
4850 * sd-bus has a new function sd_bus_open_user_machine() to open a
4851 connection to the session bus of a specific user in a local container
4852 or on the local host. This is exposed in the existing -M switch to
4853 systemctl and similar tools:
4855 systemctl --user -M lennart@foobar start foo
4857 This will connect to the user bus of a user "lennart" in container
4858 "foobar". If no container name is specified, the specified user on
4859 the host itself is connected to
4861 systemctl --user -M lennart@ start quux
4863 * sd-bus also gained a convenience function sd_bus_message_send() to
4864 simplify invocations of sd_bus_send(), taking only a single
4865 parameter: the message to send.
4867 * sd-event allows rate limits to be set on event sources, for dealing
4868 with high-priority event sources that might starve out others. See
4869 the new man page sd_event_source_set_ratelimit(3) for details.
4871 * systemd.link files gained a [Link] Promiscuous= switch, which allows
4872 the device to be raised in promiscuous mode.
4874 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
4875 number of TX and RX queues to be configured.
4877 New [Link] TransmitQueueLength= setting allows the size of the TX
4878 queue to be configured.
4880 New [Link] GenericSegmentOffloadMaxBytes= and
4881 GenericSegmentOffloadMaxSegments= allow capping the packet size and
4882 the number of segments accepted in Generic Segment Offload.
4884 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
4885 wireless routing protocol that operates on ISO/OSI Layer 2 only and
4886 uses ethernet frames to route/bridge packets. This encompasses a new
4887 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
4888 new settings in .netdev files, and a new BatmanAdvanced= setting in
4891 * systemd.network files gained a [Network] RouteTable= configuration
4892 switch to select the routing policy table.
4894 systemd.network files gained a [RoutingPolicyRule] Type=
4895 configuration switch (one of "blackhole, "unreachable", "prohibit").
4897 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
4898 RouteAllowList= settings to ignore/accept route advertisements from
4899 routers matching specified prefixes. The DenyList= setting has been
4900 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
4903 systemd.network files gained a [DHCPv6] UseAddress= setting to
4904 optionally ignore the address provided in the lease.
4906 systemd.network files gained a [DHCPv6PrefixDelegation]
4907 ManageTemporaryAddress= switch.
4909 systemd.network files gained a new ActivationPolicy= setting which
4910 allows configuring how the UP state of an interface shall be managed,
4911 i.e. whether the interface is always upped, always downed, or may be
4912 upped/downed by the user using "ip link set dev".
4914 * The default for the Broadcast= setting in .network files has slightly
4915 changed: the broadcast address will not be configured for wireguard
4918 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
4919 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
4920 configuration options for VLAN packet handling.
4922 * udev rules may now set log_level= option. This allows debug logs to
4923 be enabled for select events, e.g. just for a specific subsystem or
4924 even a single device.
4926 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
4927 DATA_PREPARED_ID properties for block devices with ISO9660 file
4930 * udev now exports decoded DMI information about installed memory slots
4931 as device properties under the /sys/class/dmi/id/ pseudo device.
4933 * /dev/ is not mounted noexec anymore. This didn't provide any
4934 significant security benefits and would conflict with the executable
4935 mappings used with /dev/sgx device nodes. The previous behaviour can
4936 be restored for individual services with NoExecPaths=/dev (or by allow-
4937 listing and excluding /dev from ExecPaths=).
4939 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
4940 and /dev/vhost-net are owned by the kvm group.
4942 * The hardware database has been extended with a list of fingerprint
4943 readers that correctly support USB auto-suspend using data from
4946 * systemd-resolved can now answer DNSSEC questions through the stub
4947 resolver interface in a way that allows local clients to do DNSSEC
4948 validation themselves. For a question with DO+CD set, it'll proxy the
4949 DNS query and respond with a mostly unmodified packet received from
4950 the upstream server.
4952 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
4953 resolved.conf. If true the service will provide caching even for DNS
4954 lookups made to an upstream DNS server on the 127.0.0.1/::1
4955 addresses. By default (and when the option is false) systemd-resolved
4956 will not cache such lookups, in order to avoid duplicate local
4957 caching, under the assumption the local upstream server caches
4960 * systemd-resolved now implements RFC5001 NSID in its local DNS
4961 stub. This may be used by local clients to determine whether they are
4962 talking to the DNS resolver stub or a different DNS server.
4964 * When resolving host names and other records resolvectl will now
4965 report where the data was acquired from (i.e. the local cache, the
4966 network, locally synthesized, …) and whether the network traffic it
4967 effected was encrypted or not. Moreover the tool acquired a number of
4968 new options --cache=, --synthesize=, --network=, --zone=,
4969 --trust-anchor=, --validate= that take booleans and may be used to
4970 tweak a lookup, i.e. whether it may be answered from cached
4971 information, locally synthesized information, information acquired
4972 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
4973 anchor, and whether DNSSEC validation shall be executed for the
4976 * systemd-nspawn gained a new --ambient-capability= setting
4977 (AmbientCapability= in .nspawn files) to configure ambient
4978 capabilities passed to the container payload.
4980 * systemd-nspawn gained the ability to configure the firewall using the
4981 nftables subsystem (in addition to the existing iptables
4982 support). Similarly, systemd-networkd's IPMasquerade= option now
4983 supports nftables as back-end, too. In both cases NAT on IPv6 is now
4984 supported too, in addition to IPv4 (the iptables back-end still is
4987 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
4988 retains its meaning, but has been deprecated. Please switch to either
4989 "ivp4" or "both" (if covering IPv6 is desired).
4991 * systemd-importd will now download .verity and .roothash.p7s files
4992 along with the machine image (as exposed via machinectl pull-raw).
4994 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
4995 setting to configure the time a unit's cgroup needs to exceed memory
4996 pressure limits before action will be taken, and a new
4997 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
5000 systemd-oomd is now considered fully supported (the usual
5001 backwards-compatibility promises apply). Swap is not required for
5002 operation, but it is still recommended.
5004 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
5005 configures the retry delay when trying to contact servers.
5007 * systemd-stdio-bridge gained --system/--user options to connect to the
5008 system bus (previous default) or the user session bus.
5010 * systemd-localed may now call locale-gen to generate missing locales
5011 on-demand (UTF-8-only). This improves integration with Debian-based
5012 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
5014 * systemctl --check-inhibitors=true may now be used to obey inhibitors
5015 even when invoked non-interactively. The old --ignore-inhibitors
5016 switch is now deprecated and replaced by --check-inhibitors=false.
5018 * systemctl import-environment will now emit a warning when called
5019 without any arguments (i.e. to import the full environment block of
5020 the called program). This command will usually be invoked from a
5021 shell, which means that it'll inherit a bunch of variables which are
5022 specific to that shell, and usually to the TTY the shell is connected
5023 to, and don't have any meaning in the global context of the system or
5024 user service manager. Instead, only specific variables should be
5025 imported into the manager environment block.
5027 Similarly, programs which update the manager environment block by
5028 directly calling the D-Bus API of the manager, should also push
5029 specific variables, and not the full inherited environment.
5031 * systemctl's status output now shows unit state with a more careful
5032 choice of Unicode characters: units in maintenance show a "○" symbol
5033 instead of the usual "●", failed units show "×", and services being
5036 * coredumpctl gained a --debugger-arguments= switch to pass arguments
5037 to the debugger. It also gained support for showing coredump info in
5038 a simple JSON format.
5040 * systemctl/loginctl/machinectl's --signal= option now accept a special
5041 value "list", which may be used to show a brief table with known
5042 process signals and their numbers.
5044 * networkctl now shows the link activation policy in status.
5046 * Various tools gained --pager/--no-pager/--json= switches to
5047 enable/disable the pager and provide JSON output.
5049 * Various tools now accept two new values for the SYSTEMD_COLORS
5050 environment variable: "16" and "256", to configure how many terminal
5051 colors are used in output.
5053 * less 568 or newer is now required for the auto-paging logic of the
5054 various tools. Hyperlink ANSI sequences in terminal output are now
5055 used even if a pager is used, and older versions of less are not able
5056 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
5057 disable this output again.
5059 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
5060 builds, non-merged-usr builds) are now officially deprecated. A
5061 warning is emitted during build. Support is slated to be removed in
5062 about a year (when the Debian Bookworm release development starts).
5064 * Systems with the legacy cgroup v1 hierarchy are now marked as
5065 "tainted", to make it clearer that using the legacy hierarchy is not
5068 * systemd-localed will now refuse to configure a keymap which is not
5069 installed in the file system. This is intended as a bug fix, but
5070 could break cases where systemd-localed was used to configure the
5071 keymap in advanced of it being installed. It is necessary to install
5072 the keymap file first.
5074 * The main git development branch has been renamed to 'main'.
5076 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
5077 for partitions, as in the vast majority of cases they contain none
5078 and are used internally by the bootloader (eg: uboot).
5080 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
5081 spawned processes to the PID of the process itself. This may be used
5082 by programs for detecting whether they were forked off by the service
5083 manager itself or are a process forked off further down the tree.
5085 * The sd-device API gained four new calls: sd_device_get_action() to
5086 determine the uevent add/remove/change/… action the device object has
5087 been seen for, sd_device_get_seqno() to determine the uevent sequence
5088 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
5089 object from stat(2) data of a device node, and sd_device_trigger() to
5090 write to the 'uevent' attribute of a device.
5092 * For most tools the --no-legend= switch has been replaced by
5093 --legend=no and --legend=yes, to force whether tables are shown with
5096 * Units acquired a new property "Markers" that takes a list of zero,
5097 one or two of the following strings: "needs-reload" and
5098 "needs-restart". These markers may be set via "systemctl
5099 set-property". Once a marker is set, "systemctl reload-or-restart
5100 --marked" may be invoked to execute the operation the units are
5101 marked for. This is useful for package managers that want to mark
5102 units for restart/reload while updating, but effect the actual
5103 operations at a later step at once.
5105 * The sd_bus_message_read_strv() API call of sd-bus may now also be
5106 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
5109 * bootctl will now report whether the UEFI firmware used a TPM2 device
5110 and measured the boot process into it.
5112 * systemd-tmpfiles learnt support for a new environment variable
5113 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
5114 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
5115 even if the root fs of the system is not itself a btrfs volume.
5117 * systemd-detect-virt/ConditionVirtualization= will now explicitly
5118 detect Docker/Podman environments where possible. Moreover, they
5119 should be able to generically detect any container manager as long as
5120 it assigns the container a cgroup.
5122 * portablectl gained a new "reattach" verb for detaching/reattaching a
5123 portable service image, useful for updating images on-the-fly.
5125 * Intel SGX enclave device nodes (which expose a security feature of
5126 newer Intel CPUs) will now be owned by a new system group "sgx".
5128 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
5129 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
5130 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
5131 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
5132 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
5133 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
5134 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
5135 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
5136 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
5137 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
5138 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
5139 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
5140 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
5141 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
5142 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
5143 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
5144 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
5145 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
5146 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
5147 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
5148 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
5149 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
5150 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
5151 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
5152 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
5153 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
5154 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
5155 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
5156 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
5157 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
5158 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
5159 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
5160 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
5161 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
5162 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
5163 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
5164 Zmicer Turok, Дамјан Георгиевски
5166 — Berlin, 2021-03-30
5170 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
5171 "bind" and "unbind" to the Linux device model. When this kernel
5172 change was made, systemd-udevd was only minimally updated to handle
5173 and propagate these new event types. The introduction of these new
5174 uevents (which are typically generated for USB devices and devices
5175 needing a firmware upload before being functional) resulted in a
5176 number of issues which we so far didn't address. We hoped the kernel
5177 maintainers would themselves address these issues in some form, but
5178 that did not happen. To handle them properly, many (if not most) udev
5179 rules files shipped in various packages need updating, and so do many
5180 programs that monitor or enumerate devices with libudev or sd-device,
5181 or otherwise process uevents. Please note that this incompatibility
5182 is not fault of systemd or udev, but caused by an incompatible kernel
5183 change that happened back in Linux 4.14, but is becoming more and
5184 more visible as the new uevents are generated by more kernel drivers.
5186 To minimize issues resulting from this kernel change (but not avoid
5187 them entirely) starting with systemd-udevd 247 the udev "tags"
5188 concept (which is a concept for marking and filtering devices during
5189 enumeration and monitoring) has been reworked: udev tags are now
5190 "sticky", meaning that once a tag is assigned to a device it will not
5191 be removed from the device again until the device itself is removed
5192 (i.e. unplugged). This makes sure that any application monitoring
5193 devices that match a specific tag is guaranteed to both see uevents
5194 where the device starts being relevant, and those where it stops
5195 being relevant (the latter now regularly happening due to the new
5196 "unbind" uevent type). The udev tags concept is hence now a concept
5197 tied to a *device* instead of a device *event* — unlike for example
5198 udev properties whose lifecycle (as before) is generally tied to a
5199 device event, meaning that the previously determined properties are
5200 forgotten whenever a new uevent is processed.
5202 With the newly redefined udev tags concept, sometimes it's necessary
5203 to determine which tags are the ones applied by the most recent
5204 uevent/database update, in order to discern them from those
5205 originating from earlier uevents/database updates of the same
5206 device. To accommodate for this a new automatic property CURRENT_TAGS
5207 has been added that works similar to the existing TAGS property but
5208 only lists tags set by the most recent uevent/database
5209 update. Similarly, the libudev/sd-device API has been updated with
5210 new functions to enumerate these 'current' tags, in addition to the
5211 existing APIs that now enumerate the 'sticky' ones.
5213 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
5214 essential that all udev rules files and applications are updated to
5215 handle the new events. Specifically:
5217 • All rule files that currently use a header guard similar to
5218 ACTION!="add|change",GOTO="xyz_end" should be updated to use
5219 ACTION=="remove",GOTO="xyz_end" instead, so that the
5220 properties/tags they add are also applied whenever "bind" (or
5221 "unbind") is seen. (This is most important for all physical device
5222 types — those for which "bind" and "unbind" are currently
5223 generated, for all other device types this change is still
5224 recommended but not as important — but certainly prepares for
5225 future kernel uevent type additions).
5227 • Similarly, all code monitoring devices that contains an 'if' branch
5228 discerning the "add" + "change" uevent actions from all other
5229 uevents actions (i.e. considering devices only relevant after "add"
5230 or "change", and irrelevant on all other events) should be reworked
5231 to instead negatively check for "remove" only (i.e. considering
5232 devices relevant after all event types, except for "remove", which
5233 invalidates the device). Note that this also means that devices
5234 should be considered relevant on "unbind", even though conceptually
5235 this — in some form — invalidates the device. Since the precise
5236 effect of "unbind" is not generically defined, devices should be
5237 considered relevant even after "unbind", however I/O errors
5238 accessing the device should then be handled gracefully.
5240 • Any code that uses device tags for deciding whether a device is
5241 relevant or not most likely needs to be updated to use the new
5242 udev_device_has_current_tag() API (or sd_device_has_current_tag()
5243 in case sd-device is used), to check whether the tag is set at the
5244 moment an uevent is seen (as opposed to the existing
5245 udev_device_has_tag() API which checks if the tag ever existed on
5246 the device, following the API concept redefinition explained
5249 We are very sorry for this breakage and the requirement to update
5250 packages using these interfaces. We'd again like to underline that
5251 this is not caused by systemd/udev changes, but result of a kernel
5254 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
5255 packages have not retriggered devices once the udev package (or any
5256 auxiliary package installing additional udev rules) is updated. We
5257 intend to work with major distributions to change this, so that
5258 "udevadm trigger -c change" is issued on such upgrades, ensuring that
5259 the updated ruleset is applied to the devices already discovered, so
5260 that (asynchronously) after the upgrade completed the udev database
5261 is consistent with the updated rule set. This means udev rules must
5262 be ready to be retriggered with a "change" action any time, and
5263 result in correct and complete udev database entries. While the
5264 majority of udev rule files known to us currently get this right,
5265 some don't. Specifically, there are udev rules files included in
5266 various packages that only set udev properties on the "add" action,
5267 but do not handle the "change" action. If a device matching those
5268 rules is retriggered with the "change" action (as is intended here)
5269 it would suddenly lose the relevant properties. This always has been
5270 problematic, but as soon as all udev devices are triggered on relevant
5271 package upgrades this will become particularly so. It is strongly
5272 recommended to fix offending rules so that they can handle a "change"
5273 action at any time, and acquire all necessary udev properties even
5274 then. Or in other words: the header guard mentioned above
5275 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
5276 this, as it makes sure rules are rerun on "change" correctly, and
5277 accumulate the correct and complete set of udev properties. udev rule
5278 definitions that cannot handle "change" events being triggered at
5279 arbitrary times should be considered buggy.
5281 * The MountAPIVFS= service file setting now defaults to on if
5282 RootImage= and RootDirectory= are used, which means that with those
5283 two settings /proc/, /sys/ and /dev/ are automatically properly set
5284 up for services. Previous behaviour may be restored by explicitly
5285 setting MountAPIVFS=off.
5287 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
5288 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
5289 latter it takes precedence over the former, similar to how most of
5290 systemd's own configuration is handled. Given that PAM stack
5291 definitions are primarily put together by OS vendors/distributions
5292 (though possibly overridden by users), this systemd release moves its
5293 own PAM stack configuration for the "systemd-user" PAM service (i.e.
5294 for the PAM session invoked by the per-user user@.service instance)
5295 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
5296 packages' vendor versions of their PAM stack definitions from
5297 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
5298 desired the location to which systemd installs its PAM stack
5299 configuration may be changed via the -Dpamconfdir Meson option.
5301 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
5302 libpwquality and libcryptsetup have been changed to be based on
5303 dlopen(): instead of regular dynamic library dependencies declared in
5304 the binary ELF headers, these libraries are now loaded on demand
5305 only, if they are available. If the libraries cannot be found the
5306 relevant operations will fail gracefully, or a suitable fallback
5307 logic is chosen. This is supposed to be useful for general purpose
5308 distributions, as it allows minimizing the list of dependencies the
5309 systemd packages pull in, permitting building of more minimal OS
5310 images, while still making use of these "weak" dependencies should
5311 they be installed. Since many package managers automatically
5312 synthesize package dependencies from ELF shared library dependencies,
5313 some additional manual packaging work has to be done now to replace
5314 those (slightly downgraded from "required" to "recommended" or
5315 whatever is conceptually suitable for the package manager). Note that
5316 this change does not alter build-time behaviour: as before the
5317 build-time dependencies have to be installed during build, even if
5318 they now are optional during runtime.
5320 * sd-event.h gained a new call sd_event_add_time_relative() for
5321 installing timers relative to the current time. This is mostly a
5322 convenience wrapper around the pre-existing sd_event_add_time() call
5323 which installs absolute timers.
5325 * sd-event event sources may now be placed in a new "exit-on-failure"
5326 mode, which may be controlled via the new
5327 sd_event_source_get_exit_on_failure() and
5328 sd_event_source_set_exit_on_failure() functions. If enabled, any
5329 failure returned by the event source handler functions will result in
5330 exiting the event loop (unlike the default behaviour of just
5331 disabling the event source but continuing with the event loop). This
5332 feature is useful to set for all event sources that define "primary"
5333 program behaviour (where failure should be fatal) in contrast to
5334 "auxiliary" behaviour (where failure should remain local).
5336 * Most event source types sd-event supports now accept a NULL handler
5337 function, in which case the event loop is exited once the event
5338 source is to be dispatched, using the userdata pointer — converted to
5339 a signed integer — as exit code of the event loop. Previously this
5340 was supported for IO and signal event sources already. Exit event
5341 sources still do not support this (simply because it makes little
5342 sense there, as the event loop is already exiting when they are
5345 * A new per-unit setting RootImageOptions= has been added which allows
5346 tweaking the mount options for any file system mounted as effect of
5347 the RootImage= setting.
5349 * Another new per-unit setting MountImages= has been added, that allows
5350 mounting additional disk images into the file system tree accessible
5353 * Timer units gained a new FixedRandomDelay= boolean setting. If
5354 enabled, the random delay configured with RandomizedDelaySec= is
5355 selected in a way that is stable on a given system (though still
5356 different for different units).
5358 * Socket units gained a new setting Timestamping= that takes "us", "ns"
5359 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
5362 * systemd-repart now generates JSON output when requested with the new
5365 * systemd-machined's OpenMachineShell() bus call will now pass
5366 additional policy metadata data fields to the PolicyKit
5367 authentication request.
5369 * systemd-tmpfiles gained a new -E switch, which is equivalent to
5370 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
5371 --exclude=/sys. It's particularly useful in combination with --root=,
5372 when operating on OS trees that do not have any of these four runtime
5373 directories mounted, as this means no files below these subtrees are
5374 created or modified, since those mount points should probably remain
5377 * systemd-tmpfiles gained a new --image= switch which is like --root=,
5378 but takes a disk image instead of a directory as argument. The
5379 specified disk image is mounted inside a temporary mount namespace
5380 and the tmpfiles.d/ drop-ins stored in the image are executed and
5381 applied to the image. systemd-sysusers similarly gained a new
5382 --image= switch, that allows the sysusers.d/ drop-ins stored in the
5383 image to be applied onto the image.
5385 * Similarly, the journalctl command also gained an --image= switch,
5386 which is a quick one-step solution to look at the log data included
5389 * journalctl's --output=cat option (which outputs the log content
5390 without any metadata, just the pure text messages) will now make use
5391 of terminal colors when run on a suitable terminal, similarly to the
5394 * JSON group records now support a "description" string that may be
5395 used to add a human-readable textual description to such groups. This
5396 is supposed to match the user's GECOS field which traditionally
5397 didn't have a counterpart for group records.
5399 * The "systemd-dissect" tool that may be used to inspect OS disk images
5400 and that was previously installed to /usr/lib/systemd/ has now been
5401 moved to /usr/bin/, reflecting its updated status of an officially
5402 supported tool with a stable interface. It gained support for a new
5403 --mkdir switch which when combined with --mount has the effect of
5404 creating the directory to mount the image to if it is missing
5405 first. It also gained two new commands --copy-from and --copy-to for
5406 copying files and directories in and out of an OS image without the
5407 need to manually mount it. It also acquired support for a new option
5408 --json= to generate JSON output when inspecting an OS image.
5410 * The cgroup2 file system is now mounted with the
5411 "memory_recursiveprot" mount option, supported since kernel 5.7. This
5412 means that the MemoryLow= and MemoryMin= unit file settings now apply
5413 recursively to whole subtrees.
5415 * systemd-homed now defaults to using the btrfs file system — if
5416 available — when creating home directories in LUKS volumes. This may
5417 be changed with the DefaultFileSystemType= setting in homed.conf.
5418 It's now the default file system in various major distributions and
5419 has the major benefit for homed that it can be grown and shrunk while
5420 mounted, unlike the other contenders ext4 and xfs, which can both be
5421 grown online, but not shrunk (in fact xfs is the technically most
5422 limited option here, as it cannot be shrunk at all).
5424 * JSON user records managed by systemd-homed gained support for
5425 "recovery keys". These are basically secondary passphrases that can
5426 unlock user accounts/home directories. They are computer-generated
5427 rather than user-chosen, and typically have greater entropy.
5428 homectl's --recovery-key= option may be used to add a recovery key to
5429 a user account. The generated recovery key is displayed as a QR code,
5430 so that it can be scanned to be kept in a safe place. This feature is
5431 particularly useful in combination with systemd-homed's support for
5432 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
5433 security tokens are lost. Recovery keys may be entered wherever the
5434 system asks for a password.
5436 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
5437 home directory which indicates that a home directory has not been
5438 deactivated cleanly when offline. This flag is useful to identify
5439 home directories for which the offline discard logic did not run when
5440 offlining, and where it would be a good idea to log in again to catch
5443 * systemctl gained a new parameter --timestamp= which may be used to
5444 change the style in which timestamps are output, i.e. whether to show
5445 them in local timezone or UTC, or whether to show µs granularity.
5447 * Alibaba's "pouch" container manager is now detected by
5448 systemd-detect-virt, ConditionVirtualization= and similar
5449 constructs. Similar, they now also recognize IBM PowerVM machine
5452 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
5453 place to use for propagating external mounts into the
5454 container. Similarly /run/host/notify is now used as the socket path
5455 for container payloads to communicate with the container manager
5456 using sd_notify(). The container manager now uses the
5457 /run/host/inaccessible/ directory to place "inaccessible" file nodes
5458 of all relevant types which may be used by the container payload as
5459 bind mount source to over-mount inodes to make them inaccessible.
5460 /run/host/container-manager will now be initialized with the same
5461 string as the $container environment variable passed to the
5462 container's PID 1. /run/host/container-uuid will be initialized with
5463 the same string as $container_uuid. This means the /run/host/
5464 hierarchy is now the primary way to make host resources available to
5465 the container. The Container Interface documents these new files and
5468 https://systemd.io/CONTAINER_INTERFACE
5470 * Support for the "ConditionNull=" unit file condition has been
5471 deprecated and undocumented for 6 years. systemd started to warn
5472 about its use 1.5 years ago. It has now been removed entirely.
5474 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
5475 a sd_bus_error struct and a list of error names, and checks if the
5476 error matches one of these names. It's a convenience wrapper that is
5477 useful in cases where multiple errors shall be handled the same way.
5479 * A new system call filter list "@known" has been added, that contains
5480 all system calls known at the time systemd was built.
5482 * Behaviour of system call filter allow lists has changed slightly:
5483 system calls that are contained in @known will result in EPERM by
5484 default, while those not contained in it result in ENOSYS. This
5485 should improve compatibility because known system calls will thus be
5486 communicated as prohibited, while unknown (and thus newer ones) will
5487 be communicated as not implemented, which hopefully has the greatest
5488 chance of triggering the right fallback code paths in client
5491 * "systemd-analyze syscall-filter" will now show two separate sections
5492 at the bottom of the output: system calls known during systemd build
5493 time but not included in any of the filter groups shown above, and
5494 system calls defined on the local kernel but known during systemd
5497 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
5498 systemd-nspawn all system call filter violations will be logged by
5499 the kernel (audit). This is useful for tracking down system calls
5500 invoked by container payloads that are prohibited by the container's
5501 system call filter policy.
5503 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
5504 systemd-nspawn (and other programs that use seccomp) all seccomp
5505 filtering is turned off.
5507 * Two new unit file settings ProtectProc= and ProcSubset= have been
5508 added that expose the hidepid= and subset= mount options of procfs.
5509 All processes of the unit will only see processes in /proc that are
5510 are owned by the unit's user. This is an important new sandboxing
5511 option that is recommended to be set on all system services. All
5512 long-running system services that are included in systemd itself set
5513 this option now. This option is only supported on kernel 5.8 and
5514 above, since the hidepid= option supported on older kernels was not a
5515 per-mount option but actually applied to the whole PID namespace.
5517 * Socket units gained a new boolean setting FlushPending=. If enabled
5518 all pending socket data/connections are flushed whenever the socket
5519 unit enters the "listening" state, i.e. after the associated service
5522 * The unit file setting NUMAMask= gained a new "all" value: when used,
5523 all existing NUMA nodes are added to the NUMA mask.
5525 * A new "credentials" logic has been added to system services. This is
5526 a simple mechanism to pass privileged data to services in a safe and
5527 secure way. It's supposed to be used to pass per-service secret data
5528 such as passwords or cryptographic keys but also associated less
5529 private information such as user names, certificates, and similar to
5530 system services. Each credential is identified by a short user-chosen
5531 name and may contain arbitrary binary data. Two new unit file
5532 settings have been added: SetCredential= and LoadCredential=. The
5533 former allows setting a credential to a literal string, the latter
5534 sets a credential to the contents of a file (or data read from a
5535 user-chosen AF_UNIX stream socket). Credentials are passed to the
5536 service via a special credentials directory, one file for each
5537 credential. The path to the credentials directory is passed in a new
5538 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
5539 are passed in the file system they may be easily referenced in
5540 ExecStart= command lines too, thus no explicit support for the
5541 credentials logic in daemons is required (though ideally daemons
5542 would look for the bits they need in $CREDENTIALS_DIRECTORY
5543 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
5544 backed by unswappable memory if privileges allow it, immutable if
5545 privileges allow it, is accessible only to the service's UID, and is
5546 automatically destroyed when the service stops.
5548 * systemd-nspawn supports the same credentials logic. It can both
5549 consume credentials passed to it via the aforementioned
5550 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
5551 to its payload. The service manager/PID 1 has been updated to match
5552 this: it can also accept credentials from the container manager that
5553 invokes it (in fact: any process that invokes it), and passes them on
5554 to its services. Thus, credentials can be propagated recursively down
5555 the tree: from a system's service manager to a systemd-nspawn
5556 service, to the service manager that runs as container payload and to
5557 the service it runs below. Credentials may also be added on the
5558 systemd-nspawn command line, using new --set-credential= and
5559 --load-credential= command line switches that match the
5560 aforementioned service settings.
5562 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
5563 the partition drop-ins which may be used to format/LUKS
5564 encrypt/populate any created partitions. The partitions are
5565 encrypted/formatted/populated before they are registered in the
5566 partition table, so that they appear atomically: either the
5567 partitions do not exist yet or they exist fully encrypted, formatted,
5568 and populated — there is no time window where they are
5569 "half-initialized". Thus the system is robust to abrupt shutdown: if
5570 the tool is terminated half-way during its operations on next boot it
5571 will start from the beginning.
5573 * systemd-repart's --size= operation gained a new "auto" value. If
5574 specified, and operating on a loopback file it is automatically sized
5575 to the minimal size the size constraints permit. This is useful to
5576 use "systemd-repart" as an image builder for minimally sized images.
5578 * systemd-resolved now gained a third IPC interface for requesting name
5579 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
5580 interface is now supported. The nss-resolve NSS module has been
5581 modified to use this new interface instead of D-Bus. Using Varlink
5582 has a major benefit over D-Bus: it works without a broker service,
5583 and thus already during earliest boot, before the dbus daemon has
5584 been started. This means name resolution via systemd-resolved now
5585 works at the same time systemd-networkd operates: from earliest boot
5586 on, including in the initrd.
5588 * systemd-resolved gained support for a new DNSStubListenerExtra=
5589 configuration file setting which may be used to specify additional IP
5590 addresses the built-in DNS stub shall listen on, in addition to the
5591 main one on 127.0.0.53:53.
5593 * Name lookups issued via systemd-resolved's D-Bus and Varlink
5594 interfaces (and thus also via glibc NSS if nss-resolve is used) will
5595 now honour a trailing dot in the hostname: if specified the search
5596 path logic is turned off. Thus "resolvectl query foo." is now
5597 equivalent to "resolvectl query --search=off foo.".
5599 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
5600 exposes how /etc/resolv.conf is currently managed: by resolved (and
5601 in which mode if so) or another subsystem. "resolvctl" will display
5602 this property in its status output.
5604 * The resolv.conf snippets systemd-resolved provides will now set "."
5605 as the search domain if no other search domain is known. This turns
5606 off the derivation of an implicit search domain by nss-dns for the
5607 hostname, when the hostname is set to an FQDN. This change is done to
5608 make nss-dns using resolv.conf provided by systemd-resolved behave
5609 more similarly to nss-resolve.
5611 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
5612 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
5613 "birth" time (btime) of a file in addition to the atime, mtime, and
5616 * systemd-analyze gained a new verb "capability" that lists all known
5617 capabilities by the systemd build and by the kernel.
5619 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
5620 advance the system clock to it at boot if it is noticed to be before
5621 that time. Previously, PID 1 would only advance the time to an epoch
5622 time that is set during build-time. With this new file OS builders
5623 can change this epoch timestamp on individual OS images without
5624 having to rebuild systemd.
5626 * systemd-logind will now listen to the KEY_RESTART key from the Linux
5627 input layer and reboot the system if it is pressed, similarly to how
5628 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
5629 was originally defined in the Multimedia context (to restart playback
5630 of a song or film), but is now primarily used in various embedded
5631 devices for "Reboot" buttons. Accordingly, systemd-logind will now
5632 honour it as such. This may configured in more detail via the new
5633 HandleRebootKey= and RebootKeyIgnoreInhibited=.
5635 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
5636 copying OS trees, for example in "systemd-nspawn --ephemeral",
5637 "systemd-nspawn --template=", "machinectl clone" and similar. This is
5638 useful when operating with OSTree images, which use hardlinks heavily
5639 throughout, and where such copies previously resulting in "exploding"
5642 * systemd-nspawn's --console= setting gained support for a new
5643 "autopipe" value, which is identical to "interactive" when invoked on
5644 a TTY, and "pipe" otherwise.
5646 * systemd-networkd's .network files gained support for explicitly
5647 configuring the multicast membership entries of bridge devices in the
5648 [BridgeMDB] section. It also gained support for the PIE queuing
5649 discipline in the [FlowQueuePIE] sections.
5651 * systemd-networkd's .netdev files may now be used to create "BareUDP"
5652 tunnels, configured in the new [BareUDP] setting.
5654 * systemd-networkd's Gateway= setting in .network files now accepts the
5655 special values "_dhcp4" and "_ipv6ra" to configure additional,
5656 locally defined, explicit routes to the gateway acquired via DHCP or
5657 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
5658 but still accepted for backwards compatibility.
5660 * systemd-networkd's [IPv6PrefixDelegation] section and
5661 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
5662 IPv6SendRA= (the old names are still accepted for backwards
5665 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
5666 boolean setting in [Network] section. If enabled, the delegated prefix
5667 gained by another link will be configured, and an address within the
5668 prefix will be assigned.
5670 * systemd-networkd's .network files gained the Announce= boolean setting
5671 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
5672 prefix will be announced through IPv6 router advertisement (IPv6 RA).
5673 The setting is enabled by default.
5675 * VXLAN tunnels may now be marked as independent of any underlying
5676 network interface via the new Independent= boolean setting.
5678 * systemctl gained support for two new verbs: "service-log-level" and
5679 "service-log-target" may be used on services that implement the
5680 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
5681 adjust the log level and target. All of systemd's long-running
5682 services support this now, but ideally all system services would
5683 implement this interface to make the system more uniformly
5686 * The SystemCallErrorNumber= unit file setting now accepts the new
5687 "kill" and "log" actions, in addition to arbitrary error number
5688 specifications as before. If "kill" the processes are killed on the
5689 event, if "log" the offending system call is audit logged.
5691 * A new SystemCallLog= unit file setting has been added that accepts a
5692 list of system calls that shall be logged about (audit).
5694 * The OS image dissection logic (as used by RootImage= in unit files or
5695 systemd-nspawn's --image= switch) has gained support for identifying
5696 and mounting explicit /usr/ partitions, which are now defined in the
5697 discoverable partition specification. This should be useful for
5698 environments where the root file system is
5699 generated/formatted/populated dynamically on first boot and combined
5700 with an immutable /usr/ tree that is supplied by the vendor.
5702 * In the final phase of shutdown, within the systemd-shutdown binary
5703 we'll now try to detach MD devices (i.e software RAID) in addition to
5704 loopback block devices and DM devices as before. This is supposed to
5705 be a safety net only, in order to increase robustness if things go
5706 wrong. Storage subsystems are expected to properly detach their
5707 storage volumes during regular shutdown already (or in case of
5708 storage backing the root file system: in the initrd hook we return to
5711 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
5712 will now log the thread ID in their log output. This is useful when
5713 working with heavily threaded programs.
5715 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
5716 not use the RDRAND CPU instruction. This is useful in environments
5717 such as replay debuggers where non-deterministic behaviour is not
5720 * The autopaging logic in systemd's various tools (such as systemctl)
5721 has been updated to turn on "secure" mode in "less"
5722 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
5723 detected. This disables invoking external programs from the pager,
5724 via the pipe logic. This behaviour may be overridden via the new
5725 $SYSTEMD_PAGERSECURE environment variable.
5727 * Units which have resource limits (.service, .mount, .swap, .slice,
5728 .socket, and .slice) gained new configuration settings
5729 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
5730 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
5731 limits and optional action taken by systemd-oomd.
5733 * A new service systemd-oomd has been added. It monitors resource
5734 contention for selected parts of the unit hierarchy using the PSI
5735 information reported by the kernel, and kills processes when memory
5736 or swap pressure is above configured limits. This service is only
5737 enabled by default in developer mode (see below) and should be
5738 considered a preview in this release. Behaviour details and option
5739 names are subject to change without the usual backwards-compatibility
5742 * A new helper oomctl has been added to introspect systemd-oomd state.
5743 It is only enabled by default in developer mode and should be
5744 considered a preview without the usual backwards-compatibility
5747 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
5748 enabled, systemd reads the system UID boundaries from /etc/login.defs
5749 at runtime, instead of using the built-in values selected during
5750 build. This is an option to improve compatibility for upgrades from
5751 old systems. It's strongly recommended not to make use of this
5752 functionality on new systems (or even enable it during build), as it
5753 makes something runtime-configurable that is mostly an implementation
5754 detail of the OS, and permits avoidable differences in deployments
5755 that create all kinds of problems in the long run.
5757 * New meson option '-Dmode=developer|release' has been added. When
5758 'developer', additional checks and features are enabled that are
5759 relevant during upstream development, e.g. verification that
5760 semi-automatically-generated documentation has been properly updated
5761 following API changes. Those checks are considered hints for
5762 developers and are not actionable in downstream builds. In addition,
5763 extra features that are not ready for general consumption may be
5764 enabled in developer mode. It is thus recommended to set
5765 '-Dmode=release' in end-user and distro builds.
5767 * systemd-cryptsetup gained support for processing detached LUKS
5768 headers specified on the kernel command line via the header=
5769 parameter of the luks.options= kernel command line option. The same
5770 device/path syntax as for key files is supported for header files
5773 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
5774 slot index data for devices that are connected through a PCI bridge
5775 where the _SUN index is associated with the bridge instead of the
5776 network device itself. Previously this would create ambiguous device
5777 naming if multiple network interfaces were connected to the same PCI
5778 bridge. Since this is a naming scheme incompatibility on systems that
5779 possess hardware like this it has been introduced as new naming
5780 scheme "v247". The previous scheme can be selected via the
5781 "net.naming-scheme=v245" kernel command line parameter.
5783 * ConditionFirstBoot= semantics have been modified to be safe towards
5784 abnormal system power-off during first boot. Specifically, the
5785 "systemd-machine-id-commit.service" service now acts as boot
5786 milestone indicating when the first boot process is sufficiently
5787 complete in order to not consider the next following boot also a
5788 first boot. If the system is reset before this unit is reached the
5789 first time, the next boot will still be considered a first boot; once
5790 it has been reached, no further boots will be considered a first
5791 boot. The "first-boot-complete.target" unit now acts as official hook
5792 point to order against this. If a service shall be run on every boot
5793 until the first boot fully succeeds it may thus be ordered before
5794 this target unit (and pull it in) and carry ConditionFirstBoot=
5797 * bootctl's set-default and set-oneshot commands now accept the three
5798 special strings "@default", "@oneshot", "@current" in place of a boot
5799 entry id. These strings are resolved to the current default and
5800 oneshot boot loader entry, as well as the currently booted one. Thus
5801 a command "bootctl set-default @current" may be used to make the
5802 currently boot menu item the new default for all subsequent boots.
5804 * "systemctl edit" has been updated to show the original effective unit
5805 contents in commented form in the text editor.
5807 * Units in user mode are now segregated into three new slices:
5808 session.slice (units that form the core of graphical session),
5809 app.slice ("normal" user applications), and background.slice
5810 (low-priority tasks). Unless otherwise configured, user units are
5811 placed in app.slice. The plan is to add resource limits and
5812 protections for the different slices in the future.
5814 * New GPT partition types for RISCV32/64 for the root and /usr
5815 partitions, and their associated Verity partitions have been defined,
5816 and are now understood by systemd-gpt-auto-generator, and the OS
5817 image dissection logic.
5819 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
5820 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
5821 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
5822 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
5823 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
5824 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
5825 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
5826 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
5827 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
5828 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
5829 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
5830 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
5831 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
5832 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
5833 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
5834 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
5835 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
5836 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
5837 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
5838 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
5839 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
5840 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
5841 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
5842 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
5843 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
5844 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
5845 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
5846 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
5847 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
5848 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
5849 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
5850 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
5851 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
5853 – Warsaw, 2020-11-26
5857 * The service manager gained basic support for cgroup v2 freezer. Units
5858 can now be suspended or resumed either using new systemctl verbs,
5859 freeze and thaw respectively, or via D-Bus.
5861 * PID 1 may now automatically load pre-compiled AppArmor policies from
5862 /etc/apparmor/earlypolicy during early boot.
5864 * The CPUAffinity= setting in service unit files now supports a new
5865 special value "numa" that causes the CPU affinity masked to be set
5866 based on the NUMA mask.
5868 * systemd will now log about all left-over processes remaining in a
5869 unit when the unit is stopped. It will now warn about services using
5870 KillMode=none, as this is generally an unsafe thing to make use of.
5872 * Two new unit file settings
5873 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
5874 added. They may be used to check whether a specific file system path
5875 resides on a block device that is encrypted on the block level
5876 (i.e. using dm-crypt/LUKS).
5878 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
5879 has been added that may be used for simple environment checks. This
5880 is particularly useful when passing in environment variables from a
5881 container manager (or from PAM in case of the systemd --user
5884 * .service unit files now accept a new setting CoredumpFilter= which
5885 allows configuration of the memory sections coredumps of the
5886 service's processes shall include.
5888 * .mount units gained a new ReadWriteOnly= boolean option. If set
5889 it will not be attempted to mount a file system read-only if mounting
5890 in read-write mode doesn't succeed. An option x-systemd.rw-only is
5891 available in /etc/fstab to control the same.
5893 * .socket units gained a new boolean setting PassPacketInfo=. If
5894 enabled, the kernel will attach additional per-packet metadata to all
5895 packets read from the socket, as an ancillary message. This controls
5896 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
5897 depending on socket type.
5899 * .service units gained a new setting RootHash= which may be used to
5900 specify the root hash for verity enabled disk images which are
5901 specified in RootImage=. RootVerity= may be used to specify a path to
5902 the Verity data matching a RootImage= file system. (The latter is
5903 only useful for images that do not contain the Verity data embedded
5904 into the same image that carries a GPT partition table following the
5905 Discoverable Partition Specification). Similarly, systemd-nspawn
5906 gained a new switch --verity-data= that takes a path to a file with
5907 the verity data of the disk image supplied in --image=, if the image
5908 doesn't contain the verity data itself.
5910 * .service units gained a new setting RootHashSignature= which takes
5911 either a base64 encoded PKCS#7 signature of the root hash specified
5912 with RootHash=, or a path to a file to read the signature from. This
5913 allows validation of the root hash against public keys available in
5914 the kernel keyring, and is only supported on recent kernels
5915 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
5916 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
5917 this mechanism has also been added to systemd-veritysetup.
5919 * .service unit files gained two new options
5920 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
5921 tune behaviour if a start or stop timeout is hit, i.e. whether to
5922 terminate the service with SIGTERM, SIGABRT or SIGKILL.
5924 * Most options in systemd that accept hexadecimal values prefixed with
5925 0x in additional to the usual decimal notation now also support octal
5926 notation when the 0o prefix is used and binary notation if the 0b
5929 * Various command line parameters and configuration file settings that
5930 configure key or certificate files now optionally take paths to
5931 AF_UNIX sockets in the file system. If configured that way a stream
5932 connection is made to the socket and the required data read from
5933 it. This is a simple and natural extension to the existing regular
5934 file logic, and permits other software to provide keys or
5935 certificates via simple IPC services, for example when unencrypted
5936 storage on disk is not desired. Specifically, systemd-networkd's
5937 Wireguard and MACSEC key file settings as well as
5938 systemd-journal-gatewayd's and systemd-journal-remote's PEM
5939 key/certificate parameters support this now.
5941 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
5942 configuration files that support specifier expansion learnt six new
5943 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
5944 resolve to the various ID fields from /etc/os-release, %l resolves to
5945 the "short" hostname of the system, i.e. the hostname configured in
5946 the kernel truncated at the first dot.
5948 * Support for the .include syntax in unit files has been removed. The
5949 concept has been obsolete for 6 years and we started warning about
5950 its pending removal 2 years ago (also see NEWS file below). It's
5953 * StandardError= and StandardOutput= in unit files no longer support
5954 the "syslog" and "syslog-console" switches. They were long removed
5955 from the documentation, but will now result in warnings when used,
5956 and be converted to "journal" and "journal+console" automatically.
5958 * If the service setting User= is set to the "nobody" user, a warning
5959 message is now written to the logs (but the value is nonetheless
5960 accepted). Setting User=nobody is unsafe, since the primary purpose
5961 of the "nobody" user is to own all files whose owner cannot be mapped
5962 locally. It's in particular used by the NFS subsystem and in user
5963 namespacing. By running a service under this user's UID it might get
5964 read and even write access to all these otherwise unmappable files,
5965 which is quite likely a major security problem.
5967 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
5968 and others) now have a size and inode limits applied (50% of RAM for
5969 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
5970 that the implicit kernel default is 50% too, so there is no change
5971 in the size limit for /tmp and /dev/shm.
5973 * nss-mymachines lost support for resolution of users and groups, and
5974 now only does resolution of hostnames. This functionality is now
5975 provided by nss-systemd. Thus, the 'mymachines' entry should be
5976 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
5977 (and 'systemd' added if it is not already there).
5979 * A new kernel command line option systemd.hostname= has been added
5980 that allows controlling the hostname that is initialized early during
5983 * A kernel command line option "udev.blockdev_read_only" has been
5984 added. If specified all hardware block devices that show up are
5985 immediately marked as read-only by udev. This option is useful for
5986 making sure that a specific boot under no circumstances modifies data
5987 on disk. Use "blockdev --setrw" to undo the effect of this, per
5990 * A new boolean kernel command line option systemd.swap= has been
5991 added, which may be used to turn off automatic activation of swap
5992 devices listed in /etc/fstab.
5994 * New kernel command line options systemd.condition-needs-update= and
5995 systemd.condition-first-boot= have been added, which override the
5996 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
5999 * A new kernel command line option systemd.clock-usec= has been added
6000 that allows setting the system clock to the specified time in µs
6001 since Jan 1st, 1970 early during boot. This is in particular useful
6002 in order to make test cases more reliable.
6004 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
6005 systemd-coredump to save core files for suid processes. When saving
6006 the core file, systemd-coredump will use the effective uid and gid of
6007 the process that faulted.
6009 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
6010 now automatically set to "Y" at boot, in order to enable pstore
6011 generation for collection with systemd-pstore.
6013 * We provide a set of udev rules to enable auto-suspend on PCI and USB
6014 devices that were tested to correctly support it. Previously, this
6015 was distributed as a set of udev rules, but has now been replaced by
6016 by a set of hwdb entries (and a much shorter udev rule to take action
6017 if the device modalias matches one of the new hwdb entries).
6019 As before, entries are periodically imported from the database
6020 maintained by the ChromiumOS project. If you have a device that
6021 supports auto-suspend correctly and where it should be enabled by
6022 default, please submit a patch that adds it to the database (see
6023 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
6025 * systemd-udevd gained the new configuration option timeout_signal= as well
6026 as a corresponding kernel command line option udev.timeout_signal=.
6027 The option can be used to configure the UNIX signal that the main
6028 daemon sends to the worker processes on timeout. Setting the signal
6029 to SIGABRT is useful for debugging.
6031 * .link files managed by systemd-udevd gained options RxFlowControl=,
6032 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
6033 order to configure various flow control parameters. They also gained
6034 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
6035 frame ring buffer sizes.
6037 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
6038 enabled systemd-networkd manages all routes configured by other tools.
6040 * .network files managed by systemd-networkd gained a new section
6041 [SR-IOV], in order to configure SR-IOV capable network devices.
6043 * systemd-networkd's [IPv6Prefix] section in .network files gained a
6044 new boolean setting Assign=. If enabled an address from the prefix is
6045 automatically assigned to the interface.
6047 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
6048 controls delegated prefixes assigned by DHCPv6 client. The section
6049 has three settings: SubnetID=, Assign=, and Token=. The setting
6050 SubnetID= allows explicit configuration of the preferred subnet that
6051 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
6052 Assign= is enabled (which is the default) an address from any acquired
6053 delegated prefix is automatically chosen and assigned to the
6054 interface. The setting Token= specifies an optional address generation
6057 * systemd-networkd's [Network] section gained a new setting
6058 IPv4AcceptLocal=. If enabled the interface accepts packets with local
6061 * systemd-networkd gained support for configuring the HTB queuing
6062 discipline in the [HierarchyTokenBucket] and
6063 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
6064 be configured in the [PFIFO] section, "GRED" in
6065 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
6066 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
6067 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
6068 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
6069 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
6070 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
6072 * systemd-networkd gained support for a new Termination= setting in the
6073 [CAN] section for configuring the termination resistor. It also
6074 gained a new ListenOnly= setting for controlling whether to only
6075 listen on CAN interfaces, without interfering with traffic otherwise
6076 (which is useful for debugging/monitoring CAN network
6077 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
6078 been added to configure various CAN-FD aspects.
6080 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
6081 When enabled, DHCPv6 will be attempted right-away without requiring an
6082 Router Advertisement packet suggesting it first (i.e. without the 'M'
6083 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
6084 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
6085 the RA packets suggest it.
6087 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
6088 which may be used to turn off use of the gateway information provided
6089 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
6090 used to configure how to process leases that lack a lifetime option.
6092 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
6093 setting SendVendorOption= allowing configuration of additional vendor
6094 options to send in the DHCP requests/responses. The [DHCPv6] section
6095 gained a new SendOption= setting for sending arbitrary DHCP
6096 options. RequestOptions= has been added to request arbitrary options
6097 from the server. UserClass= has been added to set the DHCP user class
6100 * systemd-networkd's [DHCPServer] section gained a new set of options
6101 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
6102 information about these three protocols in the DHCP lease. It also
6103 gained support for including "MUD" URLs ("Manufacturer Usage
6104 Description"). Support for "MUD" URLs was also added to the LLDP
6105 stack, configurable in the [LLDP] section in .network files.
6107 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
6108 mode. Also, the sections now support a new setting SourceMACAddress=.
6110 * systemd-networkd's .netdev files now support a new setting
6111 VLANProtocol= in the [Bridge] section that allows configuration of
6112 the VLAN protocol to use.
6114 * systemd-networkd supports a new Group= setting in the [Link] section
6115 of the .network files, to control the link group.
6117 * systemd-networkd's [Network] section gained a new
6118 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
6119 link local address is generated.
6121 * A new default .network file is now shipped that matches TUN/TAP
6122 devices that begin with "vt-" in their name. Such interfaces will
6123 have IP routing onto the host links set up automatically. This is
6124 supposed to be used by VM managers to trivially acquire a network
6125 interface which is fully set up for host communication, simply by
6126 carefully picking an interface name to use.
6128 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
6129 which sets the route priority for routes specified by the DHCP server.
6131 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
6132 which configures the vendor class information sent to DHCP server.
6134 * The BlackList= settings in .network files' [DHCPv4] and
6135 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
6136 are still understood to provide compatibility.
6138 * networkctl gained the new "forcerenew" command for forcing all DHCP
6139 server clients to renew their lease. The interface "status" output
6140 will now show numerous additional fields of information about an
6141 interface. There are new "up" and "down" commands to bring specific
6142 interfaces up or down.
6144 * systemd-resolved's DNS= configuration option now optionally accepts a
6145 port number (after ":") and a host name (after "#"). When the host
6146 name is specified, the DNS-over-TLS certificate is validated to match
6147 the specified hostname. Additionally, in case of IPv6 addresses, an
6148 interface may be specified (after "%").
6150 * systemd-resolved may be configured to forward single-label DNS names.
6151 This is not standard-conformant, but may make sense in setups where
6152 public DNS servers are not used.
6154 * systemd-resolved's DNS-over-TLS support gained SNI validation.
6156 * systemd-nspawn's --resolv-conf= switch gained a number of new
6157 supported values. Specifically, options starting with "replace-" are
6158 like those prefixed "copy-" but replace any existing resolv.conf
6159 file. And options ending in "-uplink" and "-stub" can now be used to
6160 propagate other flavours of resolv.conf into the container (as
6161 defined by systemd-resolved).
6163 * The various programs included in systemd can now optionally output
6164 their log messages on stderr prefixed with a timestamp, controlled by
6165 the $SYSTEMD_LOG_TIME environment variable.
6167 * systemctl gained a new "-P" switch that is a shortcut for "--value
6170 * "systemctl list-units" and "systemctl list-machines" no longer hide
6171 their first output column with --no-legend. To hide the first column,
6174 * "systemctl reboot" takes the option "--reboot-argument=".
6175 The optional positional argument to "systemctl reboot" is now
6176 being deprecated in favor of this option.
6178 * systemd-run gained a new switch --slice-inherit. If specified the
6179 unit it generates is placed in the same slice as the systemd-run
6182 * systemd-journald gained support for zstd compression of large fields
6183 in journal files. The hash tables in journal files have been hardened
6184 against hash collisions. This is an incompatible change and means
6185 that journal files created with new systemd versions are not readable
6186 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
6187 environment variable for systemd-journald.service is set to 0 this
6188 new hardening functionality may be turned off, so that generated
6189 journal files remain compatible with older journalctl
6192 * journalctl will now include a clickable link in the default output for
6193 each log message for which a URL with further documentation is
6194 known. This is only supported on terminal emulators that support
6195 clickable hyperlinks, and is turned off if a pager is used (since
6196 "less" still doesn't support hyperlinks,
6197 unfortunately). Documentation URLs may be included in log messages
6198 either by including a DOCUMENTATION= journal field in it, or by
6199 associating a journal message catalog entry with the log message's
6200 MESSAGE_ID, which then carries a "Documentation:" tag.
6202 * journald.conf gained a new boolean setting Audit= that may be used to
6203 control whether systemd-journald will enable audit during
6206 * when systemd-journald's log stream is broken up into multiple lines
6207 because the PID of the sender changed this is indicated in the
6208 generated log records via the _LINE_BREAK=pid-change field.
6210 * journalctl's "-o cat" output mode will now show one or more journal
6211 fields specified with --output-fields= instead of unconditionally
6212 MESSAGE=. This is useful to retrieve a very specific set of fields
6213 without any decoration.
6215 * The sd-journal.h API gained two new functions:
6216 sd_journal_enumerate_available_unique() and
6217 sd_journal_enumerate_available_data() that operate like their
6218 counterparts that lack the _available_ in the name, but skip items
6219 that cannot be read and processed by the local implementation
6220 (i.e. are compressed in an unsupported format or such),
6222 * coredumpctl gained a new --file= switch, matching the same one in
6223 journalctl: a specific journal file may be specified to read the
6226 * coredumps collected by systemd-coredump may now be compressed using
6229 * systemd-binfmt gained a new switch --unregister for unregistering all
6230 registered entries at once. This is now invoked automatically at
6231 shutdown, so that binary formats registered with the "F" flag will
6232 not block clean file system unmounting.
6234 * systemd-notify's --pid= switch gained new values: "parent", "self",
6235 "auto" for controlling which PID to send to the service manager: the
6236 systemd-notify process' PID, or the one of the process invoking it.
6238 * systemd-logind's Session bus object learnt a new method call
6239 SetType() for temporarily updating the session type of an already
6240 allocated session. This is useful for upgrading tty sessions to
6241 graphical ones once a compositor is invoked.
6243 * systemd-socket-proxy gained a new switch --exit-idle-time= for
6244 configuring an exit-on-idle time.
6246 * systemd-repart's --empty= setting gained a new value "create". If
6247 specified a new empty regular disk image file is created under the
6248 specified name. Its size may be specified with the new --size=
6249 option. The latter is also supported without the "create" mode, in
6250 order to grow existing disk image files to the specified size. These
6251 two new options are useful when creating or manipulating disk images
6252 instead of operating on actual block devices.
6254 * systemd-repart drop-ins now support a new UUID= setting to control
6255 the UUID to assign to a newly created partition.
6257 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
6260 * systemd-repart's Label= setting now support the usual, simple
6261 specifier expansion.
6263 * systemd-homed's LUKS backend gained the ability to discard empty file
6264 system blocks automatically when the user logs out. This is enabled
6265 by default to ensure that home directories take minimal space when
6266 logged out but get full size guarantees when logged in. This may be
6267 controlled with the new --luks-offline-discard= switch to homectl.
6269 * If systemd-homed detects that /home/ is encrypted as a whole it will
6270 now default to the directory or subvolume backends instead of the
6271 LUKS backend, in order to avoid double encryption. The default
6272 storage and file system may now be configured explicitly, too, via
6273 the new /etc/systemd/homed.conf configuration file.
6275 * systemd-homed now supports unlocking home directories with FIDO2
6276 security tokens that support the 'hmac-secret' extension, in addition
6277 to the existing support for PKCS#11 security token unlocking
6278 support. Note that many recent hardware security tokens support both
6279 interfaces. The FIDO2 support is accessible via homectl's
6280 --fido2-device= option.
6282 * homectl's --pkcs11-uri= setting now accepts two special parameters:
6283 if "auto" is specified and only one suitable PKCS#11 security token
6284 is plugged in, its URL is automatically determined and enrolled for
6285 unlocking the home directory. If "list" is specified a brief table of
6286 suitable PKCS#11 security tokens is shown. Similar, the new
6287 --fido2-device= option also supports these two special values, for
6288 automatically selecting and listing suitable FIDO2 devices.
6290 * The /etc/crypttab tmp option now optionally takes an argument
6291 selecting the file system to use. Moreover, the default is now
6292 changed from ext2 to ext4.
6294 * There's a new /etc/crypttab option "keyfile-erase". If specified the
6295 key file listed in the same line is removed after use, regardless if
6296 volume activation was successful or not. This is useful if the key
6297 file is only acquired transiently at runtime and shall be erased
6298 before the system continues to boot.
6300 * There's also a new /etc/crypttab option "try-empty-password". If
6301 specified, before asking the user for a password it is attempted to
6302 unlock the volume with an empty password. This is useful for
6303 installing encrypted images whose password shall be set on first boot
6304 instead of at installation time.
6306 * systemd-cryptsetup will now attempt to load the keys to unlock
6307 volumes with automatically from files in
6308 /etc/cryptsetup-keys.d/<volume>.key and
6309 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
6311 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
6312 /etc/crypttab, during boot.
6314 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
6315 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
6318 * A new generator systemd-xdg-autostart-generator has been added. It
6319 generates systemd unit files from XDG autostart .desktop files, and
6320 may be used to let the systemd user instance manage services that are
6321 started automatically as part of the desktop session.
6323 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
6324 to query and change the firmware's 'Reboot Into Firmware Interface'
6327 * systemd-firstboot gained a new switch --kernel-command-line= that may
6328 be used to initialize the /etc/kernel/cmdline file of the image. It
6329 also gained a new switch --root-password-hashed= which is like
6330 --root-password= but accepts a pre-hashed UNIX password as
6331 argument. The new option --delete-root-password may be used to unset
6332 any password for the root user (dangerous!). The --root-shell= switch
6333 may be used to control the shell to use for the root account. A new
6334 --force option may be used to override any already set settings with
6335 the parameters specified on the command line (by default, the tool
6336 will not override what has already been set before, i.e. is purely
6339 * systemd-firstboot gained support for a new --image= switch, which is
6340 similar to --root= but accepts the path to a disk image file, on
6341 which it then operates.
6343 * A new sd-path.h API has been added to libsystemd. It provides a
6344 simple API for retrieving various search paths and primary
6345 directories for various resources.
6347 * A new call sd_notify_barrier() has been added to the sd-daemon.h
6348 API. The call will block until all previously sent sd_notify()
6349 messages have been processed by the service manager. This is useful
6350 to remove races caused by a process already having disappeared at the
6351 time a notification message is processed by the service manager,
6352 making correct attribution impossible. The systemd-notify tool will
6353 now make use of this call implicitly, but this can be turned off again
6354 via the new --no-block switch.
6356 * When sending a file descriptor (fd) to the service manager to keep
6357 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
6358 may be specified. If passed the service manager will refrain from
6359 poll()ing on the file descriptor. Traditionally (and when the
6360 parameter is not specified), the service manager will poll it for
6361 POLLHUP or POLLERR events, and immediately close the fds in that
6364 * The service manager (PID1) gained a new D-Bus method call
6365 SetShowStatus() which may be used to control whether it shall show
6366 boot-time status output on the console. This method has a similar
6367 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
6369 * The sd-bus API gained a number of convenience functions that take
6370 va_list arguments rather than "...". For example, there's now
6371 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
6372 it easier to build wrappers that accept variadic arguments and want
6373 to pass a ready va_list structure to sd-bus.
6375 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
6376 flag which alters how the userdata pointer to pass to the callbacks
6377 is determined. When the flag is set, the offset field is converted
6378 as-is into a pointer, without adding it to the object pointer the
6379 vtable is associated with.
6381 * sd-bus now exposes four new functions:
6382 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
6383 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
6384 validate strings to check if they qualify as various D-Bus concepts.
6386 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
6387 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
6388 that simplify adding argument names to D-Bus methods and signals.
6390 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
6392 * Various D-Bus APIs of systemd daemons now have man pages that
6393 document the methods, signals and properties.
6395 * The expectations on user/group name syntax are now documented in
6396 detail; documentation on how classic home directories may be
6397 converted into home directories managed by homed has been added;
6398 documentation regarding integration of homed/userdb functionality in
6399 desktops has been added:
6401 https://systemd.io/USER_NAMES
6402 https://systemd.io/CONVERTING_TO_HOMED
6403 https://systemd.io/USERDB_AND_DESKTOPS
6405 * Documentation for the on-disk Journal file format has been updated
6406 and has now moved to:
6408 https://systemd.io/JOURNAL_FILE_FORMAT
6410 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
6411 has been extended by a set of environment variables that expose
6412 select fields from the host's os-release file to the container
6413 payload. Similarly, host's os-release files can be mounted into the
6414 container underneath /run/host. Together, those mechanisms provide a
6415 standardized way to expose information about the host to the
6416 container payload. Both interfaces are implemented in systemd-nspawn.
6418 * All D-Bus services shipped in systemd now implement the generic
6419 LogControl1 D-Bus API which allows clients to change log level +
6420 target of the service during runtime.
6422 * Only relevant for developers: the mkosi.default symlink has been
6423 dropped from version control. Please create a symlink to one of the
6424 distribution-specific defaults in .mkosi/ based on your preference.
6426 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
6427 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
6428 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
6429 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
6430 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
6431 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
6432 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
6433 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
6434 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
6435 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
6436 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
6437 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
6438 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
6439 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
6440 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
6441 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
6442 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
6443 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
6444 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
6445 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
6446 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
6447 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
6448 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
6449 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
6450 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
6451 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
6452 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
6453 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
6454 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
6455 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
6456 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
6457 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
6458 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
6459 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
6460 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
6461 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
6462 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
6463 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
6465 – Warsaw, 2020-07-30
6469 * A new tool "systemd-repart" has been added, that operates as an
6470 idempotent declarative repartitioner for GPT partition tables.
6471 Specifically, a set of partitions that must or may exist can be
6472 configured via drop-in files, and during every boot the partition
6473 table on disk is compared with these files, creating missing
6474 partitions or growing existing ones based on configurable relative
6475 and absolute size constraints. The tool is strictly incremental,
6476 i.e. does not delete, shrink or move partitions, but only adds and
6477 grows them. The primary use-case is OS images that ship in minimized
6478 form, that on first boot are grown to the size of the underlying
6479 block device or augmented with additional partitions. For example,
6480 the root partition could be extended to cover the whole disk, or a
6481 swap or /home partitions could be added on first boot. It can also be
6482 used for systems that use an A/B update scheme but ship images with
6483 just the A partition, with B added on first boot. The tool is
6484 primarily intended to be run in the initrd, shortly before
6485 transitioning into the host OS, but can also be run after the
6486 transition took place. It automatically discovers the disk backing
6487 the root file system, and should hence not require any additional
6488 configuration besides the partition definition drop-ins. If no
6489 configuration drop-ins are present, no action is taken.
6491 * A new component "userdb" has been added, along with a small daemon
6492 "systemd-userdbd.service" and a client tool "userdbctl". The framework
6493 allows defining rich user and group records in a JSON format,
6494 extending on the classic "struct passwd" and "struct group"
6495 structures. Various components in systemd have been updated to
6496 process records in this format, including systemd-logind and
6497 pam-systemd. The user records are intended to be extensible, and
6498 allow setting various resource management, security and runtime
6499 parameters that shall be applied to processes and sessions of the
6500 user as they log in. This facility is intended to allow associating
6501 such metadata directly with user/group records so that they can be
6502 produced, extended and consumed in unified form. We hope that
6503 eventually frameworks such as sssd will generate records this way, so
6504 that for the first time resource management and various other
6505 per-user settings can be configured in LDAP directories and then
6506 provided to systemd (specifically to systemd-logind and pam-system)
6507 to apply on login. For further details see:
6509 https://systemd.io/USER_RECORD
6510 https://systemd.io/GROUP_RECORD
6511 https://systemd.io/USER_GROUP_API
6513 * A small new service systemd-homed.service has been added, that may be
6514 used to securely manage home directories with built-in encryption.
6515 The complete user record data is unified with the home directory,
6516 thus making home directories naturally migratable. Its primary
6517 back-end is based on LUKS volumes, but fscrypt, plain directories,
6518 and other storage schemes are also supported. This solves a couple of
6519 problems we saw with traditional ways to manage home directories, in
6520 particular when it comes to encryption. For further discussion of
6521 this, see the video of Lennart's talk at AllSystemsGo! 2019:
6523 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
6525 For further details about the format and expectations on home
6526 directories this new daemon makes, see:
6528 https://systemd.io/HOME_DIRECTORY
6530 * systemd-journald is now multi-instantiable. In addition to the main
6531 instance systemd-journald.service there's now a template unit
6532 systemd-journald@.service, with each instance defining a new named
6533 log 'namespace' (whose name is specified via the instance part of the
6534 unit name). A new unit file setting LogNamespace= has been added,
6535 taking such a namespace name, that assigns services to the specified
6536 log namespaces. As each log namespace is serviced by its own
6537 independent journal daemon, this functionality may be used to improve
6538 performance and increase isolation of applications, at the price of
6539 losing global message ordering. Each instance of journald has a
6540 separate set of configuration files, with possibly different disk
6541 usage limitations and other settings.
6543 journalctl now takes a new option --namespace= to show logs from a
6544 specific log namespace. The sd-journal.h API gained
6545 sd_journal_open_namespace() for opening the log stream of a specific
6546 log namespace. systemd-journald also gained the ability to exit on
6547 idle, which is useful in the context of log namespaces, as this means
6548 log daemons for log namespaces can be activated automatically on
6549 demand and will stop automatically when no longer used, minimizing
6552 * When systemd-tmpfiles copies a file tree using the 'C' line type it
6553 will now label every copied file according to the SELinux database.
6555 * When systemd/PID 1 detects it is used in the initrd it will now boot
6556 into initrd.target rather than default.target by default. This should
6557 make it simpler to build initrds with systemd as for many cases the
6558 only difference between a host OS image and an initrd image now is
6559 the presence of the /etc/initrd-release file.
6561 * A new kernel command line option systemd.cpu_affinity= is now
6562 understood. It's equivalent to the CPUAffinity= option in
6563 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
6564 itself and the default for all other processes.
6566 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
6567 equivalent), the SELinux database is now reloaded, ensuring that
6568 sockets and other file system objects are generated taking the new
6569 database into account.
6571 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
6572 "quiet" has been changed to imply that instead of
6573 "systemd.show-status=auto". In this mode, only messages about errors
6574 and significant delays in boot are shown on the console.
6576 * The sd-event.h API gained native support for the new Linux "pidfd"
6577 concept. This permits watching processes using file descriptors
6578 instead of PID numbers, which fixes a number of races and makes
6579 process supervision more robust and efficient. All of systemd's
6580 components will now use pidfds if the kernel supports it for process
6581 watching, with the exception of PID 1 itself, unfortunately. We hope
6582 to move PID 1 to exclusively using pidfds too eventually, but this
6583 requires some more kernel work first. (Background: PID 1 watches
6584 processes using waitid() with the P_ALL flag, and that does not play
6585 together nicely with pidfds yet.)
6587 * Closely related to this, the sd-event.h API gained two new calls
6588 sd_event_source_send_child_signal() (for sending a signal to a
6589 watched process) and sd_event_source_get_child_process_own() (for
6590 marking a process so that it is killed automatically whenever the
6591 event source watching it is freed).
6593 * systemd-networkd gained support for configuring Token Bucket Filter
6594 (TBF) parameters in its qdisc configuration support. Similarly,
6595 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
6596 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
6598 * systemd-networkd gained support for Intermediate Functional Block
6599 (IFB) network devices.
6601 * systemd-networkd gained support for configuring multi-path IP routes,
6602 using the new MultiPathRoute= setting in the [Route] section.
6604 * systemd-networkd's DHCPv4 client has been updated to support a new
6605 SendDecline= option. If enabled, duplicate address detection is done
6606 after a DHCP offer is received from the server. If a conflict is
6607 detected, the address is declined. The DHCPv4 client also gained
6608 support for a new RouteMTUBytes= setting that allows to configure the
6609 MTU size to be used for routes generated from DHCPv4 leases.
6611 * The PrefixRoute= setting in systemd-networkd's [Address] section of
6612 .network files has been deprecated, and replaced by AddPrefixRoute=,
6613 with its sense inverted.
6615 * The Gateway= setting of [Route] sections of .network files gained
6616 support for a special new value "_dhcp". If set, the configured
6617 static route uses the gateway host configured via DHCP.
6619 * New User= and SuppressPrefixLength= settings have been implemented
6620 for the [RoutingPolicyRule] section of .network files to configure
6621 source routing based on UID ranges and prefix length, respectively.
6623 * The Type= match property of .link files has been generalized to
6624 always match the device type shown by 'networkctl status', even for
6625 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
6628 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
6629 D-Bus message object as "sensitive". Those objects are erased from
6630 memory when they are freed. This concept is intended to be used for
6631 messages that contain security sensitive data. A new flag
6632 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
6633 in sd-bus vtables, causing any incoming and outgoing messages of
6634 those methods to be implicitly marked as "sensitive".
6636 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
6637 contents of a message (or parts thereof) to standard output for
6640 * systemd-sysusers gained support for creating users with the primary
6641 group named differently than the user.
6643 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
6644 gained support for growing XFS partitions. Previously it supported
6645 only ext4 and btrfs partitions.
6647 * The support for /etc/crypttab gained a new x-initrd.attach option. If
6648 set, the specified encrypted volume is unlocked already in the
6649 initrd. This concept corresponds to the x-initrd.mount option in
6652 * systemd-cryptsetup gained native support for unlocking encrypted
6653 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
6654 encryption of volumes to YubiKeys. This is exposed in the new
6655 pkcs11-uri= option in /etc/crypttab.
6657 * The /etc/fstab support in systemd now supports two new mount options
6658 x-systemd.{required,wanted}-by=, for explicitly configuring the units
6659 that the specified mount shall be pulled in by, in place of
6660 the usual local-fs.target/remote-fs.target.
6662 * The https://systemd.io/ web site has been relaunched, directly
6663 populated with most of the documentation included in the systemd
6664 repository. systemd also acquired a new logo, thanks to Tobias
6667 * systemd-udevd gained support for managing "alternative" network
6668 interface names, as supported by new Linux kernels. For the first
6669 time this permits assigning multiple (and longer!) names to a network
6670 interface. systemd-udevd will now by default assign the names
6671 generated via all supported naming schemes to each interface. This
6672 may be further tweaked with .link files and the AlternativeName= and
6673 AlternativeNamesPolicy= settings. Other components of systemd have
6674 been updated to support the new alternative names wherever
6675 appropriate. For example, systemd-nspawn will now generate
6676 alternative interface names for the host-facing side of container
6677 veth links based on the full container name without truncation.
6679 * systemd-nspawn interface naming logic has been updated in another way
6680 too: if the main interface name (i.e. as opposed to new-style
6681 "alternative" names) based on the container name is truncated, a
6682 simple hashing scheme is used to give different interface names to
6683 multiple containers whose names all begin with the same prefix. Since
6684 this changes the primary interface names pointing to containers if
6685 truncation happens, the old scheme may still be requested by
6686 selecting an older naming scheme, via the net.naming-scheme= kernel
6687 command line option.
6689 * PrivateUsers= in service files now works in services run by the
6690 systemd --user per-user instance of the service manager.
6692 * A new per-service sandboxing option ProtectClock= has been added that
6693 locks down write access to the system clock. It takes away device
6694 node access to /dev/rtc as well as the system calls that set the
6695 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
6696 Note that this option does not affect access to auxiliary services
6697 that allow changing the clock, for example access to
6700 * The systemd-id128 tool gained a new "show" verb for listing or
6701 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
6702 GPT partition table types.
6704 * The Discoverable Partitions Specification has been updated to support
6705 /var and /var/tmp partition discovery. Support for this has been
6706 added to systemd-gpt-auto-generator. For details see:
6708 https://systemd.io/DISCOVERABLE_PARTITIONS
6710 * "systemctl list-unit-files" has been updated to show a new column
6711 with the suggested enablement state based on the vendor preset files
6712 for the respective units.
6714 * "systemctl" gained a new option "--with-dependencies". If specified
6715 commands such as "systemctl status" or "systemctl cat" will now show
6716 all specified units along with all units they depend on.
6718 * networkctl gained support for showing per-interface logs in its
6721 * systemd-networkd-wait-online gained support for specifying the maximum
6722 operational state to wait for, and to wait for interfaces to
6725 * The [Match] section of .link and .network files now supports a new
6726 option PermanentMACAddress= which may be used to check against the
6727 permanent MAC address of a network device even if a randomized MAC
6730 * The [TrafficControlQueueingDiscipline] section in .network files has
6731 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
6732 dropped from the individual setting names.
6734 * Any .link and .network files that have an empty [Match] section (this
6735 also includes empty and commented-out files) will now be
6736 rejected. systemd-udev and systemd-networkd started warning about
6737 such files in version 243.
6739 * systemd-logind will now validate access to the operation of changing
6740 the virtual terminal via a polkit action. By default, only users
6741 with at least one session on a local VT are granted permission.
6743 * When systemd sets up PAM sessions that invoked service processes
6744 shall run in, the pam_setcred() API is now invoked, thus permitting
6745 PAM modules to set additional credentials for the processes.
6747 * portablectl attach/detach verbs now accept --now and --enable options
6748 to combine attachment with enablement and invocation, or detachment
6749 with stopping and disablement.
6751 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
6752 fixed, which in turn exposed bugs in unit configuration of services
6753 which have Type=oneshot and should only run once, but do not have
6754 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
6755 service may be started again after exiting successfully, for example
6756 as a dependency in another transaction. Affected services included
6757 some internal systemd services (most notably
6758 systemd-vconsole-setup.service, which was updated to have
6759 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
6760 plymouth has been suitably updated or patched before upgrading to
6761 this systemd release. See
6762 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
6763 additional discussion.
6765 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
6766 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
6767 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
6768 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
6769 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
6770 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
6771 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
6772 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
6773 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
6774 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
6775 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
6776 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
6777 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
6778 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
6779 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
6780 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
6781 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
6782 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
6783 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
6784 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
6785 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
6786 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
6787 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
6788 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
6791 – Warsaw, 2020-03-06
6795 * Support for the cpuset cgroups v2 controller has been added.
6796 Processes may be restricted to specific CPUs using the new
6797 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
6798 AllowedMemoryNodes= setting.
6800 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
6801 now be configured using a new RestartKillSignal= setting. This
6802 allows units which signals to request termination to implement
6803 different behaviour when stopping in preparation for a restart.
6805 * "systemctl clean" may now be used also for socket, mount, and swap
6808 * systemd will also read configuration options from the EFI variable
6809 SystemdOptions. This may be used to configure systemd behaviour when
6810 modifying the kernel command line is inconvenient, but configuration
6811 on disk is read too late, for example for the options related to
6812 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
6813 set the EFI variable.
6815 * systemd will now disable printk ratelimits in early boot. This should
6816 allow us to capture more logs from the early boot phase where normal
6817 storage is not available and the kernel ring buffer is used for
6818 logging. Configuration on the kernel command line has higher priority
6819 and overrides the systemd setting.
6821 systemd programs which log to /dev/kmsg directly use internal
6822 ratelimits to prevent runaway logging. (Normally this is only used
6823 during early boot, so in practice this change has very little
6826 * Unit files now support top level dropin directories of the form
6827 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
6828 that affects all corresponding unit files.
6830 * systemctl gained support for 'stop --job-mode=triggering' which will
6831 stop the specified unit and any units which could trigger it.
6833 * Unit status display now includes units triggering and triggered by
6834 the unit being shown.
6836 * The RuntimeMaxSec= setting is now supported by scopes, not just
6837 .service units. This is particularly useful for PAM sessions which
6838 create a scope unit for the user login. systemd.runtime_max_sec=
6839 setting may used with the pam_systemd module to limit the duration
6840 of the PAM session, for example for time-limited logins.
6842 * A new @pkey system call group is now defined to make it easier to
6843 allow-list memory protection syscalls for containers and services
6844 which need to use them.
6846 * systemd-udevd: removed the 30s timeout for killing stale workers on
6847 exit. systemd-udevd now waits for workers to finish. The hard-coded
6848 exit timeout of 30s was too short for some large installations, where
6849 driver initialization could be prematurely interrupted during initrd
6850 processing if the root file system had been mounted and init was
6851 preparing to switch root. If udevd is run without systemd and workers
6852 are hanging while udevd receives an exit signal, udevd will now exit
6853 when udev.event_timeout is reached for the last hanging worker. With
6854 systemd, the exit timeout can additionally be configured using
6855 TimeoutStopSec= in systemd-udevd.service.
6857 * udev now provides a program (fido_id) that identifies FIDO CTAP1
6858 ("U2F")/CTAP2 security tokens based on the usage declared in their
6859 report and descriptor and outputs suitable environment variables.
6860 This replaces the externally maintained allow lists of all known
6861 security tokens that were used previously.
6863 * Automatically generated autosuspend udev rules for allow-listed
6864 devices have been imported from the Chromium OS project. This should
6865 improve power saving with many more devices.
6867 * udev gained a new "CONST{key}=value" setting that allows matching
6868 against system-wide constants without forking a helper binary.
6869 Currently "arch" and "virt" keys are supported.
6871 * udev now opens CDROMs in non-exclusive mode when querying their
6872 capabilities. This should fix issues where other programs trying to
6873 use the CDROM cannot gain access to it, but carries a risk of
6874 interfering with programs writing to the disk, if they did not open
6875 the device in exclusive mode as they should.
6877 * systemd-networkd does not create a default route for IPv4 link local
6878 addressing anymore. The creation of the route was unexpected and was
6879 breaking routing in various cases, but people who rely on it being
6880 created implicitly will need to adjust. Such a route may be requested
6881 with DefaultRouteOnDevice=yes.
6883 Similarly, systemd-networkd will not assign a link-local IPv6 address
6884 when IPv6 link-local routing is not enabled.
6886 * Receive and transmit buffers may now be configured on links with
6887 the new RxBufferSize= and TxBufferSize= settings.
6889 * systemd-networkd may now advertise additional IPv6 routes. A new
6890 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
6893 * systemd-networkd may now configure "next hop" routes using the
6894 [NextHop] section and Gateway= and Id= settings.
6896 * systemd-networkd will now retain DHCP config on restarts by default
6897 (but this may be overridden using the KeepConfiguration= setting).
6898 The default for SendRelease= has been changed to true.
6900 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
6901 received from the server.
6903 The client will use the received SIP server list if UseSIP=yes is
6906 The client may be configured to request specific options from the
6907 server using a new RequestOptions= setting.
6909 The client may be configured to send arbitrary options to the server
6910 using a new SendOption= setting.
6912 A new IPServiceType= setting has been added to configure the "IP
6913 service type" value used by the client.
6915 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
6916 request prefix hints in the DHCPv6 solicitation.
6918 * The DHCPv4 server may be configured to send arbitrary options using
6919 a new SendOption= setting.
6921 * The DHCPv4 server may now be configured to emit SIP server list using
6922 the new EmitSIP= and SIP= settings.
6924 * systemd-networkd and networkctl may now renew DHCP leases on demand.
6925 networkctl has a new 'networkctl renew' verb.
6927 * systemd-networkd may now reconfigure links on demand. networkctl
6928 gained two new verbs: "reload" will reload the configuration, and
6929 "reconfigure DEVICE…" will reconfigure one or more devices.
6931 * .network files may now match on SSID and BSSID of a wireless network,
6932 i.e. the access point name and hardware address using the new SSID=
6933 and BSSID= options. networkctl will display the current SSID and
6934 BSSID for wireless links.
6936 .network files may also match on the wireless network type using the
6937 new WLANInterfaceType= option.
6939 * systemd-networkd now includes default configuration that enables
6940 link-local addressing when connected to an ad-hoc wireless network.
6942 * systemd-networkd may configure the Traffic Control queueing
6943 disciplines in the kernel using the new
6944 [TrafficControlQueueingDiscipline] section and Parent=,
6945 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
6946 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
6947 NetworkEmulatorDuplicateRate= settings.
6949 * systemd-tmpfiles gained a new w+ setting to append to files.
6951 * systemd-analyze dump will now report when the memory configuration in
6952 the kernel does not match what systemd has configured (usually,
6953 because some external program has modified the kernel configuration
6956 * systemd-analyze gained a new --base-time= switch instructs the
6957 'calendar' verb to resolve times relative to that timestamp instead
6958 of the present time.
6960 * journalctl --update-catalog now produces deterministic output (making
6961 reproducible image builds easier).
6963 * A new devicetree-overlay setting is now documented in the Boot Loader
6966 * The default value of the WatchdogSec= setting used in systemd
6967 services (the ones bundled with the project itself) may be set at
6968 configuration time using the -Dservice-watchdog= setting. If set to
6969 empty, the watchdogs will be disabled.
6971 * systemd-resolved validates IP addresses in certificates now when GnuTLS
6974 * libcryptsetup >= 2.0.1 is now required.
6976 * A configuration option -Duser-path= may be used to override the $PATH
6977 used by the user service manager. The default is again to use the same
6978 path as the system manager.
6980 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
6981 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
6984 * Service units gained a new sandboxing option ProtectKernelLogs= which
6985 makes sure the program cannot get direct access to the kernel log
6986 buffer anymore, i.e. the syslog() system call (not to be confused
6987 with the API of the same name in libc, which is not affected), the
6988 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
6989 inaccessible to the service. It's recommended to enable this setting
6990 for all services that should not be able to read from or write to the
6991 kernel log buffer, which are probably almost all.
6993 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
6994 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
6995 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
6996 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
6997 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
6998 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
6999 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
7000 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
7001 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
7002 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
7003 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
7004 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
7005 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
7006 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
7007 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
7008 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
7009 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
7010 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
7011 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
7012 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
7013 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
7014 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
7015 Zach Smith, Zbigniew Jędrzejewski-Szmek
7017 – Warsaw, 2019-11-29
7021 * This release enables unprivileged programs (i.e. requiring neither
7022 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
7023 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
7024 kernel for the whole UNIX group range, i.e. all processes. This
7025 change should be reasonably safe, as the kernel support for it was
7026 specifically implemented to allow safe access to ICMP Echo for
7027 processes lacking any privileges. If this is not desirable, it can be
7028 disabled again by setting the parameter to "1 0".
7030 * Previously, filters defined with SystemCallFilter= would have the
7031 effect that any calling of an offending system call would terminate
7032 the calling thread. This behaviour never made much sense, since
7033 killing individual threads of unsuspecting processes is likely to
7034 create more problems than it solves. With this release the default
7035 action changed from killing the thread to killing the whole
7036 process. For this to work correctly both a kernel version (>= 4.14)
7037 and a libseccomp version (>= 2.4.0) supporting this new seccomp
7038 action is required. If an older kernel or libseccomp is used the old
7039 behaviour continues to be used. This change does not affect any
7040 services that have no system call filters defined, or that use
7041 SystemCallErrorNumber= (and thus see EPERM or another error instead
7042 of being killed when calling an offending system call). Note that
7043 systemd documentation always claimed that the whole process is
7044 killed. With this change behaviour is thus adjusted to match the
7047 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
7048 4194304 by default, i.e. the full 22bit range the kernel allows, up
7049 from the old 16-bit range. This should improve security and
7050 robustness, as PID collisions are made less likely (though certainly
7051 still possible). There are rumours this might create compatibility
7052 problems, though at this moment no practical ones are known to
7053 us. Downstream distributions are hence advised to undo this change in
7054 their builds if they are concerned about maximum compatibility, but
7055 for everybody else we recommend leaving the value bumped. Besides
7056 improving security and robustness this should also simplify things as
7057 the maximum number of allowed concurrent tasks was previously bounded
7058 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
7059 only a single knob is left ("kernel.threads-max"). There have been
7060 concerns that usability is affected by this change because larger PID
7061 numbers are harder to type, but we believe the change from 5 digits
7062 to 7 digits doesn't hamper usability.
7064 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
7065 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
7066 hierarchically set default memory protection values for a particular
7067 subtree of the unit hierarchy.
7069 * Memory protection directives can now take a value of zero, allowing
7070 explicit opting out of a default value propagated by an ancestor.
7072 * systemd now defaults to the "unified" cgroup hierarchy setup during
7073 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
7074 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
7075 change reflects the fact that cgroupsv2 support has matured
7076 substantially in both systemd and in the kernel, and is clearly the
7077 way forward. Downstream production distributions might want to
7078 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
7079 their builds as unfortunately the popular container managers have not
7080 caught up with the kernel API changes.
7082 * Man pages are not built by default anymore (html pages were already
7083 disabled by default), to make development builds quicker. When
7084 building systemd for a full installation with documentation, meson
7085 should be called with -Dman=true and/or -Dhtml=true as appropriate.
7086 The default was changed based on the assumption that quick one-off or
7087 repeated development builds are much more common than full optimized
7088 builds for installation, and people need to pass various other
7089 options to when doing "proper" builds anyway, so the gain from making
7090 development builds quicker is bigger than the one time disruption for
7093 Two scripts are created in the *build* directory to generate and
7094 preview man and html pages on demand, e.g.:
7096 build/man/man systemctl
7097 build/man/html systemd.index
7099 * libidn2 is used by default if both libidn2 and libidn are installed.
7100 Please use -Dlibidn=true if libidn is preferred.
7102 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
7103 big-endian machines. Before, bytes were written and read in native
7104 machine order as exposed by the native libc __cpu_mask interface.
7105 Now, little-endian order is always used (CPUs 0–7 are described by
7106 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
7107 This change fixes D-Bus calls that cross endianness boundary.
7109 The presentation format used for CPUAffinity= by "systemctl show" and
7110 "systemd-analyze dump" is changed to present CPU indices instead of
7111 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
7112 shown as CPUAffinity=03000000000000000000000000000… (on
7113 little-endian) or CPUAffinity=00000000000000300000000000000… (on
7114 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
7115 input format. The maximum integer that will be printed in the new
7116 format is 8191 (four digits), while the old format always used a very
7117 long number (with the length varying by architecture), so they can be
7118 unambiguously distinguished.
7120 * /usr/sbin/halt.local is no longer supported. Implementation in
7121 distributions was inconsistent and it seems this functionality was
7124 To replace this functionality, users should:
7125 - either define a new unit and make it a dependency of final.target
7126 (systemctl add-wants final.target my-halt-local.service)
7127 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
7128 and ensure that it accepts "halt", "poweroff", "reboot", and
7129 "kexec" as an argument, see the description in systemd-shutdown(8).
7131 * When a [Match] section in .link or .network file is empty (contains
7132 no match patterns), a warning will be emitted. Please add any "match
7133 all" pattern instead, e.g. OriginalName=* or Name=* in case all
7134 interfaces should really be matched.
7136 * A new setting NUMAPolicy= may be used to set process memory
7137 allocation policy. This setting can be specified in
7138 /etc/systemd/system.conf and hence will set the default policy for
7139 PID1. The default policy can be overridden on a per-service
7140 basis. The related setting NUMAMask= is used to specify NUMA node
7141 mask that should be associated with the selected policy.
7143 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
7144 generates when processes it manages are reaching their memory limits,
7145 and will place their units in a special state, and optionally kill or
7146 stop the whole unit.
7148 * The service manager will now expose bus properties for the IO
7149 resources used by units. This information is also shown in "systemctl
7150 status" now (for services that have IOAccounting=yes set). Moreover,
7151 the IO accounting data is included in the resource log message
7152 generated whenever a unit stops.
7154 * Units may now configure an explicit timeout to wait for when killed
7155 with SIGABRT, for example when a service watchdog is hit. Previously,
7156 the regular TimeoutStopSec= timeout was applied in this case too —
7157 now a separate timeout may be set using TimeoutAbortSec=.
7159 * Services may now send a special WATCHDOG=trigger message with
7160 sd_notify() to trigger an immediate "watchdog missed" event, and thus
7161 trigger service termination. This is useful both for testing watchdog
7162 handling, but also for defining error paths in services, that shall
7163 be handled the same way as watchdog events.
7165 * There are two new per-unit settings IPIngressFilterPath= and
7166 IPEgressFilterPath= which allow configuration of a BPF program
7167 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
7168 to apply to the IP packet ingress/egress path of all processes of a
7169 unit. This is useful to allow running systemd services with BPF
7170 programs set up externally.
7172 * systemctl gained a new "clean" verb for removing the state, cache,
7173 runtime or logs directories of a service while it is terminated. The
7174 new verb may also be used to remove the state maintained on disk for
7175 timer units that have Persistent= configured.
7177 * During the last phase of shutdown systemd will now automatically
7178 increase the log level configured in the "kernel.printk" sysctl so
7179 that any relevant loggable events happening during late shutdown are
7180 made visible. Previously, loggable events happening so late during
7181 shutdown were generally lost if the "kernel.printk" sysctl was set to
7182 high thresholds, as regular logging daemons are terminated at that
7183 time and thus nothing is written to disk.
7185 * If processes terminated during the last phase of shutdown do not exit
7186 quickly systemd will now show their names after a short time, to make
7187 debugging easier. After a longer timeout they are forcibly killed,
7190 * journalctl (and the other tools that display logs) will now highlight
7191 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
7192 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
7193 are now shown in blue color, to separate them visually from regular
7194 logs. References to configuration files are now turned into clickable
7195 links on terminals that support that.
7197 * systemd-journald will now stop logging to /var/log/journal during
7198 shutdown when /var/ is on a separate mount, so that it can be
7199 unmounted safely during shutdown.
7201 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
7203 * systemd-resolved "Cache=" configuration option in resolved.conf has
7204 been extended to also accept the 'no-negative' value. Previously,
7205 only a boolean option was allowed (yes/no), having yes as the
7206 default. If this option is set to 'no-negative', negative answers are
7207 not cached while the old cache heuristics are used positive answers.
7208 The default remains unchanged.
7210 * The predictable naming scheme for network devices now supports
7211 generating predictable names for "netdevsim" devices.
7213 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
7216 Those two changes form a new net.naming-policy-scheme= entry.
7217 Distributions which want to preserve naming stability may want to set
7218 the -Ddefault-net-naming-scheme= configuration option.
7220 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
7221 interfaces natively.
7223 * systemd-networkd's bridge FDB support now allows configuration of a
7224 destination address for each entry (Destination=), as well as the
7225 VXLAN VNI (VNI=), as well as an option to declare what an entry is
7226 associated with (AssociatedWith=).
7228 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
7229 option for configuring the maximum number of DHCP lease requests. It
7230 also learnt a new BlackList= option for deny-listing DHCP servers (a
7231 similar setting has also been added to the IPv6 RA client), as well
7232 as a SendRelease= option for configuring whether to send a DHCP
7233 RELEASE message when terminating.
7235 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
7236 separately in the [DHCPv4] and [DHCPv6] sections.
7238 * systemd-networkd's DHCP support will now optionally create an
7239 implicit host route to the DNS server specified in the DHCP lease, in
7240 addition to the routes listed explicitly in the lease. This should
7241 ensure that in multi-homed systems DNS traffic leaves the systems on
7242 the interface that acquired the DNS server information even if other
7243 routes such as default routes exist. This behaviour may be turned on
7244 with the new RoutesToDNS= option.
7246 * systemd-networkd's VXLAN support gained a new option
7247 GenericProtocolExtension= for enabling VXLAN Generic Protocol
7248 Extension support, as well as IPDoNotFragment= for setting the IP
7249 "Don't fragment" bit on outgoing packets. A similar option has been
7250 added to the GENEVE support.
7252 * In systemd-networkd's [Route] section you may now configure
7253 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
7254 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
7255 propagation. The Type= setting now supports local, broadcast,
7256 anycast, multicast, any, xresolve routes, too.
7258 * systemd-networkd's [Network] section learnt a new option
7259 DefaultRouteOnDevice= for automatically configuring a default route
7260 onto the network device.
7262 * systemd-networkd's bridging support gained two new options ProxyARP=
7263 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
7264 MulticastRouter= for configuring multicast routing behaviour. A new
7265 option MulticastIGMPVersion= may be used to change bridge's multicast
7266 Internet Group Management Protocol (IGMP) version.
7268 * systemd-networkd's FooOverUDP support gained the ability to configure
7269 local and peer IP addresses via Local= and Peer=. A new option
7270 PeerPort= may be used to configure the peer's IP port.
7272 * systemd-networkd's TUN support gained a new setting VnetHeader= for
7273 tweaking Generic Segment Offload support.
7275 * The address family for policy rules may be specified using the new
7276 Family= option in the [RoutingPolicyRule] section.
7278 * networkctl gained a new "delete" command for removing virtual network
7279 devices, as well as a new "--stats" switch for showing device
7282 * networkd.conf gained a new setting SpeedMeter= and
7283 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
7284 measured speed may be shown by 'networkctl status'.
7286 * "networkctl status" now displays MTU and queue lengths, and more
7287 detailed information about VXLAN and bridge devices.
7289 * systemd-networkd's .network and .link files gained a new Property=
7290 setting in the [Match] section, to match against devices with
7291 specific udev properties.
7293 * systemd-networkd's tunnel support gained a new option
7294 AssignToLoopback= for selecting whether to use the loopback device
7295 "lo" as underlying device.
7297 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
7298 been renamed to LinkLayerAddress=, and it now allows configuration of
7301 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
7302 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
7303 IPv6 configuration (static or DHCPv6) was found for a given
7304 interface. It will not touch the sysctl otherwise.
7306 * The order of entries is $PATH used by the user manager instance was
7307 changed to put bin/ entries before the corresponding sbin/ entries.
7308 It is recommended to not rely on this order, and only ever have one
7309 binary with a given name in the system paths under /usr.
7311 * A new tool systemd-network-generator has been added that may generate
7312 .network, .netdev and .link files from IP configuration specified on
7313 the kernel command line in the format used by Dracut.
7315 * The CriticalConnection= setting in .network files is now deprecated,
7316 and replaced by a new KeepConfiguration= setting which allows more
7317 detailed configuration of the IP configuration to keep in place.
7319 * systemd-analyze gained a few new verbs:
7321 - "systemd-analyze timestamp" parses and converts timestamps. This is
7322 similar to the existing "systemd-analyze calendar" command which
7323 does the same for recurring calendar events.
7325 - "systemd-analyze timespan" parses and converts timespans (i.e.
7326 durations as opposed to points in time).
7328 - "systemd-analyze condition" will parse and test ConditionXYZ=
7331 - "systemd-analyze exit-status" will parse and convert exit status
7332 codes to their names and back.
7334 - "systemd-analyze unit-files" will print a list of all unit
7335 file paths and unit aliases.
7337 * SuccessExitStatus=, RestartPreventExitStatus=, and
7338 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
7339 is equivalent to "65"). Those exit status name mappings may be
7340 displayed with the systemd-analyze exit-status verb describe above.
7342 * systemd-logind now exposes a per-session SetBrightness() bus call,
7343 which may be used to securely change the brightness of a kernel
7344 brightness device, if it belongs to the session's seat. By using this
7345 call unprivileged clients can make changes to "backlight" and "leds"
7346 devices securely with strict requirements on session membership.
7347 Desktop environments may use this to generically make brightness
7348 changes to such devices without shipping private SUID binaries or
7349 udev rules for that purpose.
7351 * "udevadm info" gained a --wait-for-initialization switch to wait for
7352 a device to be initialized.
7354 * systemd-hibernate-resume-generator will now look for resumeflags= on
7355 the kernel command line, which is similar to rootflags= and may be
7356 used to configure device timeout for the hibernation device.
7358 * sd-event learnt a new API call sd_event_source_disable_unref() for
7359 disabling and unref'ing an event source in a single function. A
7360 related call sd_event_source_disable_unrefp() has been added for use
7361 with gcc's cleanup extension.
7363 * The sd-id128.h public API gained a new definition
7364 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
7367 * "busctl introspect" gained a new switch --xml-interface for dumping
7368 XML introspection data unmodified.
7370 * PID 1 may now show the unit name instead of the unit description
7371 string in its status output during boot. This may be configured in
7372 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
7373 kernel command line option systemd.status_unit_format=.
7375 * PID 1 now understands a new option KExecWatchdogSec= in
7376 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
7377 Previously watchdog functionality was only available for regular
7378 reboots. The new setting defaults to off, because we don't know in
7379 the general case if the watchdog will be reset after kexec (some
7380 drivers do reset it, but not all), and the new userspace might not be
7381 configured to handle the watchdog.
7383 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
7384 RebootWatchdogSec= to more clearly communicate what it is about. The
7385 old name is still accepted for compatibility.
7387 * The systemd.debug_shell kernel command line option now optionally
7388 takes a tty name to spawn the debug shell on, which allows a
7389 different tty to be selected than the built-in default.
7391 * Service units gained a new ExecCondition= setting which will run
7392 before ExecStartPre= and either continue execution of the unit (for
7393 clean exit codes), stop execution without marking the unit failed
7394 (for exit codes 1 through 254), or stop execution and fail the unit
7395 (for exit code 255 or abnormal termination).
7397 * A new service systemd-pstore.service has been added that pulls data
7398 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
7401 * timedatectl gained new verbs for configuring per-interface NTP
7402 service configuration for systemd-timesyncd.
7404 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
7405 2019. (You can set non-UTF-8 locales though, if you know their name.)
7407 * If variable assignments in sysctl.d/ files are prefixed with "-" any
7408 failures to apply them are now ignored.
7410 * systemd-random-seed.service now optionally credits entropy when
7411 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
7412 true for the service to enable this behaviour, but please consult the
7413 documentation first, since this comes with a couple of caveats.
7415 * systemd-random-seed.service is now a synchronization point for full
7416 initialization of the kernel's entropy pool. Services that require
7417 /dev/urandom to be correctly initialized should be ordered after this
7420 * The systemd-boot boot loader has been updated to optionally maintain
7421 a random seed file in the EFI System Partition (ESP). During the boot
7422 phase, this random seed is read and updated with a new seed
7423 cryptographically derived from it. Another derived seed is passed to
7424 the OS. The latter seed is then credited to the kernel's entropy pool
7425 very early during userspace initialization (from PID 1). This allows
7426 systems to boot up with a fully initialized kernel entropy pool from
7427 earliest boot on, and thus entirely removes all entropy pool
7428 initialization delays from systems using systemd-boot. Special care
7429 is taken to ensure different seeds are derived on system images
7430 replicated to multiple systems. "bootctl status" will show whether
7431 a seed was received from the boot loader.
7433 * bootctl gained two new verbs:
7435 - "bootctl random-seed" will generate the file in ESP and an EFI
7436 variable to allow a random seed to be passed to the OS as described
7439 - "bootctl is-installed" checks whether systemd-boot is currently
7442 * bootctl will warn if it detects that boot entries are misconfigured
7443 (for example if the kernel image was removed without purging the
7446 * A new document has been added describing systemd's use and support
7447 for the kernel's entropy pool subsystem:
7449 https://systemd.io/RANDOM_SEEDS
7451 * When the system is hibernated the swap device to write the
7452 hibernation image to is now automatically picked from all available
7453 swap devices, preferring the swap device with the highest configured
7454 priority over all others, and picking the device with the most free
7455 space if there are multiple devices with the highest priority.
7457 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
7458 option that permits selecting the timeout how long to wait for a
7459 device with an encryption key before asking for the password.
7461 * IOWeight= has learnt to properly set the IO weight when using the
7462 BFQ scheduler officially found in kernels 5.0+.
7464 * A new mailing list has been created for reporting of security issues:
7465 systemd-security@redhat.com. For mode details, see
7466 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
7468 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
7469 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
7470 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
7471 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
7472 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
7473 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
7474 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
7475 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
7476 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
7477 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
7478 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
7479 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
7480 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
7481 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
7482 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
7483 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
7484 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
7485 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
7486 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
7487 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
7488 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
7489 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
7490 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
7491 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
7492 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
7493 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
7494 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
7495 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
7496 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
7497 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
7499 – Camerino, 2019-09-03
7503 * In .link files, MACAddressPolicy=persistent (the default) is changed
7504 to cover more devices. For devices like bridges, tun, tap, bond, and
7505 similar interfaces that do not have other identifying information,
7506 the interface name is used as the basis for persistent seed for MAC
7507 and IPv4LL addresses. The way that devices that were handled
7508 previously is not changed, and this change is about covering more
7509 devices then previously by the "persistent" policy.
7511 MACAddressPolicy=random may be used to force randomized MACs and
7512 IPv4LL addresses for a device if desired.
7514 Hint: the log output from udev (at debug level) was enhanced to
7515 clarify what policy is followed and which attributes are used.
7516 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
7517 may be used to view this.
7519 Hint: if a bridge interface is created without any slaves, and gains
7520 a slave later, then now the bridge does not inherit slave's MAC.
7521 To inherit slave's MAC, for example, create the following file:
7523 # /etc/systemd/network/98-bridge-inherit-mac.link
7528 MACAddressPolicy=none
7531 * The .device units generated by systemd-fstab-generator and other
7532 generators do not automatically pull in the corresponding .mount unit
7533 as a Wants= dependency. This means that simply plugging in the device
7534 will not cause the mount unit to be started automatically. But please
7535 note that the mount unit may be started for other reasons, in
7536 particular if it is part of local-fs.target, and any unit which
7537 (transitively) depends on local-fs.target is started.
7539 * networkctl list/status/lldp now accept globbing wildcards for network
7540 interface names to match against all existing interfaces.
7542 * The $PIDFILE environment variable is set to point the absolute path
7543 configured with PIDFile= for processes of that service.
7545 * The fallback DNS server list was augmented with Cloudflare public DNS
7546 servers. Use `-Ddns-servers=` to set a different fallback.
7548 * A new special target usb-gadget.target will be started automatically
7549 when a USB Device Controller is detected (which means that the system
7550 is a USB peripheral).
7552 * A new unit setting CPUQuotaPeriodSec= assigns the time period
7553 relatively to which the CPU time quota specified by CPUQuota= is
7556 * A new unit setting ProtectHostname= may be used to prevent services
7557 from modifying hostname information (even if they otherwise would
7558 have privileges to do so).
7560 * A new unit setting NetworkNamespacePath= may be used to specify a
7561 namespace for service or socket units through a path referring to a
7562 Linux network namespace pseudo-file.
7564 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
7565 have an effect on .socket units: when used the listening socket is
7566 created within the configured network namespace instead of the host
7569 * ExecStart= command lines in unit files may now be prefixed with ':'
7570 in which case environment variable substitution is
7571 disabled. (Supported for the other ExecXYZ= settings, too.)
7573 * .timer units gained two new boolean settings OnClockChange= and
7574 OnTimezoneChange= which may be used to also trigger a unit when the
7575 system clock is changed or the local timezone is
7576 modified. systemd-run has been updated to make these options easily
7577 accessible from the command line for transient timers.
7579 * Two new conditions for units have been added: ConditionMemory= may be
7580 used to conditionalize a unit based on installed system
7581 RAM. ConditionCPUs= may be used to conditionalize a unit based on
7582 installed CPU cores.
7584 * The @default system call filter group understood by SystemCallFilter=
7585 has been updated to include the new rseq() system call introduced in
7588 * A new time-set.target has been added that indicates that the system
7589 time has been set from a local source (possibly imprecise). The
7590 existing time-sync.target is stronger and indicates that the time has
7591 been synchronized with a precise external source. Services where
7592 approximate time is sufficient should use the new target.
7594 * "systemctl start" (and related commands) learnt a new
7595 --show-transaction option. If specified brief information about all
7596 jobs queued because of the requested operation is shown.
7598 * systemd-networkd recognizes a new operation state 'enslaved', used
7599 (instead of 'degraded' or 'carrier') for interfaces which form a
7600 bridge, bond, or similar, and an new 'degraded-carrier' operational
7601 state used for the bond or bridge master interface when one of the
7602 enslaved devices is not operational.
7604 * .network files learnt the new IgnoreCarrierLoss= option for leaving
7605 networks configured even if the carrier is lost.
7607 * The RequiredForOnline= setting in .network files may now specify a
7608 minimum operational state required for the interface to be considered
7609 "online" by systemd-networkd-wait-online. Related to this
7610 systemd-networkd-wait-online gained a new option --operational-state=
7611 to configure the same, and its --interface= option was updated to
7612 optionally also take an operational state specific for an interface.
7614 * systemd-networkd-wait-online gained a new setting --any for waiting
7615 for only one of the requested interfaces instead of all of them.
7617 * systemd-networkd now implements L2TP tunnels.
7619 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
7620 may be used to cause autonomous and onlink prefixes received in IPv6
7621 Router Advertisements to be ignored.
7623 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
7624 file settings may be used to tweak bridge behaviour.
7626 * The new TripleSampling= option in .network files may be used to
7627 configure CAN triple sampling.
7629 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
7630 used to point to private or preshared key for a WireGuard interface.
7632 * /etc/crypttab now supports the same-cpu-crypt and
7633 submit-from-crypt-cpus options to tweak encryption work scheduling
7636 * systemd-tmpfiles will now take a BSD file lock before operating on a
7637 contents of directory. This may be used to temporarily exclude
7638 directories from aging by taking the same lock (useful for example
7639 when extracting a tarball into /tmp or /var/tmp as a privileged user,
7640 which might create files with really old timestamps, which
7641 nevertheless should not be deleted). For further details, see:
7643 https://systemd.io/TEMPORARY_DIRECTORIES
7645 * systemd-tmpfiles' h line type gained support for the
7646 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
7647 controlling project quota inheritance.
7649 * sd-boot and bootctl now implement support for an Extended Boot Loader
7650 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
7651 addition to the ESP partition mounted to /efi or /boot/efi.
7652 Configuration file fragments, kernels, initrds and other EFI images
7653 to boot will be loaded from both the ESP and XBOOTLDR partitions.
7654 The XBOOTLDR partition was previously described by the Boot Loader
7655 Specification, but implementation was missing in sd-boot. Support for
7656 this concept allows using the sd-boot boot loader in more
7657 conservative scenarios where the boot loader itself is placed in the
7658 ESP but the kernels to boot (and their metadata) in a separate
7661 * A system may now be booted with systemd.volatile=overlay on the
7662 kernel command line, which causes the root file system to be set up
7663 an overlayfs mount combining the root-only root directory with a
7664 writable tmpfs. In this setup, the underlying root device is not
7665 modified, and any changes are lost at reboot.
7667 * Similar, systemd-nspawn can now boot containers with a volatile
7668 overlayfs root with the new --volatile=overlay switch.
7670 * systemd-nspawn can now consume OCI runtime bundles using a new
7671 --oci-bundle= option. This implementation is fully usable, with most
7672 features in the specification implemented, but since this a lot of
7673 new code and functionality, this feature should most likely not
7674 be used in production yet.
7676 * systemd-nspawn now supports various options described by the OCI
7677 runtime specification on the command-line and in .nspawn files:
7678 --inaccessible=/Inaccessible= may be used to mask parts of the file
7679 system tree, --console=/--pipe may be used to configure how standard
7680 input, output, and error are set up.
7682 * busctl learned the `emit` verb to generate D-Bus signals.
7684 * systemd-analyze cat-config may be used to gather and display
7685 configuration spread over multiple files, for example system and user
7686 presets, tmpfiles.d, sysusers.d, udev rules, etc.
7688 * systemd-analyze calendar now takes an optional new parameter
7689 --iterations= which may be used to show a maximum number of iterations
7690 the specified expression will elapse next.
7692 * The sd-bus C API gained support for naming method parameters in the
7695 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
7696 the reboot() system call expects.
7698 * journalctl learnt a new --cursor-file= option that points to a file
7699 from which a cursor should be loaded in the beginning and to which
7700 the updated cursor should be stored at the end.
7702 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
7703 detected by systemd-detect-virt (and may also be used in
7704 ConditionVirtualization=).
7706 * The behaviour of systemd-logind may now be modified with environment
7707 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
7708 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
7709 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
7710 skip the relevant operation completely (when set to false), or to
7711 create a flag file in /run/systemd (when set to true), instead of
7712 actually commencing the real operation when requested. The presence
7713 of /run/systemd/reboot-to-firmware-setup,
7714 /run/systemd/reboot-to-boot-loader-menu, and
7715 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
7716 boot loader implementations to replace some steps logind performs
7717 during reboot with their own operations.
7719 * systemctl can be used to request a reboot into the boot loader menu
7720 or a specific boot loader entry with the new --boot-load-menu= and
7721 --boot-loader-entry= options to a reboot command. (This requires a
7722 boot loader that supports this, for example sd-boot.)
7724 * kernel-install will no longer unconditionally create the output
7725 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
7726 snippets, but will do only if the machine-specific parent directory
7727 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
7728 to create this parent directory during sd-boot installation.
7730 This makes it easier to use kernel-install with plugins which support
7731 a different layout of the bootloader partitions (for example grub2).
7733 * During package installation (with `ninja install`), we would create
7734 symlinks for getty@tty1.service, systemd-networkd.service,
7735 systemd-networkd.socket, systemd-resolved.service,
7736 remote-cryptsetup.target, remote-fs.target,
7737 systemd-networkd-wait-online.service, and systemd-timesyncd.service
7738 in /etc, as if `systemctl enable` was called for those units, to make
7739 the system usable immediately after installation. Now this is not
7740 done anymore, and instead calling `systemctl preset-all` is
7741 recommended after the first installation of systemd.
7743 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
7744 is built on seccomp. When turned on creation of SUID/SGID files is
7747 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
7748 implied if DynamicUser= is turned on for a service. This hardens
7749 these services, so that they neither can benefit from nor create
7750 SUID/SGID executables. This is a minor compatibility breakage, given
7751 that when DynamicUser= was first introduced SUID/SGID behaviour was
7752 unaffected. However, the security benefit of these two options is
7753 substantial, and the setting is still relatively new, hence we opted
7754 to make it mandatory for services with dynamic users.
7756 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
7757 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
7758 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
7759 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
7760 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
7761 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
7762 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
7763 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
7764 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
7765 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
7766 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
7767 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
7768 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
7769 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
7770 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
7771 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
7772 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
7773 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
7775 — Warsaw, 2019-04-11
7779 * The default locale can now be configured at compile time. Otherwise,
7780 a suitable default will be selected automatically (one of C.UTF-8,
7781 en_US.UTF-8, and C).
7783 * The version string shown by systemd and other tools now includes the
7784 git commit hash when built from git. An override may be specified
7785 during compilation, which is intended to be used by distributions to
7786 include the package release information.
7788 * systemd-cat can now filter standard input and standard error streams
7789 for different syslog priorities using the new --stderr-priority=
7792 * systemd-journald and systemd-journal-remote reject entries which
7793 contain too many fields (CVE-2018-16865) and set limits on the
7794 process' command line length (CVE-2018-16864).
7796 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
7799 * A new network device NamePolicy "keep" is implemented for link files,
7800 and used by default in 99-default.link (the fallback configuration
7801 provided by systemd). With this policy, if the network device name
7802 was already set by userspace, the device will not be renamed again.
7803 This matches the naming scheme that was implemented before
7804 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
7805 is also enabled by default, even if not specified. Effectively, this
7806 means that if naming-scheme >= 240 is specified, network devices will
7807 be renamed according to the configuration, even if they have been
7808 renamed already, if "keep" is not specified as the naming policy in
7809 the .link file. The 99-default.link file provided by systemd includes
7810 "keep" for backwards compatibility, but it is recommended for user
7811 installed .link files to *not* include it.
7813 The "kernel" policy, which keeps kernel names declared to be
7814 "persistent", now works again as documented.
7816 * kernel-install script now optionally takes the paths to one or more
7817 initrd files, and passes them to all plugins.
7819 * The mincore() system call has been dropped from the @system-service
7820 system call filter group, as it is pretty exotic and may potentially
7821 used for side-channel attacks.
7823 * -fPIE is dropped from compiler and linker options. Please specify
7824 -Db_pie=true option to meson to build position-independent
7825 executables. Note that the meson option is supported since meson-0.49.
7827 * The fs.protected_regular and fs.protected_fifos sysctls, which were
7828 added in Linux 4.19 to make some data spoofing attacks harder, are
7829 now enabled by default. While this will hopefully improve the
7830 security of most installations, it is technically a backwards
7831 incompatible change; to disable these sysctls again, place the
7832 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
7834 fs.protected_regular = 0
7835 fs.protected_fifos = 0
7837 Note that the similar hardlink and symlink protection has been
7838 enabled since v199, and may be disabled likewise.
7840 * The files read from the EnvironmentFile= setting in unit files now
7841 parse backslashes inside quotes literally, matching the behaviour of
7844 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
7845 now automatically become NOPs when run in a chroot() environment.
7847 * The tmpfiles.d/ "C" line type will now copy directory trees not only
7848 when the destination is so far missing, but also if it already exists
7849 as a directory and is empty. This is useful to cater for systems
7850 where directory trees are put together from multiple separate mount
7851 points but otherwise empty.
7853 * A new function sd_bus_close_unref() (and the associated
7854 sd_bus_close_unrefp()) has been added to libsystemd, that combines
7855 sd_bus_close() and sd_bus_unref() in one.
7857 * udevadm control learnt a new option for --ping for testing whether a
7858 systemd-udevd instance is running and reacting.
7860 * udevadm trigger learnt a new option for --wait-daemon for waiting
7861 systemd-udevd daemon to be initialized.
7863 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
7864 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
7865 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
7866 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
7867 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
7868 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
7869 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
7870 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
7871 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
7872 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
7873 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
7874 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
7875 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
7876 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
7877 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
7878 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
7879 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
7881 — Berlin, 2019-02-14
7885 * NoNewPrivileges=yes has been set for all long-running services
7886 implemented by systemd. Previously, this was problematic due to
7887 SELinux (as this would also prohibit the transition from PID1's label
7888 to the service's label). This restriction has since been lifted, but
7889 an SELinux policy update is required.
7890 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
7892 * DynamicUser=yes is dropped from systemd-networkd.service,
7893 systemd-resolved.service and systemd-timesyncd.service, which was
7894 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
7895 and since v236 for systemd-timesyncd.service. The users and groups
7896 systemd-network, systemd-resolve and systemd-timesync are created
7897 by systemd-sysusers again. Distributors or system administrators
7898 may need to create these users and groups if they not exist (or need
7899 to re-enable DynamicUser= for those units) while upgrading systemd.
7900 Also, the clock file for systemd-timesyncd may need to move from
7901 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
7903 * When unit files are loaded from disk, previously systemd would
7904 sometimes (depending on the unit loading order) load units from the
7905 target path of symlinks in .wants/ or .requires/ directories of other
7906 units. This meant that unit could be loaded from different paths
7907 depending on whether the unit was requested explicitly or as a
7908 dependency of another unit, not honouring the priority of directories
7909 in search path. It also meant that it was possible to successfully
7910 load and start units which are not found in the unit search path, as
7911 long as they were requested as a dependency and linked to from
7912 .wants/ or .requires/. The target paths of those symlinks are not
7913 used for loading units anymore and the unit file must be found in
7916 * A new service type has been added: Type=exec. It's very similar to
7917 Type=simple but ensures the service manager will wait for both fork()
7918 and execve() of the main service binary to complete before proceeding
7919 with follow-up units. This is primarily useful so that the manager
7920 propagates any errors in the preparation phase of service execution
7921 back to the job that requested the unit to be started. For example,
7922 consider a service that has ExecStart= set to a file system binary
7923 that doesn't exist. With Type=simple starting the unit would be
7924 considered instantly successful, as only fork() has to complete
7925 successfully and the manager does not wait for execve(), and hence
7926 its failure is seen "too late". With the new Type=exec service type
7927 starting the unit will fail, as the manager will wait for the
7928 execve() and notice its failure, which is then propagated back to the
7931 NOTE: with the next release 241 of systemd we intend to change the
7932 systemd-run tool to default to Type=exec for transient services
7933 started by it. This should be mostly safe, but in specific corner
7934 cases might result in problems, as the systemd-run tool will then
7935 block on NSS calls (such as user name look-ups due to User=) done
7936 between the fork() and execve(), which under specific circumstances
7937 might cause problems. It is recommended to specify "-p Type=simple"
7938 explicitly in the few cases where this applies. For regular,
7939 non-transient services (i.e. those defined with unit files on disk)
7940 we will continue to default to Type=simple.
7942 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
7943 userspace processes is set to 1024 (soft) and 4096
7944 (hard). Previously, systemd passed this on unmodified to all
7945 processes it forked off. With this systemd release the hard limit
7946 systemd passes on is increased to 512K, overriding the kernel's
7947 defaults and substantially increasing the number of simultaneous file
7948 descriptors unprivileged userspace processes can allocate. Note that
7949 the soft limit remains at 1024 for compatibility reasons: the
7950 traditional UNIX select() call cannot deal with file descriptors >=
7951 1024 and increasing the soft limit globally might thus result in
7952 programs unexpectedly allocating a high file descriptor and thus
7953 failing abnormally when attempting to use it with select() (of
7954 course, programs shouldn't use select() anymore, and prefer
7955 poll()/epoll, but the call unfortunately remains undeservedly popular
7956 at this time). This change reflects the fact that file descriptor
7957 handling in the Linux kernel has been optimized in more recent
7958 kernels and allocating large numbers of them should be much cheaper
7959 both in memory and in performance than it used to be. Programs that
7960 want to take benefit of the increased limit have to "opt-in" into
7961 high file descriptors explicitly by raising their soft limit. Of
7962 course, when they do that they must acknowledge that they cannot use
7963 select() anymore (and neither can any shared library they use — or
7964 any shared library used by any shared library they use and so on).
7965 Which default hard limit is most appropriate is of course hard to
7966 decide. However, given reports that ~300K file descriptors are used
7967 in real-life applications we believe 512K is sufficiently high as new
7968 default for now. Note that there are also reports that using very
7969 high hard limits (e.g. 1G) is problematic: some software allocates
7970 large arrays with one element for each potential file descriptor
7971 (Java, …) — a high hard limit thus triggers excessively large memory
7972 allocations in these applications. Hopefully, the new default of 512K
7973 is a good middle ground: higher than what real-life applications
7974 currently need, and low enough for avoid triggering excessively large
7975 allocations in problematic software. (And yes, somebody should fix
7978 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
7979 to the highest possible values, as separate accounting of file
7980 descriptors is no longer necessary, as memcg tracks them correctly as
7981 part of the memory accounting anyway. Thus, from the four limits on
7982 file descriptors currently enforced (fs.file-max, fs.nr_open,
7983 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
7984 and keep only the latter two. A set of build-time options
7985 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
7986 has been added to revert this change in behaviour, which might be
7987 an option for systems that turn off memcg in the kernel.
7989 * When no /etc/locale.conf file exists (and hence no locale settings
7990 are in place), systemd will now use the "C.UTF-8" locale by default,
7991 and set LANG= to it. This locale is supported by various
7992 distributions including Fedora, with clear indications that upstream
7993 glibc is going to make it available too. This locale enables UTF-8
7994 mode by default, which appears appropriate for 2018.
7996 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
7997 default. This effectively switches the RFC3704 Reverse Path filtering
7998 from Strict mode to Loose mode. This is more appropriate for hosts
7999 that have multiple links with routes to the same networks (e.g.
8000 a client with a Wi-Fi and Ethernet both connected to the internet).
8002 Consult the kernel documentation for details on this sysctl:
8003 https://docs.kernel.org/networking/ip-sysctl.html
8005 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
8008 * CPUAccounting=yes no longer enables the CPU controller when using
8009 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
8010 statistics are now provided independently from the CPU controller.
8012 * Support for disabling a particular cgroup controller within a sub-tree
8013 has been added through the DisableControllers= directive.
8015 * cgroup_no_v1=all on the kernel command line now also implies
8016 using the unified cgroup hierarchy, unless one explicitly passes
8017 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
8019 * The new "MemoryMin=" unit file property may now be used to set the
8020 memory usage protection limit of processes invoked by the unit. This
8021 controls the cgroup v2 memory.min attribute. Similarly, the new
8022 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
8023 cgroup v2 io.latency cgroup property for configuring per-service I/O
8026 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
8027 to the cgroup v1 "devices" cgroup controller.
8029 * systemd-escape now is able to combine --unescape with --template. It
8030 also learnt a new option --instance for extracting and unescaping the
8031 instance part of a unit name.
8033 * sd-bus now provides the sd_bus_message_readv() which is similar to
8034 sd_bus_message_read() but takes a va_list object. The pair
8035 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
8036 has been added for configuring the default method call timeout to
8037 use. sd_bus_error_move() may be used to efficiently move the contents
8038 from one sd_bus_error structure to another, invalidating the
8039 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
8040 be used to control whether a bus connection object is automatically
8041 flushed when an sd-event loop is exited.
8043 * When processing classic BSD syslog log messages, journald will now
8044 save the original time-stamp string supplied in the new
8045 SYSLOG_TIMESTAMP= journal field. This permits consumers to
8046 reconstruct the original BSD syslog message more correctly.
8048 * StandardOutput=/StandardError= in service files gained support for
8049 new "append:…" parameters, for connecting STDOUT/STDERR of a service
8050 to a file, and appending to it.
8052 * The signal to use as last step of killing of unit processes is now
8053 configurable. Previously it was hard-coded to SIGKILL, which may now
8054 be overridden with the new KillSignal= setting. Note that this is the
8055 signal used when regular termination (i.e. SIGTERM) does not suffice.
8056 Similarly, the signal used when aborting a program in case of a
8057 watchdog timeout may now be configured too (WatchdogSignal=).
8059 * The XDG_SESSION_DESKTOP environment variable may now be configured in
8060 the pam_systemd argument line, using the new desktop= switch. This is
8061 useful to initialize it properly from a display manager without
8062 having to touch C code.
8064 * Most configuration options that previously accepted percentage values
8065 now also accept permille values with the '‰' suffix (instead of '%').
8067 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
8070 * systemd-resolved's configuration file resolved.conf gained a new
8071 option ReadEtcHosts= which may be used to turn off processing and
8072 honoring /etc/hosts entries.
8074 * The "--wait" switch may now be passed to "systemctl
8075 is-system-running", in which case the tool will synchronously wait
8076 until the system finished start-up.
8078 * hostnamed gained a new bus call to determine the DMI product UUID.
8080 * On x86-64 systemd will now prefer using the RDRAND processor
8081 instruction over /dev/urandom whenever it requires randomness that
8082 neither has to be crypto-grade nor should be reproducible. This
8083 should substantially reduce the amount of entropy systemd requests
8084 from the kernel during initialization on such systems, though not
8085 reduce it to zero. (Why not zero? systemd still needs to allocate
8086 UUIDs and such uniquely, which require high-quality randomness.)
8088 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
8089 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
8090 for forcing the "Other Information" bit in IPv6 RA messages. The
8091 bonding logic gained four new options AdActorSystemPriority=,
8092 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
8093 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
8094 shuffling of flows. The tunnel logic gained a new
8095 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
8096 Deployment. The policy rule logic gained four new options IPProtocol=,
8097 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
8098 support for the MulticastToUnicast= option. networkd also gained
8099 support for configuring static IPv4 ARP or IPv6 neighbor entries.
8101 * .preset files (as read by 'systemctl preset') may now be used to
8102 instantiate services.
8104 * /etc/crypttab now understands the sector-size= option to configure
8105 the sector size for an encrypted partition.
8107 * Key material for encrypted disks may now be placed on a formatted
8108 medium, and referenced from /etc/crypttab by the UUID of the file
8109 system, followed by "=" suffixed by the path to the key file.
8111 * The "collect" udev component has been removed without replacement, as
8112 it is neither used nor maintained.
8114 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
8115 LogsDirectory=, ConfigurationDirectory= settings are used in a
8116 service the executed processes will now receive a set of environment
8117 variables containing the full paths of these directories.
8118 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
8119 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
8120 are used. Note that these options may be used multiple times per
8121 service in which case the resulting paths will be concatenated and
8122 separated by colons.
8124 * Predictable interface naming has been extended to cover InfiniBand
8125 NICs. They will be exposed with an "ib" prefix.
8127 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
8128 which case the respective line failing is ignored.
8130 * .link files may now be used to configure the equivalent to the
8131 "ethtool advertise" commands.
8133 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
8134 alternative to libudev.h. Previously, the latter was just an internal
8135 wrapper around the former, but now these two APIs are exposed
8138 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
8139 which calculates an app-specific boot ID similar to how
8140 sd_id128_get_machine_app_specific() generates an app-specific machine
8143 * A new tool systemd-id128 has been added that can be used to determine
8144 and generate various 128-bit IDs.
8146 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
8149 * systemd-hibernate-resume-generator will now honor the "noresume"
8150 kernel command line option, in which case it will bypass resuming
8151 from any hibernated image.
8153 * The systemd-sleep.conf configuration file gained new options
8154 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
8155 AllowHybridSleep= for prohibiting specific sleep modes even if the
8156 kernel exports them.
8158 * portablectl is now officially supported and has thus moved to
8161 * bootctl learnt the two new commands "set-default" and "set-oneshot"
8162 for setting the default boot loader item to boot to (either
8163 persistently or only for the next boot). This is currently only
8164 compatible with sd-boot, but may be implemented on other boot loaders
8165 too, that follow the boot loader interface. The updated interface is
8166 now documented here:
8168 https://systemd.io/BOOT_LOADER_INTERFACE
8170 * A new kernel command line option systemd.early_core_pattern= is now
8171 understood which may be used to influence the core_pattern PID 1
8172 installs during early boot.
8174 * busctl learnt two new options -j and --json= for outputting method
8175 call replies, properties and monitoring output in JSON.
8177 * journalctl's JSON output now supports simple ANSI coloring as well as
8178 a new "json-seq" mode for generating RFC7464 output.
8180 * Unit files now support the %g/%G specifiers that resolve to the UNIX
8181 group/GID of the service manager runs as, similar to the existing
8182 %u/%U specifiers that resolve to the UNIX user/UID.
8184 * systemd-logind learnt a new global configuration option
8185 UserStopDelaySec= that may be set in logind.conf. It specifies how
8186 long the systemd --user instance shall remain started after a user
8187 logs out. This is useful to speed up repetitive re-connections of the
8188 same user, as it means the user's service manager doesn't have to be
8189 stopped/restarted on each iteration, but can be reused between
8190 subsequent options. This setting defaults to 10s. systemd-logind also
8191 exports two new properties on its Manager D-Bus objects indicating
8192 whether the system's lid is currently closed, and whether the system
8195 * systemd gained support for a generic boot counting logic, which
8196 generically permits automatic reverting to older boot loader entries
8197 if newer updated ones don't work. The boot loader side is implemented
8198 in sd-boot, but is kept open for other boot loaders too. For details
8201 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
8203 * The SuccessAction=/FailureAction= unit file settings now learnt two
8204 new parameters: "exit" and "exit-force", which result in immediate
8205 exiting of the service manager, and are only useful in systemd --user
8206 and container environments.
8208 * Unit files gained support for a pair of options
8209 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
8210 exit status to use as service manager exit status when
8211 SuccessAction=/FailureAction= is set to exit or exit-force.
8213 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
8214 options may now be used to configure the log rate limiting applied by
8215 journald per-service.
8217 * systemd-analyze gained a new verb "timespan" for parsing and
8218 normalizing time span values (i.e. strings like "5min 7s 8us").
8220 * systemd-analyze also gained a new verb "security" for analyzing the
8221 security and sand-boxing settings of services in order to determine an
8222 "exposure level" for them, indicating whether a service would benefit
8223 from more sand-boxing options turned on for them.
8225 * "systemd-analyze syscall-filter" will now also show system calls
8226 supported by the local kernel but not included in any of the defined
8229 * .nspawn files now understand the Ephemeral= setting, matching the
8230 --ephemeral command line switch.
8232 * sd-event gained the new APIs sd_event_source_get_floating() and
8233 sd_event_source_set_floating() for controlling whether a specific
8234 event source is "floating", i.e. destroyed along with the even loop
8237 * Unit objects on D-Bus gained a new "Refs" property that lists all
8238 clients that currently have a reference on the unit (to ensure it is
8241 * The JoinControllers= option in system.conf is no longer supported, as
8242 it didn't work correctly, is hard to support properly, is legacy (as
8243 the concept only exists on cgroup v1) and apparently wasn't used.
8245 * Journal messages that are generated whenever a unit enters the failed
8246 state are now tagged with a unique MESSAGE_ID. Similarly, messages
8247 generated whenever a service process exits are now made recognizable,
8248 too. A tagged message is also emitted whenever a unit enters the
8249 "dead" state on success.
8251 * systemd-run gained a new switch --working-directory= for configuring
8252 the working directory of the service to start. A shortcut -d is
8253 equivalent, setting the working directory of the service to the
8254 current working directory of the invoking program. The new --shell
8255 (or just -S) option has been added for invoking the $SHELL of the
8256 caller as a service, and implies --pty --same-dir --wait --collect
8257 --service-type=exec. Or in other words, "systemd-run -S" is now the
8258 quickest way to quickly get an interactive in a fully clean and
8259 well-defined system service context.
8261 * machinectl gained a new verb "import-fs" for importing an OS tree
8262 from a directory. Moreover, when a directory or tarball is imported
8263 and single top-level directory found with the OS itself below the OS
8264 tree is automatically mangled and moved one level up.
8266 * systemd-importd will no longer set up an implicit btrfs loop-back
8267 file system on /var/lib/machines. If one is already set up, it will
8268 continue to be used.
8270 * A new generator "systemd-run-generator" has been added. It will
8271 synthesize a unit from one or more program command lines included in
8272 the kernel command line. This is very useful in container managers
8275 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
8277 This will run "systemd-nspawn" on an image, invoke the specified
8278 command line and immediately shut down the container again, returning
8279 the command line's exit code.
8281 * The block device locking logic is now documented:
8283 https://systemd.io/BLOCK_DEVICE_LOCKING
8285 * loginctl and machinectl now optionally output the various tables in
8286 JSON using the --output= switch. It is our intention to add similar
8287 support to systemctl and all other commands.
8289 * udevadm's query and trigger verb now optionally take a .device unit
8292 * systemd-udevd's network naming logic now understands a new
8293 net.naming-scheme= kernel command line switch, which may be used to
8294 pick a specific version of the naming scheme. This helps stabilizing
8295 interface names even as systemd/udev are updated and the naming logic
8298 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
8299 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
8300 initialize one to all 0xFF.
8302 * After loading the SELinux policy systemd will now recursively relabel
8303 all files and directories listed in
8304 /run/systemd/relabel-extra.d/*.relabel (which should be simple
8305 newline separated lists of paths) in addition to the ones it already
8306 implicitly relabels in /run, /dev and /sys. After the relabelling is
8307 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
8308 removed. This is useful to permit initrds (i.e. code running before
8309 the SELinux policy is in effect) to generate files in the host
8310 filesystem safely and ensure that the correct label is applied during
8311 the transition to the host OS.
8313 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
8314 mknod() handling in user namespaces. Previously mknod() would always
8315 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
8316 but device nodes generated that way cannot be opened, and attempts to
8317 open them result in EPERM. This breaks the "graceful fallback" logic
8318 in systemd's PrivateDevices= sand-boxing option. This option is
8319 implemented defensively, so that when systemd detects it runs in a
8320 restricted environment (such as a user namespace, or an environment
8321 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
8322 where device nodes cannot be created the effect of PrivateDevices= is
8323 bypassed (following the logic that 2nd-level sand-boxing is not
8324 essential if the system systemd runs in is itself already sand-boxed
8325 as a whole). This logic breaks with 4.18 in container managers where
8326 user namespacing is used: suddenly PrivateDevices= succeeds setting
8327 up a private /dev/ file system containing devices nodes — but when
8328 these are opened they don't work.
8330 At this point it is recommended that container managers utilizing
8331 user namespaces that intend to run systemd in the payload explicitly
8332 block mknod() with seccomp or similar, so that the graceful fallback
8335 We are very sorry for the breakage and the requirement to change
8336 container configurations for newer kernels. It's purely caused by an
8337 incompatible kernel change. The relevant kernel developers have been
8338 notified about this userspace breakage quickly, but they chose to
8341 * PermissionsStartOnly= setting is deprecated (but is still supported
8342 for backwards compatibility). The same functionality is provided by
8343 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
8346 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
8347 pam_systemd anymore.
8349 * The naming scheme for network devices was changed to always rename
8350 devices, even if they were already renamed by userspace. The "kernel"
8351 policy was changed to only apply as a fallback, if no other naming
8354 * The requirements to build systemd is bumped to meson-0.46 and
8357 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
8358 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
8359 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
8360 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
8361 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
8362 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
8363 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
8364 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
8365 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
8366 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
8367 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
8368 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
8369 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
8370 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
8371 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
8372 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
8373 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
8374 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
8375 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
8376 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
8377 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
8378 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
8379 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
8380 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
8381 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
8382 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
8383 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
8384 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
8385 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
8386 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
8387 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
8388 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
8389 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
8390 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
8391 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
8392 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
8393 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
8394 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
8395 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
8396 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
8397 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
8398 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
8399 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
8400 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
8401 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
8403 — Warsaw, 2018-12-21
8407 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
8408 builtin will name network interfaces differently than in previous
8409 versions for virtual network interfaces created with SR-IOV and NPAR
8410 and for devices where the PCI network controller device does not have
8411 a slot number associated.
8413 SR-IOV virtual devices are now named based on the name of the parent
8414 interface, with a suffix of "v<N>", where <N> is the virtual device
8415 number. Previously those virtual devices were named as if completely
8418 The ninth and later NPAR virtual devices will be named following the
8419 scheme used for the first eight NPAR partitions. Previously those
8420 devices were not renamed and the kernel default (eth<n>) was used.
8422 "net_id" will also generate names for PCI devices where the PCI
8423 network controller device does not have an associated slot number
8424 itself, but one of its parents does. Previously those devices were
8425 not renamed and the kernel default (eth<n>) was used.
8427 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
8428 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
8429 the unit. So, it is expected that the default behavior of
8430 systemd-logind is not changed. However, if distribution packagers or
8431 administrators disabled or modified IPAddressDeny= setting by a
8432 drop-in config file, then it may be necessary to update the file to
8433 re-enable AF_INET and AF_INET6 to support network user name services,
8436 * When the RestrictNamespaces= unit property is specified multiple
8437 times, then the specified types are merged now. Previously, only the
8438 last assignment was used. So, if distribution packagers or
8439 administrators modified the setting by a drop-in config file, then it
8440 may be necessary to update the file.
8442 * When OnFailure= is used in combination with Restart= on a service
8443 unit, then the specified units will no longer be triggered on
8444 failures that result in restarting. Previously, the specified units
8445 would be activated each time the unit failed, even when the unit was
8446 going to be restarted automatically. This behaviour contradicted the
8447 documentation. With this release the code is adjusted to match the
8450 * systemd-tmpfiles will now print a notice whenever it encounters
8451 tmpfiles.d/ lines referencing the /var/run/ directory. It will
8452 recommend reworking them to use the /run/ directory instead (for
8453 which /var/run/ is simply a symlinked compatibility alias). This way
8454 systemd-tmpfiles can properly detect line conflicts and merge lines
8455 referencing the same file by two paths, without having to access
8458 * systemctl disable/unmask/preset/preset-all cannot be used with
8459 --runtime. Previously this was allowed, but resulted in unintuitive
8460 behaviour that wasn't useful. systemctl disable/unmask will now undo
8461 both runtime and persistent enablement/masking, i.e. it will remove
8462 any relevant symlinks both in /run and /etc.
8464 * Note that all long-running system services shipped with systemd will
8465 now default to a system call allow list (rather than a deny list, as
8466 before). In particular, systemd-udevd will now enforce one too. For
8467 most cases this should be safe, however downstream distributions
8468 which disabled sandboxing of systemd-udevd (specifically the
8469 MountFlags= setting), might want to disable this security feature
8470 too, as the default allow-listing will prohibit all mount, swap,
8471 reboot and clock changing operations from udev rules.
8473 * sd-boot acquired new loader configuration settings to optionally turn
8474 off Windows and MacOS boot partition discovery as well as
8475 reboot-into-firmware menu items. It is also able to pick a better
8476 screen resolution for HiDPI systems, and now provides loader
8477 configuration settings to change the resolution explicitly.
8479 * systemd-resolved now supports DNS-over-TLS. It's still
8480 turned off by default, use DNSOverTLS=opportunistic to turn it on in
8481 resolved.conf. We intend to make this the default as soon as couple
8482 of additional techniques for optimizing the initial latency caused by
8483 establishing a TLS/TCP connection are implemented.
8485 * systemd-resolved.service and systemd-networkd.service now set
8486 DynamicUser=yes. The users systemd-resolve and systemd-network are
8487 not created by systemd-sysusers anymore.
8489 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
8490 that embed a network facing module into any process using getpwuid()
8491 or related call: the dynamic allocation of the user ID for
8492 systemd-resolved.service means the service manager has to check NSS
8493 if the user name is already taken when forking off the service. Since
8494 the user in the common case won't be defined in /etc/passwd the
8495 lookup is likely to trigger nss-ldap which in turn might use NSS to
8496 ask systemd-resolved for hostname lookups. This will hence result in
8497 a deadlock: a user name lookup in order to start
8498 systemd-resolved.service will result in a hostname lookup for which
8499 systemd-resolved.service needs to be started already. There are
8500 multiple ways to work around this problem: pre-allocate the
8501 "systemd-resolve" user on such systems, so that nss-ldap won't be
8502 triggered; or use a different NSS package that doesn't do networking
8503 in-process but provides a local asynchronous name cache; or configure
8504 the NSS package to avoid lookups for UIDs in the range `pkg-config
8505 systemd --variable=dynamicuidmin` … `pkg-config systemd
8506 --variable=dynamicuidmax`, so that it does not consider itself
8507 authoritative for the same UID range systemd allocates dynamic users
8510 * The systemd-resolve tool has been renamed to resolvectl (it also
8511 remains available under the old name, for compatibility), and its
8512 interface is now verb-based, similar in style to the other <xyz>ctl
8513 tools, such as systemctl or loginctl.
8515 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
8516 compatibility. It may be symlinked under the 'resolvconf' name, in
8517 which case it will take arguments and input compatible with the
8518 Debian and FreeBSD resolvconf tool.
8520 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
8521 where the system initially suspends, and after a timeout resumes and
8524 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
8525 set the client will only send a DUID as client identifier. (EDIT: the
8526 option was broken, and was dropped in v255.)
8528 * The nss-systemd glibc NSS module will now enumerate dynamic users and
8529 groups in effect. Previously, it could resolve UIDs/GIDs to user
8530 names/groups and vice versa, but did not support enumeration.
8532 * journald's Compress= configuration setting now optionally accepts a
8533 byte threshold value. All journal objects larger than this threshold
8534 will be compressed, smaller ones will not. Previously this threshold
8535 was not configurable and set to 512.
8537 * A new system.conf setting NoNewPrivileges= is now available which may
8538 be used to turn off acquisition of new privileges system-wide
8539 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
8540 for all its children). Note that turning this option on means setuid
8541 binaries and file system capabilities lose their special powers.
8542 While turning on this option is a big step towards a more secure
8543 system, doing so is likely to break numerous pre-existing UNIX tools,
8544 in particular su and sudo.
8546 * A new service systemd-time-sync-wait.service has been added. If
8547 enabled it will delay the time-sync.target unit at boot until time
8548 synchronization has been received from the network. This
8549 functionality is useful on systems lacking a local RTC or where it is
8550 acceptable that the boot process shall be delayed by external network
8553 * When hibernating, systemd will now inform the kernel of the image
8554 write offset, on kernels new enough to support this. This means swap
8555 files should work for hibernation now.
8557 * When loading unit files, systemd will now look for drop-in unit files
8558 extensions in additional places. Previously, for a unit file name
8559 "foo-bar-baz.service" it would look for dropin files in
8560 "foo-bar-baz.service.d/*.conf". Now, it will also look in
8561 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
8562 service name truncated after all inner dashes. This scheme allows
8563 writing drop-ins easily that apply to a whole set of unit files at
8564 once. It's particularly useful for mount and slice units (as their
8565 naming is prefix based), but is also useful for service and other
8566 units, for packages that install multiple unit files at once,
8567 following a strict naming regime of beginning the unit file name with
8568 the package's name. Two new specifiers are now supported in unit
8569 files to match this: %j and %J are replaced by the part of the unit
8570 name following the last dash.
8572 * Unit files and other configuration files that support specifier
8573 expansion now understand another three new specifiers: %T and %V will
8574 resolve to /tmp and /var/tmp respectively, or whatever temporary
8575 directory has been set for the calling user. %E will expand to either
8576 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
8578 * The ExecStart= lines of unit files are no longer required to
8579 reference absolute paths. If non-absolute paths are specified the
8580 specified binary name is searched within the service manager's
8581 built-in $PATH, which may be queried with 'systemd-path
8582 search-binaries-default'. It's generally recommended to continue to
8583 use absolute paths for all binaries specified in unit files.
8585 * Units gained a new load state "bad-setting", which is used when a
8586 unit file was loaded, but contained fatal errors which prevent it
8587 from being started (for example, a service unit has been defined
8588 lacking both ExecStart= and ExecStop= lines).
8590 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
8591 support alternative debuggers, for example lldb. The old name
8592 continues to be available however, for compatibility reasons. Use the
8593 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
8594 to pick an alternative debugger instead of the default gdb.
8596 * systemctl and the other tools will now output escape sequences that
8597 generate proper clickable hyperlinks in various terminal emulators
8598 where useful (for example, in the "systemctl status" output you can
8599 now click on the unit file name to quickly open it in the
8600 editor/viewer of your choice). Note that not all terminal emulators
8601 support this functionality yet, but many do. Unfortunately, the
8602 "less" pager doesn't support this yet, hence this functionality is
8603 currently automatically turned off when a pager is started (which
8604 happens quite often due to auto-paging). We hope to remove this
8605 limitation as soon as "less" learns these escape sequences. This new
8606 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
8607 environment variable. For details on these escape sequences see:
8608 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
8610 * networkd's .network files now support a new IPv6MTUBytes= option for
8611 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
8612 option in the [Route] section to configure the MTU to use for
8613 specific routes. It also gained support for configuration of the DHCP
8614 "UserClass" option through the new UserClass= setting. It gained
8615 three new options in the new [CAN] section for configuring CAN
8616 networks. The MULTICAST and ALLMULTI interface flags may now be
8617 controlled explicitly with the new Multicast= and AllMulticast=
8620 * networkd will now automatically make use of the kernel's route
8621 expiration feature, if it is available.
8623 * udevd's .link files now support setting the number of receive and
8624 transmit channels, using the RxChannels=, TxChannels=,
8625 OtherChannels=, CombinedChannels= settings.
8627 * Support for UDPSegmentationOffload= has been removed, given its
8628 limited support in hardware, and waning software support.
8630 * networkd's .netdev files now support creating "netdevsim" interfaces.
8632 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
8633 to query the unit belonging to a specific kernel control group.
8635 * systemd-analyze gained a new verb "cat-config", which may be used to
8636 dump the contents of any configuration file, with all its matching
8637 drop-in files added in, and honouring the usual search and masking
8638 logic applied to systemd configuration files. For example use
8639 "systemd-analyze cat-config systemd/system.conf" to get the complete
8640 system configuration file of systemd how it would be loaded by PID 1
8641 itself. Similar to this, various tools such as systemd-tmpfiles or
8642 systemd-sysusers, gained a new option "--cat-config", which does the
8643 corresponding operation for their own configuration settings. For
8644 example, "systemd-tmpfiles --cat-config" will now output the full
8645 list of tmpfiles.d/ lines in place.
8647 * timedatectl gained three new verbs: "show" shows bus properties of
8648 systemd-timedated, "timesync-status" shows the current NTP
8649 synchronization state of systemd-timesyncd, and "show-timesync"
8650 shows bus properties of systemd-timesyncd.
8652 * systemd-timesyncd gained a bus interface on which it exposes details
8655 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
8656 understood by systemd-timedated. It takes a colon-separated list of
8657 unit names of NTP client services. The list is used by
8658 "timedatectl set-ntp".
8660 * systemd-nspawn gained a new --rlimit= switch for setting initial
8661 resource limits for the container payload. There's a new switch
8662 --hostname= to explicitly override the container's hostname. A new
8663 --no-new-privileges= switch may be used to control the
8664 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
8665 --oom-score-adjust= switch controls the OOM scoring adjustment value
8666 for the payload. The new --cpu-affinity= switch controls the CPU
8667 affinity of the container payload. The new --resolv-conf= switch
8668 allows more detailed control of /etc/resolv.conf handling of the
8669 container. Similarly, the new --timezone= switch allows more detailed
8670 control of /etc/localtime handling of the container.
8672 * systemd-detect-virt gained a new --list switch, which will print a
8673 list of all currently known VM and container environments.
8675 * Support for "Portable Services" has been added, see
8676 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
8677 experimental, but this is expected to change soon. Reflecting this
8678 experimental state, the "portablectl" binary is not installed into
8679 /usr/bin yet. The binary has to be called with the full path
8680 /usr/lib/systemd/portablectl instead.
8682 * journalctl's and systemctl's -o switch now knows a new log output
8683 mode "with-unit". The output it generates is very similar to the
8684 regular "short" mode, but displays the unit name instead of the
8685 syslog tag for each log line. Also, the date is shown with timezone
8686 information. This mode is probably more useful than the classic
8687 "short" output mode for most purposes, except where pixel-perfect
8688 compatibility with classic /var/log/messages formatting is required.
8690 * A new --dump-bus-properties switch has been added to the systemd
8691 binary, which may be used to dump all supported D-Bus properties.
8692 (Options which are still supported, but are deprecated, are *not*
8695 * sd-bus gained a set of new calls:
8696 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
8697 enable/disable the "floating" state of a bus slot object,
8698 i.e. whether the slot object pins the bus it is allocated for into
8699 memory or if the bus slot object gets disconnected when the bus goes
8700 away. sd_bus_open_with_description(),
8701 sd_bus_open_user_with_description(),
8702 sd_bus_open_system_with_description() may be used to allocate bus
8703 objects and set their description string already during allocation.
8705 * sd-event gained support for watching inotify events from the event
8706 loop, in an efficient way, sharing inotify handles between multiple
8707 users. For this a new function sd_event_add_inotify() has been added.
8709 * sd-event and sd-bus gained support for calling special user-supplied
8710 destructor functions for userdata pointers associated with
8711 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
8712 functions sd_bus_slot_set_destroy_callback,
8713 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
8714 sd_bus_track_get_destroy_callback,
8715 sd_event_source_set_destroy_callback,
8716 sd_event_source_get_destroy_callback have been added.
8718 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
8720 * PID 1 will now automatically reschedule .timer units whenever the
8721 local timezone changes. (They previously got rescheduled
8722 automatically when the system clock changed.)
8724 * New documentation has been added to document cgroups delegation,
8725 portable services and the various code quality tools we have set up:
8727 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
8728 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
8729 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
8731 * The Boot Loader Specification has been added to the source tree.
8733 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
8735 While moving it into our source tree we have updated it and further
8736 changes are now accepted through the usual github PR workflow.
8738 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
8739 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
8740 earlier PAM modules. The data in these fields is used to initialize
8741 the session scope's resource properties. Thus external PAM modules
8742 may now configure per-session limits, for example sourced from
8743 external user databases.
8745 * socket units with Accept=yes will now maintain a "refused" counter in
8746 addition to the existing "accepted" counter, counting connections
8747 refused due to the enforced limits.
8749 * The "systemd-path search-binaries-default" command may now be use to
8750 query the default, built-in $PATH PID 1 will pass to the services it
8753 * A new unit file setting PrivateMounts= has been added. It's a boolean
8754 option. If enabled the unit's processes are invoked in their own file
8755 system namespace. Note that this behaviour is also implied if any
8756 other file system namespacing options (such as PrivateTmp=,
8757 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
8758 primarily useful for services that do not use any of the other file
8759 system namespacing options. One such service is systemd-udevd.service
8760 where this is now used by default.
8762 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
8763 when the system is booted in UEFI "secure mode".
8765 * A new unit "system-update-pre.target" is added, which defines an
8766 optional synchronization point for offline system updates, as
8767 implemented by the pre-existing "system-update.target" unit. It
8768 allows ordering services before the service that executes the actual
8769 update process in a generic way.
8771 * Systemd now emits warnings whenever .include syntax is used.
8773 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
8774 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
8775 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
8776 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
8777 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
8778 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
8779 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
8780 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
8781 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
8782 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
8783 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
8784 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
8785 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
8786 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
8787 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
8788 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
8789 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
8790 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
8791 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
8792 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
8793 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
8794 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
8795 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
8796 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
8797 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
8798 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
8799 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
8800 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
8801 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8803 — Berlin, 2018-06-22
8807 * The MemoryAccounting= unit property now defaults to on. After
8808 discussions with the upstream control group maintainers we learnt
8809 that the negative impact of cgroup memory accounting on current
8810 kernels is finally relatively minimal, so that it should be safe to
8811 enable this by default without affecting system performance. Besides
8812 memory accounting only task accounting is turned on by default, all
8813 other forms of resource accounting (CPU, IO, IP) remain off for now,
8814 because it's not clear yet that their impact is small enough to move
8815 from opt-in to opt-out. We recommend downstreams to leave memory
8816 accounting on by default if kernel 4.14 or higher is primarily
8817 used. On very resource constrained systems or when support for old
8818 kernels is a necessity, -Dmemory-accounting-default=false can be used
8819 to revert this change.
8821 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
8822 %udev_rules_update) and the journal catalog (%journal_catalog_update)
8823 from the upgrade scriptlets of individual packages now do nothing.
8824 Transfiletriggers have been added which will perform those updates
8825 once at the end of the transaction.
8827 Similar transfiletriggers have been added to execute any sysctl.d
8828 and binfmt.d rules. Thus, it should be unnecessary to provide any
8829 scriptlets to execute this configuration from package installation
8832 * systemd-sysusers gained a mode where the configuration to execute is
8833 specified on the command line, but this configuration is not executed
8834 directly, but instead it is merged with the configuration on disk,
8835 and the result is executed. This is useful for package installation
8836 scripts which want to create the user before installing any files on
8837 disk (in case some of those files are owned by that user), while
8838 still allowing local admin overrides.
8840 This functionality is exposed to rpm scriptlets through a new
8841 %sysusers_create_package macro. Old %sysusers_create and
8842 %sysusers_create_inline macros are deprecated.
8844 A transfiletrigger for sysusers.d configuration is now installed,
8845 which means that it should be unnecessary to call systemd-sysusers from
8846 package installation scripts, unless the package installs any files
8847 owned by those newly-created users, in which case
8848 %sysusers_create_package should be used.
8850 * Analogous change has been done for systemd-tmpfiles: it gained a mode
8851 where the command-line configuration is merged with the configuration
8852 on disk. This is exposed as the new %tmpfiles_create_package macro,
8853 and %tmpfiles_create is deprecated. A transfiletrigger is installed
8854 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
8855 from package installation scripts.
8857 * sysusers.d configuration for a user may now also specify the group
8858 number, in addition to the user number ("u username 123:456"), or
8859 without the user number ("u username -:456").
8861 * Configution items for systemd-sysusers can now be specified as
8862 positional arguments when the new --inline switch is used.
8864 * The login shell of users created through sysusers.d may now be
8865 specified (previously, it was always /bin/sh for root and
8866 /sbin/nologin for other users).
8868 * systemd-analyze gained a new --global switch to look at global user
8869 configuration. It also gained a unit-paths verb to list the unit load
8870 paths that are compiled into systemd (which can be used with
8871 --systemd, --user, or --global).
8873 * udevadm trigger gained a new --settle/-w option to wait for any
8874 triggered events to finish (but just those, and not any other events
8875 which are triggered meanwhile).
8877 * The action that systemd-logind takes when the lid is closed and the
8878 machine is connected to external power can now be configured using
8879 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
8880 was determined by HandleLidSwitch=, and, for backwards compatibility,
8881 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
8883 * journalctl will periodically call sd_journal_process() to make it
8884 resilient against inotify queue overruns when journal files are
8885 rotated very quickly.
8887 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
8888 sd_bus_get_n_queued_write — may be used to check the number of
8889 pending bus messages.
8891 * systemd gained a new
8892 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
8893 which can be used to migrate foreign processes to scope and service
8894 units. The primary user for this new API is systemd itself: the
8895 systemd --user instance uses this call of the systemd --system
8896 instance to migrate processes if it itself gets the request to
8897 migrate processes and the kernel refuses this due to access
8898 restrictions. Thanks to this "systemd-run --scope --user …" works
8899 again in pure cgroup v2 environments when invoked from the user
8902 * A new TemporaryFileSystem= setting can be used to mask out part of
8903 the real file system tree with tmpfs mounts. This may be combined
8904 with BindPaths= and BindReadOnlyPaths= to hide files or directories
8905 not relevant to the unit, while still allowing some paths lower in
8906 the tree to be accessed.
8908 ProtectHome=tmpfs may now be used to hide user home and runtime
8909 directories from units, in a way that is mostly equivalent to
8910 "TemporaryFileSystem=/home /run/user /root".
8912 * Non-service units are now started with KeyringMode=shared by default.
8913 This means that mount and swapon and other mount tools have access
8914 to keys in the main keyring.
8916 * /sys/fs/bpf is now mounted automatically.
8918 * QNX virtualization is now detected by systemd-detect-virt and may
8919 be used in ConditionVirtualization=.
8921 * IPAccounting= may now be enabled also for slice units.
8923 * A new -Dsplit-bin= build configuration switch may be used to specify
8924 whether bin and sbin directories are merged, or if they should be
8925 included separately in $PATH and various listings of executable
8926 directories. The build configuration scripts will try to autodetect
8927 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
8928 system, but distributions are encouraged to configure this
8931 * A new -Dok-color= build configuration switch may be used to change
8932 the colour of "OK" status messages.
8934 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
8935 PrivateNetwork=yes was buggy in previous versions of systemd. This
8936 means that after the upgrade and daemon-reexec, any such units must
8939 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
8940 will not exclude read-only files owned by root from cleanup.
8942 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
8943 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
8944 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
8945 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
8946 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
8947 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
8948 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
8949 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
8950 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
8951 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
8952 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
8953 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
8954 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
8955 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
8956 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
8957 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
8959 — Warsaw, 2018-03-05
8963 * Some keyboards come with a zoom see-saw or rocker which until now got
8964 mapped to the Linux "zoomin/out" keys in hwdb. However, these
8965 keycodes are not recognized by any major desktop. They now produce
8966 Up/Down key events so that they can be used for scrolling.
8968 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
8969 slightly: previously, if an argument was specified for lines of this
8970 type (i.e. the right-most column was set) this string was appended to
8971 existing files each time systemd-tmpfiles was run. This behaviour was
8972 different from what the documentation said, and not particularly
8973 useful, as repeated systemd-tmpfiles invocations would not be
8974 idempotent and grow such files without bounds. With this release
8975 behaviour has been altered to match what the documentation says:
8976 lines of this type only have an effect if the indicated files don't
8977 exist yet, and only then the argument string is written to the file.
8979 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
8980 systemd-tmpfiles behaviour: previously, read-only files owned by root
8981 were always excluded from the file "aging" algorithm (i.e. the
8982 automatic clean-up of directories like /tmp based on
8983 atime/mtime/ctime). We intend to drop this restriction, and age files
8984 by default even when owned by root and read-only. This behaviour was
8985 inherited from older tools, but there have been requests to remove
8986 it, and it's not obvious why this restriction was made in the first
8987 place. Please speak up now, if you are aware of software that requires
8988 this behaviour, otherwise we'll remove the restriction in v238.
8990 * A new environment variable $SYSTEMD_OFFLINE is now understood by
8991 systemctl. It takes a boolean argument. If on, systemctl assumes it
8992 operates on an "offline" OS tree, and will not attempt to talk to the
8993 service manager. Previously, this mode was implicitly enabled if a
8994 chroot() environment was detected, and this new environment variable
8995 now provides explicit control.
8997 * .path and .socket units may now be created transiently, too.
8998 Previously only service, mount, automount and timer units were
8999 supported as transient units. The systemd-run tool has been updated
9000 to expose this new functionality, you may hence use it now to bind
9001 arbitrary commands to path or socket activation on-the-fly from the
9002 command line. Moreover, almost all properties are now exposed for the
9003 unit types that already supported transient operation.
9005 * The systemd-mount command gained support for a new --owner= parameter
9006 which takes a user name, which is then resolved and included in uid=
9007 and gid= mount options string of the file system to mount.
9009 * A new unit condition ConditionControlGroupController= has been added
9010 that checks whether a specific cgroup controller is available.
9012 * Unit files, udev's .link files, and systemd-networkd's .netdev and
9013 .network files all gained support for a new condition
9014 ConditionKernelVersion= for checking against specific kernel
9017 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
9018 support for configuring device flags in the Flags= setting. In the
9019 same files, the [Tunnel] section gained support for configuring
9020 AllowLocalRemote=. The [Route] section in .network files gained
9021 support for configuring InitialCongestionWindow=,
9022 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
9023 understands RapidCommit=.
9025 * systemd-networkd's DHCPv6 support gained support for Prefix
9028 * sd-bus gained support for a new "watch-bind" feature. When this
9029 feature is enabled, an sd_bus connection may be set up to connect to
9030 an AF_UNIX socket in the file system as soon as it is created. This
9031 functionality is useful for writing early-boot services that
9032 automatically connect to the system bus as soon as it is started,
9033 without ugly time-based polling. systemd-networkd and
9034 systemd-resolved have been updated to make use of this
9035 functionality. busctl exposes this functionality in a new
9036 --watch-bind= command line switch.
9038 * sd-bus will now optionally synthesize a local "Connected" signal as
9039 soon as a D-Bus connection is set up fully. This message mirrors the
9040 already existing "Disconnected" signal which is synthesized when the
9041 connection is terminated. This signal is generally useful but
9042 particularly handy in combination with the "watch-bind" feature
9043 described above. Synthesizing of this message has to be requested
9044 explicitly through the new API call sd_bus_set_connected_signal(). In
9045 addition a new call sd_bus_is_ready() has been added that checks
9046 whether a connection is fully set up (i.e. between the "Connected" and
9047 "Disconnected" signals).
9049 * sd-bus gained two new calls sd_bus_request_name_async() and
9050 sd_bus_release_name_async() for asynchronously registering bus
9051 names. Similar, there is now sd_bus_add_match_async() for installing
9052 a signal match asynchronously. All of systemd's own services have
9053 been updated to make use of these calls. Doing these operations
9054 asynchronously has two benefits: it reduces the risk of deadlocks in
9055 case of cyclic dependencies between bus services, and it speeds up
9056 service initialization since synchronization points for bus
9057 round-trips are removed.
9059 * sd-bus gained two new calls sd_bus_match_signal() and
9060 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
9061 and sd_bus_add_match_async() but instead of taking a D-Bus match
9062 string take match fields as normal function parameters.
9064 * sd-bus gained two new calls sd_bus_set_sender() and
9065 sd_bus_message_set_sender() for setting the sender name of outgoing
9066 messages (either for all outgoing messages or for just one specific
9067 one). These calls are only useful in direct connections as on
9068 brokered connections the broker fills in the sender anyway,
9069 overwriting whatever the client filled in.
9071 * sd-event gained a new pseudo-handle that may be specified on all API
9072 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
9073 used this refers to the default event loop object of the calling
9074 thread. Note however that this does not implicitly allocate one —
9075 which has to be done prior by using sd_event_default(). Similarly
9076 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
9077 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
9078 to the default bus of the specified type of the calling thread. Here
9079 too this does not implicitly allocate bus connection objects, this
9080 has to be done prior with sd_bus_default() and friends.
9082 * sd-event gained a new call pair
9083 sd_event_source_{get|set}_io_fd_own(). This may be used to request
9084 automatic closure of the file descriptor an IO event source watches
9085 when the event source is destroyed.
9087 * systemd-networkd gained support for natively configuring WireGuard
9090 * In previous versions systemd synthesized user records both for the
9091 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
9092 internally. In order to simplify distribution-wide renames of the
9093 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
9094 new transitional flag file has been added: if
9095 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
9096 user and group record within the systemd codebase is disabled.
9098 * systemd-notify gained a new --uid= option for selecting the source
9099 user/UID to use for notification messages sent to the service
9102 * journalctl gained a new --grep= option to list only entries in which
9103 the message matches a certain pattern. By default matching is case
9104 insensitive if the pattern is lowercase, and case sensitive
9105 otherwise. Option --case-sensitive=yes|no can be used to override
9106 this an specify case sensitivity or case insensitivity.
9108 * There's now a "systemd-analyze service-watchdogs" command for printing
9109 the current state of the service runtime watchdog, and optionally
9110 enabling or disabling the per-service watchdogs system-wide if given a
9111 boolean argument (i.e. the concept you configure in WatchdogSec=), for
9112 debugging purposes. There's also a kernel command line option
9113 systemd.service_watchdogs= for controlling the same.
9115 * Two new "log-level" and "log-target" options for systemd-analyze were
9116 added that merge the now deprecated get-log-level, set-log-level and
9117 get-log-target, set-log-target pairs. The deprecated options are still
9118 understood for backwards compatibility. The two new options print the
9119 current value when no arguments are given, and set them when a
9120 level/target is given as an argument.
9122 * sysusers.d's "u" lines now optionally accept both a UID and a GID
9123 specification, separated by a ":" character, in order to create users
9124 where UID and GID do not match.
9126 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
9127 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
9128 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
9129 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
9130 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
9131 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
9132 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
9133 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
9134 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
9135 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
9136 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
9137 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
9138 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
9139 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
9140 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
9141 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
9142 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
9143 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
9144 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
9145 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
9152 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
9153 in v235 has been extended to also set the dummy.ko module option
9154 numdummies=0, preventing the kernel from automatically creating
9155 dummy0. All dummy interfaces must now be explicitly created.
9157 * Unknown '%' specifiers in configuration files are now rejected. This
9158 applies to units and tmpfiles.d configuration. Any percent characters
9159 that are followed by a letter or digit that are not supposed to be
9160 interpreted as the beginning of a specifier should be escaped by
9161 doubling ("%%"). (So "size=5%" is still accepted, as well as
9162 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
9163 valid specifiers today.)
9165 * systemd-resolved now maintains a new dynamic
9166 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
9167 recommended to make /etc/resolv.conf a symlink to it. This file
9168 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
9169 includes dynamically acquired search domains, achieving more correct
9170 DNS resolution by software that bypasses local DNS APIs such as NSS.
9172 * The "uaccess" udev tag has been dropped from /dev/kvm and
9173 /dev/dri/renderD*. These devices now have the 0666 permissions by
9174 default (but this may be changed at build-time). /dev/dri/renderD*
9175 will now be owned by the "render" group along with /dev/kfd.
9177 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
9178 systemd-journal-gatewayd.service and
9179 systemd-journal-upload.service. This means "nss-systemd" must be
9180 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
9181 services are resolved properly.
9183 * In /etc/fstab two new mount options are now understood:
9184 x-systemd.makefs and x-systemd.growfs. The former has the effect that
9185 the configured file system is formatted before it is mounted, the
9186 latter that the file system is resized to the full block device size
9187 after it is mounted (i.e. if the file system is smaller than the
9188 partition it resides on, it's grown). This is similar to the fsck
9189 logic in /etc/fstab, and pulls in systemd-makefs@.service and
9190 systemd-growfs@.service as necessary, similar to
9191 systemd-fsck@.service. Resizing is currently only supported on ext4
9194 * In systemd-networkd, the IPv6 RA logic now optionally may announce
9195 DNS server and domain information.
9197 * Support for the LUKS2 on-disk format for encrypted partitions has
9198 been added. This requires libcryptsetup2 during compilation and
9201 * The systemd --user instance will now signal "readiness" when its
9202 basic.target unit has been reached, instead of when the run queue ran
9203 empty for the first time.
9205 * Tmpfiles.d with user configuration are now also supported.
9206 systemd-tmpfiles gained a new --user switch, and snippets placed in
9207 ~/.config/user-tmpfiles.d/ and corresponding directories will be
9208 executed by systemd-tmpfiles --user running in the new
9209 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
9210 running in the user session.
9212 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
9213 %S resolves to the top-level state directory (/var/lib for the system
9214 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
9215 top-level cache directory (/var/cache for the system instance,
9216 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
9217 logs directory (/var/log for the system instance,
9218 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
9219 existing %t specifier, that resolves to the top-level runtime
9220 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
9223 * journalctl learnt a new parameter --output-fields= for limiting the
9224 set of journal fields to output in verbose and JSON output modes.
9226 * systemd-timesyncd's configuration file gained a new option
9227 RootDistanceMaxSec= for setting the maximum root distance of servers
9228 it'll use, as well as the new options PollIntervalMinSec= and
9229 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
9231 * bootctl gained a new command "list" for listing all available boot
9232 menu items on systems that follow the boot loader specification.
9234 * systemctl gained a new --dry-run switch that shows what would be done
9235 instead of doing it, and is currently supported by the shutdown and
9238 * ConditionSecurity= can now detect the TOMOYO security module.
9240 * Unit file [Install] sections are now also respected in unit drop-in
9241 files. This is intended to be used by drop-ins under /usr/lib/.
9243 * systemd-firstboot may now also set the initial keyboard mapping.
9245 * Udev "changed" events for devices which are exposed as systemd
9246 .device units are now propagated to units specified in
9247 ReloadPropagatedFrom= as reload requests.
9249 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
9250 unit template name (i.e. a name in the form of 'foobar@.service',
9251 without the instance component between the '@' and - the '.'), then
9252 the escaped sysfs path of the device is automatically used as the
9255 * SystemCallFilter= in unit files has been extended so that an "errno"
9256 can be specified individually for each system call. Example:
9257 SystemCallFilter=~uname:EILSEQ.
9259 * The cgroup delegation logic has been substantially updated. Delegate=
9260 now optionally takes a list of controllers (instead of a boolean, as
9261 before), which lists the controllers to delegate at least.
9263 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
9265 * A new LogLevelMax= setting configures the maximum log level any
9266 process of the service may log at (i.e. anything with a lesser
9267 priority than what is specified is automatically dropped). A new
9268 LogExtraFields= setting allows configuration of additional journal
9269 fields to attach to all log records generated by any of the unit's
9272 * New StandardInputData= and StandardInputText= settings along with the
9273 new option StandardInput=data may be used to configure textual or
9274 binary data that shall be passed to the executed service process via
9275 standard input, encoded in-line in the unit file.
9277 * StandardInput=, StandardOutput= and StandardError= may now be used to
9278 connect stdin/stdout/stderr of executed processes directly with a
9279 file or AF_UNIX socket in the file system, using the new "file:" option.
9281 * A new unit file option CollectMode= has been added, that allows
9282 tweaking the garbage collection logic for units. It may be used to
9283 tell systemd to garbage collect units that have failed automatically
9284 (normally it only GCs units that exited successfully). systemd-run
9285 and systemd-mount expose this new functionality with a new -G option.
9287 * "machinectl bind" may now be used to bind mount non-directories
9288 (i.e. regularfiles, devices, fifos, sockets).
9290 * systemd-analyze gained a new verb "calendar" for validating and
9291 testing calendar time specifications to use for OnCalendar= in timer
9292 units. Besides validating the expression it will calculate the next
9293 time the specified expression would elapse.
9295 * In addition to the pre-existing FailureAction= unit file setting
9296 there's now SuccessAction=, for configuring a shutdown action to
9297 execute when a unit completes successfully. This is useful in
9298 particular inside containers that shall terminate after some workload
9299 has been completed. Also, both options are now supported for all unit
9300 types, not just services.
9302 * networkds's IP rule support gained two new options
9303 IncomingInterface= and OutgoingInterface= for configuring the incoming
9304 and outgoing interfaces of configured rules. systemd-networkd also
9305 gained support for "vxcan" network devices.
9307 * networkd gained a new setting RequiredForOnline=, taking a
9308 boolean. If set, systemd-wait-online will take it into consideration
9309 when determining that the system is up, otherwise it will ignore the
9310 interface for this purpose.
9312 * The sd_notify() protocol gained support for a new operation: with
9313 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
9314 store again, ahead of POLLHUP or POLLERR when they are removed
9317 * A new document doc/UIDS-GIDS.md has been added to the source tree,
9318 that documents the UID/GID range and assignment assumptions and
9319 requirements of systemd.
9321 * The watchdog device PID 1 will ping may now be configured through the
9322 WatchdogDevice= configuration file setting, or by setting the
9323 systemd.watchdog_service= kernel command line option.
9325 * systemd-resolved's gained support for registering DNS-SD services on
9326 the local network using MulticastDNS. Services may either be
9327 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
9328 the same dir below /run, /usr/lib), or through its D-Bus API.
9330 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
9331 extend the effective start, runtime, and stop time. The service must
9332 continue to send EXTEND_TIMEOUT_USEC within the period specified to
9333 prevent the service manager from making the service as timedout.
9335 * systemd-resolved's DNSSEC support gained support for RFC 8080
9336 (Ed25519 keys and signatures).
9338 * The systemd-resolve command line tool gained a new set of options
9339 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
9340 --set-nta= and --revert to configure per-interface DNS configuration
9341 dynamically during runtime. It's useful for pushing DNS information
9342 into systemd-resolved from DNS hook scripts that various interface
9343 managing software supports (such as pppd).
9345 * systemd-nspawn gained a new --network-namespace-path= command line
9346 option, which may be used to make a container join an existing
9347 network namespace, by specifying a path to a "netns" file.
9349 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
9350 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
9351 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
9352 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
9353 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
9354 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
9355 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
9356 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
9357 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
9358 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
9359 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
9360 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
9361 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
9362 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
9363 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
9364 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
9365 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
9366 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
9367 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
9368 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
9369 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
9370 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
9371 Jędrzejewski-Szmek, Zeal Jagannatha
9373 — Berlin, 2017-12-14
9377 * INCOMPATIBILITY: systemd-logind.service and other long-running
9378 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
9379 communication with the outside. This generally improves security of
9380 the system, and is in almost all cases a safe and good choice, as
9381 these services do not and should not provide any network-facing
9382 functionality. However, systemd-logind uses the glibc NSS API to
9383 query the user database. This creates problems on systems where NSS
9384 is set up to directly consult network services for user database
9385 lookups. In particular, this creates incompatibilities with the
9386 "nss-nis" module, which attempts to directly contact the NIS/YP
9387 network servers it is configured for, and will now consistently
9388 fail. In such cases, it is possible to turn off IP sandboxing for
9389 systemd-logind.service (set IPAddressDeny= in its [Service] section
9390 to the empty string, via a .d/ unit file drop-in). Downstream
9391 distributions might want to update their nss-nis packaging to include
9392 such a drop-in snippet, accordingly, to hide this incompatibility
9393 from the user. Another option is to make use of glibc's nscd service
9394 to proxy such network requests through a privilege-separated, minimal
9395 local caching daemon, or to switch to more modern technologies such
9396 sssd, whose NSS hook-ups generally do not involve direct network
9397 access. In general, we think it's definitely time to question the
9398 implementation choices of nss-nis, i.e. whether it's a good idea
9399 today to embed a network-facing loadable module into all local
9400 processes that need to query the user database, including the most
9401 trivial and benign ones, such as "ls". For more details about
9402 IPAddressDeny= see below.
9404 * A new modprobe.d drop-in is now shipped by default that sets the
9405 bonding module option max_bonds=0. This overrides the kernel default,
9406 to avoid conflicts and ambiguity as to whether or not bond0 should be
9407 managed by systemd-networkd or not. This resolves multiple issues
9408 with bond0 properties not being applied, when bond0 is configured
9409 with systemd-networkd. Distributors may choose to not package this,
9410 however in that case users will be prevented from correctly managing
9411 bond0 interface using systemd-networkd.
9413 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
9414 which print the logging level and target of the system manager. They
9415 complement the existing "set-log-level" and "set-log-target" verbs
9416 used to change those values.
9418 * journald.conf gained a new boolean setting ReadKMsg= which defaults
9419 to on. If turned off kernel log messages will not be read by
9420 systemd-journald or included in the logs. It also gained a new
9421 setting LineMax= for configuring the maximum line length in
9422 STDOUT/STDERR log streams. The new default for this value is 48K, up
9423 from the previous hardcoded 2048.
9425 * A new unit setting RuntimeDirectoryPreserve= has been added, which
9426 allows more detailed control of what to do with a runtime directory
9427 configured with RuntimeDirectory= (i.e. a directory below /run or
9428 $XDG_RUNTIME_DIR) after a unit is stopped.
9430 * The RuntimeDirectory= setting for units gained support for creating
9431 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
9432 one top-level directory.
9434 * Units gained new options StateDirectory=, CacheDirectory=,
9435 LogsDirectory= and ConfigurationDirectory= which are closely related
9436 to RuntimeDirectory= but manage per-service directories below
9437 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
9438 possible to write unit files which when activated automatically gain
9439 properly owned service specific directories in these locations, thus
9440 making unit files self-contained and increasing compatibility with
9441 stateless systems and factory reset where /etc or /var are
9442 unpopulated at boot. Matching these new settings there's also
9443 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
9444 ConfigurationDirectoryMode= for configuring the access mode of these
9445 directories. These settings are particularly useful in combination
9446 with DynamicUser=yes as they provide secure, properly-owned,
9447 writable, and stateful locations for storage, excluded from the
9448 sandbox that such services live in otherwise.
9450 * Automake support has been removed from this release. systemd is now
9453 * systemd-journald will now aggressively cache client metadata during
9454 runtime, speeding up log write performance under pressure. This comes
9455 at a small price though: as much of the metadata is read
9456 asynchronously from /proc/ (and isn't implicitly attached to log
9457 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
9458 metadata stored alongside a log entry might be slightly
9459 out-of-date. Previously it could only be slightly newer than the log
9460 message. The time window is small however, and given that the kernel
9461 is unlikely to be improved anytime soon in this regard, this appears
9464 * nss-myhostname/systemd-resolved will now by default synthesize an
9465 A/AAAA resource record for the "_gateway" hostname, pointing to the
9466 current default IP gateway. Previously it did that for the "gateway"
9467 name, hampering adoption, as some distributions wanted to leave that
9468 hostname open for local use. The old behaviour may still be
9469 requested at build time.
9471 * systemd-networkd's [Address] section in .network files gained a new
9472 Scope= setting for configuring the IP address scope. The [Network]
9473 section gained a new boolean setting ConfigureWithoutCarrier= that
9474 tells systemd-networkd to ignore link sensing when configuring the
9475 device. The [DHCP] section gained a new Anonymize= boolean option for
9476 turning on a number of options suggested in RFC 7844. A new
9477 [RoutingPolicyRule] section has been added for configuring the IP
9478 routing policy. The [Route] section has gained support for a new
9479 Type= setting which permits configuring
9480 blackhole/unreachable/prohibit routes.
9482 * The [VRF] section in .netdev files gained a new Table= setting for
9483 configuring the routing table to use. The [Tunnel] section gained a
9484 new Independent= boolean field for configuring tunnels independent of
9485 an underlying network interface. The [Bridge] section gained a new
9486 GroupForwardMask= option for configuration of propagation of link
9487 local frames between bridge ports.
9489 * The WakeOnLan= setting in .link files gained support for a number of
9490 new modes. A new TCP6SegmentationOffload= setting has been added for
9491 configuring TCP/IPv6 hardware segmentation offload.
9493 * The IPv6 RA sender implementation may now optionally send out RDNSS
9494 and RDNSSL records to supply DNS configuration to peers.
9496 * systemd-nspawn gained support for a new --system-call-filter= command
9497 line option for adding and removing entries in the default system
9498 call filter it applies. Moreover systemd-nspawn has been changed to
9499 implement a system call allow list instead of a deny list.
9501 * systemd-run gained support for a new --pipe command line option. If
9502 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
9503 are directly passed on to the activated transient service
9504 executable. This allows invoking arbitrary processes as systemd
9505 services (for example to take benefit of dependency management,
9506 accounting management, resource management or log management that is
9507 done automatically for services) — while still allowing them to be
9508 integrated in a classic UNIX shell pipeline.
9510 * When a service sends RELOAD=1 via sd_notify() and reload propagation
9511 using ReloadPropagationTo= is configured, a reload is now propagated
9512 to configured units. (Previously this was only done on explicitly
9513 requested reloads, using "systemctl reload" or an equivalent
9516 * For each service unit a restart counter is now kept: it is increased
9517 each time the service is restarted due to Restart=, and may be
9518 queried using "systemctl show -p NRestarts …".
9520 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
9521 @signal and @timer have been added, for usage with SystemCallFilter=
9522 in unit files and the new --system-call-filter= command line option
9523 of systemd-nspawn (see above).
9525 * ExecStart= lines in unit files gained two new modifiers: when a
9526 command line is prefixed with "!" the command will be executed as
9527 configured, except for the credentials applied by
9528 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
9529 "+", but does still apply namespacing options unlike "+". There's
9530 also "!!" now, which is mostly identical, but becomes a NOP on
9531 systems that support ambient capabilities. This is useful to write
9532 unit files that work with ambient capabilities where possible but
9533 automatically fall back to traditional privilege dropping mechanisms
9534 on systems where this is not supported.
9536 * ListenNetlink= settings in socket units now support RDMA netlink
9539 * A new unit file setting LockPersonality= has been added which permits
9540 locking down the chosen execution domain ("personality") of a service
9543 * A new special target "getty-pre.target" has been added, which is
9544 ordered before all text logins, and may be used to order services
9545 before textual logins acquire access to the console.
9547 * systemd will now attempt to load the virtio-rng.ko kernel module very
9548 early on if a VM environment supporting this is detected. This should
9549 improve entropy during early boot in virtualized environments.
9551 * A _netdev option is now supported in /etc/crypttab that operates in a
9552 similar way as the same option in /etc/fstab: it permits configuring
9553 encrypted devices that need to be ordered after the network is up.
9554 Following this logic, two new special targets
9555 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
9556 added that are to cryptsetup.target what remote-fs.target and
9557 remote-fs-pre.target are to local-fs.target.
9559 * Service units gained a new UnsetEnvironment= setting which permits
9560 unsetting specific environment variables for services that are
9561 normally passed to it (for example in order to mask out locale
9562 settings for specific services that can't deal with it).
9564 * Units acquired a new boolean option IPAccounting=. When turned on, IP
9565 traffic accounting (packet count as well as byte count) is done for
9566 the service, and shown as part of "systemctl status" or "systemd-run
9569 * Service units acquired two new options IPAddressAllow= and
9570 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
9571 for configuring a simple IP access control list for all sockets of
9572 the unit. These options are available also on .slice and .socket
9573 units, permitting flexible access list configuration for individual
9574 services as well as groups of services (as defined by a slice unit),
9575 including system-wide. Note that IP ACLs configured this way are
9576 enforced on every single IPv4 and IPv6 socket created by any process
9577 of the service unit, and apply to ingress as well as egress traffic.
9579 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
9580 structured log message is generated each time the unit is stopped,
9581 containing information about the consumed resources of this
9584 * A new setting KeyringMode= has been added to unit files, which may be
9585 used to control how the kernel keyring is set up for executed
9588 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
9589 "systemctl kexec" and "systemctl exit" are now always asynchronous in
9590 behaviour (that is: these commands return immediately after the
9591 operation was enqueued instead of waiting for the operation to
9592 complete). Previously, "systemctl poweroff" and "systemctl reboot"
9593 were asynchronous on systems using systemd-logind (i.e. almost
9594 always, and like they were on sysvinit), and the other three commands
9595 were unconditionally synchronous. With this release this is cleaned
9596 up, and callers will see the same asynchronous behaviour on all
9597 systems for all five operations.
9599 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
9602 * .timer units now accept calendar specifications in other timezones
9603 than UTC or the local timezone.
9605 * The tmpfiles snippet var.conf has been changed to create
9606 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
9607 the "utmp" group already, and it appears to be generally understood
9608 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
9609 databases. Previously this was implemented correctly for all these
9610 databases excepts btmp, which has been opened up like this now
9611 too. Note that while the other databases are world-readable
9612 (i.e. 0644), btmp is not and remains more restrictive.
9614 * The systemd-resolve tool gained a new --reset-server-features
9615 switch. When invoked like this systemd-resolved will forget
9616 everything it learnt about the features supported by the configured
9617 upstream DNS servers, and restarts the feature probing logic on the
9618 next resolver look-up for them at the highest feature level
9621 * The status dump systemd-resolved sends to the logs upon receiving
9622 SIGUSR1 now also includes information about all DNS servers it is
9623 configured to use, and the features levels it probed for them.
9625 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
9626 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
9627 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
9628 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
9629 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
9630 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
9631 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
9632 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
9633 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
9634 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
9635 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
9636 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
9637 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
9638 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
9639 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
9640 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
9641 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
9642 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
9643 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
9644 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9646 — Berlin, 2017-10-06
9650 * Meson is now supported as build system in addition to Automake. It is
9651 our plan to remove Automake in one of our next releases, so that
9652 Meson becomes our exclusive build system. Hence, please start using
9653 the Meson build system in your downstream packaging. There's plenty
9654 of documentation around how to use Meson, the extremely brief
9657 ./autogen.sh && ./configure && make && sudo make install
9661 meson build && ninja -C build && sudo ninja -C build install
9663 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
9664 which permits configuring a timeout on the time a job is
9665 running. This is particularly useful for setting timeouts on jobs for
9668 * Unit files gained two new options ConditionUser= and ConditionGroup=
9669 for conditionalizing units based on the identity of the user/group
9670 running a systemd user instance.
9672 * systemd-networkd now understands a new FlowLabel= setting in the
9673 [VXLAN] section of .network files, as well as a Priority= in
9674 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
9675 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
9676 gained support for configuration of GENEVE links, and IPv6 address
9677 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
9679 * .link files now understand a new Port= setting.
9681 * systemd-networkd's DHCP support gained support for DHCP option 119
9682 (domain search list).
9684 * systemd-networkd gained support for serving IPv6 address ranges using
9685 the Router Advertisement protocol. The new .network configuration
9686 section [IPv6Prefix] may be used to configure the ranges to
9687 serve. This is implemented based on a new, minimal, native server
9688 implementation of RA.
9690 * journalctl's --output= switch gained support for a new parameter
9691 "short-iso-precise" for a mode where timestamps are shown as precise
9694 * systemd-udevd's "net_id" builtin may now generate stable network
9695 interface names from IBM PowerVM VIO devices as well as ACPI platform
9698 * MulticastDNS support in systemd-resolved may now be explicitly
9699 enabled/disabled using the new MulticastDNS= configuration file
9702 * systemd-resolved may now optionally use libidn2 instead of the libidn
9703 for processing internationalized domain names. Support for libidn2
9704 should be considered experimental and should not be enabled by
9707 * "machinectl pull-tar" and related call may now do verification of
9708 downloaded images using SUSE-style .sha256 checksum files in addition
9709 to the already existing support for validating using Ubuntu-style
9712 * sd-bus gained support for a new sd_bus_message_appendv() call which
9713 is va_list equivalent of sd_bus_message_append().
9715 * sd-boot gained support for validating images using SHIM/MOK.
9717 * The SMACK code learnt support for "onlycap".
9719 * systemd-mount --umount is now much smarter in figuring out how to
9720 properly unmount a device given its mount or device path.
9722 * The code to call libnss_dns as a fallback from libnss_resolve when
9723 the communication with systemd-resolved fails was removed. This
9724 fallback was redundant and interfered with the [!UNAVAIL=return]
9725 suffix. See nss-resolve(8) for the recommended configuration.
9727 * systemd-logind may now be restarted without losing state. It stores
9728 the file descriptors for devices it manages in the system manager
9729 using the FDSTORE= mechanism. Please note that further changes in
9730 other components may be required to make use of this (for example
9731 Xorg has code to listen for stops of systemd-logind and terminate
9732 itself when logind is stopped or restarted, in order to avoid using
9733 stale file descriptors for graphical devices, which is now
9734 counterproductive and must be reverted in order for restarts of
9735 systemd-logind to be safe. See
9736 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
9738 * All kernel-install plugins are called with the environment variable
9739 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
9740 /etc/machine-id. If the machine ID could not be determined,
9741 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
9742 anything in the entry directory (passed as the second argument) if
9743 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
9744 temporary directory is passed as the entry directory and removed
9745 after all the plugins exit.
9747 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
9748 will now use its value as the machine ID instead of the machine ID
9749 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
9750 /etc/machine-info and no machine ID is set in /etc/machine-id,
9751 kernel-install will try to store the current machine ID there as
9752 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
9753 will generate a new UUID, store it in /etc/machine-info as
9754 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
9756 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
9757 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
9758 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
9759 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
9760 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
9761 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
9762 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
9763 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
9764 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
9765 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
9766 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
9767 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
9768 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
9769 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
9770 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
9771 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
9772 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
9773 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
9774 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
9775 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
9776 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
9777 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
9778 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
9779 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
9780 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
9781 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
9782 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
9783 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
9784 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
9787 — Berlin, 2017-07-12
9791 * The "hybrid" control group mode has been modified to improve
9792 compatibility with "legacy" cgroups-v1 setups. Specifically, the
9793 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
9794 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
9795 cgroups-v1 hierarchy), the only externally visible change being that
9796 the cgroups-v2 hierarchy is also mounted, to
9797 /sys/fs/cgroup/unified. This should provide a large degree of
9798 compatibility with "legacy" cgroups-v1, while taking benefit of the
9799 better management capabilities of cgroups-v2.
9801 * The default control group setup mode may be selected both a boot-time
9802 via a set of kernel command line parameters (specifically:
9803 systemd.unified_cgroup_hierarchy= and
9804 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
9805 default selected on the configure command line
9806 (--with-default-hierarchy=). The upstream default is "hybrid"
9807 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
9808 this will change in a future systemd version to be "unified" (pure
9809 cgroups-v2 mode). The third option for the compile time option is
9810 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
9811 distributions to default to "hybrid" mode for release distributions,
9812 starting with v233. We recommend "unified" for development
9813 distributions (specifically: distributions such as Fedora's rawhide)
9814 as that's where things are headed in the long run. Use "legacy" for
9815 greatest stability and compatibility only.
9817 * Note one current limitation of "unified" and "hybrid" control group
9818 setup modes: the kernel currently does not permit the systemd --user
9819 instance (i.e. unprivileged code) to migrate processes between two
9820 disconnected cgroup subtrees, even if both are managed and owned by
9821 the user. This effectively means "systemd-run --user --scope" doesn't
9822 work when invoked from outside of any "systemd --user" service or
9823 scope. Specifically, it is not supported from session scopes. We are
9824 working on fixing this in a future systemd version. (See #3388 for
9825 further details about this.)
9827 * DBus policy files are now installed into /usr rather than /etc. Make
9828 sure your system has dbus >= 1.9.18 running before upgrading to this
9829 version, or override the install path with --with-dbuspolicydir= .
9831 * All python scripts shipped with systemd (specifically: the various
9832 tests written in Python) now require Python 3.
9834 * systemd unit tests can now run standalone (without the source or
9835 build directories), and can be installed into /usr/lib/systemd/tests/
9836 with 'make install-tests'.
9838 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
9839 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
9842 * Support for the %c, %r, %R specifiers in unit files has been
9843 removed. Specifiers are not supposed to be dependent on configuration
9844 in the unit file itself (so that they resolve the same regardless
9845 where used in the unit files), but these specifiers were influenced
9846 by the Slice= option.
9848 * The shell invoked by debug-shell.service now defaults to /bin/sh in
9849 all cases. If distributions want to use a different shell for this
9850 purpose (for example Fedora's /sbin/sushell) they need to specify
9851 this explicitly at configure time using --with-debug-shell=.
9853 * The confirmation spawn prompt has been reworked to offer the
9856 (c)ontinue, proceed without asking anymore
9857 (D)ump, show the state of the unit
9858 (f)ail, don't execute the command and pretend it failed
9860 (i)nfo, show a short summary of the unit
9861 (j)obs, show jobs that are in progress
9862 (s)kip, don't execute the command and pretend it succeeded
9863 (y)es, execute the command
9865 The 'n' choice for the confirmation spawn prompt has been removed,
9866 because its meaning was confusing.
9868 The prompt may now also be redirected to an alternative console by
9869 specifying the console as parameter to systemd.confirm_spawn=.
9871 * Services of Type=notify require a READY=1 notification to be sent
9872 during startup. If no such message is sent, the service now fails,
9873 even if the main process exited with a successful exit code.
9875 * Services that fail to start up correctly now always have their
9876 ExecStopPost= commands executed. Previously, they'd enter "failed"
9877 state directly, without executing these commands.
9879 * The option MulticastDNS= of network configuration files has acquired
9880 an actual implementation. With MulticastDNS=yes a host can resolve
9881 names of remote hosts and reply to mDNS A and AAAA requests.
9883 * When units are about to be started an additional check is now done to
9884 ensure that all dependencies of type BindsTo= (when used in
9885 combination with After=) have been started.
9887 * systemd-analyze gained a new verb "syscall-filter" which shows which
9888 system call groups are defined for the SystemCallFilter= unit file
9889 setting, and which system calls they contain.
9891 * A new system call filter group "@filesystem" has been added,
9892 consisting of various file system related system calls. Group
9893 "@reboot" has been added, covering reboot, kexec and shutdown related
9894 calls. Finally, group "@swap" has been added covering swap
9895 configuration related calls.
9897 * A new unit file option RestrictNamespaces= has been added that may be
9898 used to restrict access to the various process namespace types the
9899 Linux kernel provides. Specifically, it may be used to take away the
9900 right for a service unit to create additional file system, network,
9901 user, and other namespaces. This sandboxing option is particularly
9902 relevant due to the high amount of recently discovered namespacing
9903 related vulnerabilities in the kernel.
9905 * systemd-udev's .link files gained support for a new AutoNegotiation=
9906 setting for configuring Ethernet auto-negotiation.
9908 * systemd-networkd's .network files gained support for a new
9909 ListenPort= setting in the [DHCP] section to explicitly configure the
9910 UDP client port the DHCP client shall listen on.
9912 * .network files gained a new Unmanaged= boolean setting for explicitly
9913 excluding one or more interfaces from management by systemd-networkd.
9915 * The systemd-networkd ProxyARP= option has been renamed to
9916 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
9917 renamed to ReduceARPProxy=. The old names continue to be available
9920 * systemd-networkd gained support for configuring IPv6 Proxy NDP
9921 addresses via the new IPv6ProxyNDPAddress= .network file setting.
9923 * systemd-networkd's bonding device support gained support for two new
9924 configuration options ActiveSlave= and PrimarySlave=.
9926 * The various options in the [Match] section of .network files gained
9927 support for negative matching.
9929 * New systemd-specific mount options are now understood in /etc/fstab:
9931 x-systemd.mount-timeout= may be used to configure the maximum
9932 permitted runtime of the mount command.
9934 x-systemd.device-bound may be set to bind a mount point to its
9935 backing device unit, in order to automatically remove a mount point
9936 if its backing device is unplugged. This option may also be
9937 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
9938 on the block device, which is now automatically set for all CDROM
9939 drives, so that mounted CDs are automatically unmounted when they are
9940 removed from the drive.
9942 x-systemd.after= and x-systemd.before= may be used to explicitly
9943 order a mount after or before another unit or mount point.
9945 * Enqueued start jobs for device units are now automatically garbage
9946 collected if there are no jobs waiting for them anymore.
9948 * systemctl list-jobs gained two new switches: with --after, for every
9949 queued job the jobs it's waiting for are shown; with --before the
9950 jobs which it's blocking are shown.
9952 * systemd-nspawn gained support for ephemeral boots from disk images
9953 (or in other words: --ephemeral and --image= may now be
9954 combined). Moreover, ephemeral boots are now supported for normal
9955 directories, even if the backing file system is not btrfs. Of course,
9956 if the file system does not support file system snapshots or
9957 reflinks, the initial copy operation will be relatively expensive, but
9958 this should still be suitable for many use cases.
9960 * Calendar time specifications in .timer units now support
9961 specifications relative to the end of a month by using "~" instead of
9962 "-" as separator between month and day. For example, "*-02~03" means
9963 "the third last day in February". In addition a new syntax for
9964 repeated events has been added using the "/" character. For example,
9965 "9..17/2:00" means "every two hours from 9am to 5pm".
9967 * systemd-socket-proxyd gained a new parameter --connections-max= for
9968 configuring the maximum number of concurrent connections.
9970 * sd-id128 gained a new API for generating unique IDs for the host in a
9971 way that does not leak the machine ID. Specifically,
9972 sd_id128_get_machine_app_specific() derives an ID based on the
9973 machine ID in a well-defined, non-reversible, stable way. This is
9974 useful whenever an identifier for the host is needed but where the
9975 identifier shall not be useful to identify the system beyond the
9976 scope of the application itself. (Internally this uses HMAC-SHA256 as
9977 keyed hash function using the machine ID as input.)
9979 * NotifyAccess= gained a new supported value "exec". When set
9980 notifications are accepted from all processes systemd itself invoked,
9981 including all control processes.
9983 * .nspawn files gained support for defining overlay mounts using the
9984 Overlay= and OverlayReadOnly= options. Previously this functionality
9985 was only available on the systemd-nspawn command line.
9987 * systemd-nspawn's --bind= and --overlay= options gained support for
9988 bind/overlay mounts whose source lies within the container tree by
9989 prefixing the source path with "+".
9991 * systemd-nspawn's --bind= and --overlay= options gained support for
9992 automatically allocating a temporary source directory in /var/tmp
9993 that is removed when the container dies. Specifically, if the source
9994 directory is specified as empty string this mechanism is selected. An
9995 example usage is --overlay=+/var::/var, which creates an overlay
9996 mount based on the original /var contained in the image, overlaid
9997 with a temporary directory in the host's /var/tmp. This way changes
9998 to /var are automatically flushed when the container shuts down.
10000 * systemd-nspawn --image= option does now permit raw file system block
10001 devices (in addition to images containing partition tables, as
10004 * The disk image dissection logic in systemd-nspawn gained support for
10005 automatically setting up LUKS encrypted as well as Verity protected
10006 partitions. When a container is booted from an encrypted image the
10007 passphrase is queried at start-up time. When a container with Verity
10008 data is started, the root hash is search in a ".roothash" file
10009 accompanying the disk image (alternatively, pass the root hash via
10010 the new --root-hash= command line option).
10012 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
10013 be used to dissect disk images the same way as systemd-nspawn does
10014 it, following the Bootable Partition Specification. It may even be
10015 used to mount disk images with complex partition setups (including
10016 LUKS and Verity partitions) to a local host directory, in order to
10017 inspect them. This tool is not considered public API (yet), and is
10018 thus not installed into /usr/bin. Please do not rely on its
10019 existence, since it might go away or be changed in later systemd
10022 * A new generator "systemd-verity-generator" has been added, similar in
10023 style to "systemd-cryptsetup-generator", permitting automatic setup of
10024 Verity root partitions when systemd boots up. In order to make use of
10025 this your partition setup should follow the Discoverable Partitions
10026 Specification, and the GPT partition ID of the root file system
10027 partition should be identical to the upper 128-bit of the Verity root
10028 hash. The GPT partition ID of the Verity partition protecting it
10029 should be the lower 128-bit of the Verity root hash. If the partition
10030 image follows this model it is sufficient to specify a single
10031 "roothash=" kernel command line argument to both configure which root
10032 image and verity partition to use as well as the root hash for
10033 it. Note that systemd-nspawn's Verity support follows the same
10034 semantics, meaning that disk images with proper Verity data in place
10035 may be booted in containers with systemd-nspawn as well as on
10036 physical systems via the verity generator. Also note that the "mkosi"
10037 tool available at https://github.com/systemd/mkosi has been updated
10038 to generate Verity protected disk images following this scheme. In
10039 fact, it has been updated to generate disk images that optionally
10040 implement a complete UEFI SecureBoot trust chain, involving a signed
10041 kernel and initrd image that incorporates such a root hash as well as
10042 a Verity-enabled root partition.
10044 * The hardware database (hwdb) udev supports has been updated to carry
10045 accelerometer quirks.
10047 * All system services are now run with a fresh kernel keyring set up
10048 for them. The invocation ID is stored by default in it, thus
10049 providing a safe, non-overridable way to determine the invocation
10050 ID of each service.
10052 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
10053 options for bind mounting arbitrary paths in a service-specific
10054 way. When these options are used, arbitrary host or service files and
10055 directories may be mounted to arbitrary locations in the service's
10058 * Documentation has been added that lists all of systemd's low-level
10059 environment variables:
10061 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
10063 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
10064 whether a specific socket file descriptor matches a specified socket
10067 * systemd-firstboot has been updated to check for the
10068 systemd.firstboot= kernel command line option. It accepts a boolean
10069 and when set to false the first boot questions are skipped.
10071 * systemd-fstab-generator has been updated to check for the
10072 systemd.volatile= kernel command line option, which either takes an
10073 optional boolean parameter or the special value "state". If used the
10074 system may be booted in a "volatile" boot mode. Specifically,
10075 "systemd.volatile" is used, the root directory will be mounted as
10076 tmpfs, and only /usr is mounted from the actual root file system. If
10077 "systemd.volatile=state" is used, the root directory will be mounted
10078 as usual, but /var is mounted as tmpfs. This concept provides similar
10079 functionality as systemd-nspawn's --volatile= option, but provides it
10080 on physical boots. Use this option for implementing stateless
10081 systems, or testing systems with all state and/or configuration reset
10082 to the defaults. (Note though that many distributions are not
10083 prepared to boot up without a populated /etc or /var, though.)
10085 * systemd-gpt-auto-generator gained support for LUKS encrypted root
10086 partitions. Previously it only supported LUKS encrypted partitions
10087 for all other uses, except for the root partition itself.
10089 * Socket units gained support for listening on AF_VSOCK sockets for
10090 communication in virtualized QEMU environments.
10092 * The "configure" script gained a new option --with-fallback-hostname=
10093 for specifying the fallback hostname to use if none is configured in
10094 /etc/hostname. For example, by specifying
10095 --with-fallback-hostname=fedora it is possible to default to a
10096 hostname of "fedora" on pristine installations.
10098 * systemd-cgls gained support for a new --unit= switch for listing only
10099 the control groups of a specific unit. Similar --user-unit= has been
10100 added for listing only the control groups of a specific user unit.
10102 * systemd-mount gained a new --umount switch for unmounting a mount or
10103 automount point (and all mount/automount points below it).
10105 * systemd will now refuse full configuration reloads (via systemctl
10106 daemon-reload and related calls) unless at least 16MiB of free space
10107 are available in /run. This is a safety precaution in order to ensure
10108 that generators can safely operate after the reload completed.
10110 * A new unit file option RootImage= has been added, which has a similar
10111 effect as RootDirectory= but mounts the service's root directory from
10112 a disk image instead of plain directory. This logic reuses the same
10113 image dissection and mount logic that systemd-nspawn already uses,
10114 and hence supports any disk images systemd-nspawn supports, including
10115 those following the Discoverable Partition Specification, as well as
10116 Verity enabled images. This option enables systemd to run system
10117 services directly off disk images acting as resource bundles,
10118 possibly even including full integrity data.
10120 * A new MountAPIVFS= unit file option has been added, taking a boolean
10121 argument. If enabled /proc, /sys and /dev (collectively called the
10122 "API VFS") will be mounted for the service. This is only relevant if
10123 RootDirectory= or RootImage= is used for the service, as these mounts
10124 are of course in place in the host mount namespace anyway.
10126 * systemd-nspawn gained support for a new --pivot-root= switch. If
10127 specified the root directory within the container image is pivoted to
10128 the specified mount point, while the original root disk is moved to a
10129 different place. This option enables booting of ostree images
10130 directly with systemd-nspawn.
10132 * The systemd build scripts will no longer complain if the NTP server
10133 addresses are not changed from the defaults. Google now supports
10134 these NTP servers officially. We still recommend downstreams to
10135 properly register an NTP pool with the NTP pool project though.
10137 * coredumpctl gained a new "--reverse" option for printing the list
10138 of coredumps in reverse order.
10140 * coredumpctl will now show additional information about truncated and
10141 inaccessible coredumps, as well as coredumps that are still being
10142 processed. It also gained a new --quiet switch for suppressing
10143 additional informational message in its output.
10145 * coredumpctl gained support for only showing coredumps newer and/or
10146 older than specific timestamps, using the new --since= and --until=
10147 options, reminiscent of journalctl's options by the same name.
10149 * The systemd-coredump logic has been improved so that it may be reused
10150 to collect backtraces in non-compiled languages, for example in
10151 scripting languages such as Python.
10153 * machinectl will now show the UID shift of local containers, if user
10154 namespacing is enabled for them.
10156 * systemd will now optionally run "environment generator" binaries at
10157 configuration load time. They may be used to add environment
10158 variables to the environment block passed to services invoked. One
10159 user environment generator is shipped by default that sets up
10160 environment variables based on files dropped into /etc/environment.d
10161 and ~/.config/environment.d/.
10163 * systemd-resolved now includes the new, recently published 2017 DNSSEC
10166 * hostnamed has been updated to report a new chassis type of
10167 "convertible" to cover "foldable" laptops that can both act as a
10168 tablet and as a laptop, such as various Lenovo Yoga devices.
10170 Contributions from: Adrián López, Alexander Galanin, Alexander
10171 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
10172 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
10173 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
10174 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
10175 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
10176 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
10177 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
10178 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
10179 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
10180 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
10181 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
10182 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
10183 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
10184 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
10185 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
10186 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
10187 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
10188 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
10189 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
10190 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
10191 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
10192 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
10193 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
10194 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
10195 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
10196 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
10197 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
10200 — Berlin, 2017-03-01
10204 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
10205 RestrictAddressFamilies= enabled. These sandboxing options should
10206 generally be compatible with the various external udev call-out
10207 binaries we are aware of, however there may be exceptions, in
10208 particular when exotic languages for these call-outs are used. In
10209 this case, consider turning off these settings locally.
10211 * The new RemoveIPC= option can be used to remove IPC objects owned by
10212 the user or group of a service when that service exits.
10214 * The new ProtectKernelModules= option can be used to disable explicit
10215 load and unload operations of kernel modules by a service. In
10216 addition access to /usr/lib/modules is removed if this option is set.
10218 * ProtectSystem= option gained a new value "strict", which causes the
10219 whole file system tree with the exception of /dev, /proc, and /sys,
10220 to be remounted read-only for a service.
10222 * The new ProtectKernelTunables= option can be used to disable
10223 modification of configuration files in /sys and /proc by a service.
10224 Various directories and files are remounted read-only, so access is
10225 restricted even if the file permissions would allow it.
10227 * The new ProtectControlGroups= option can be used to disable write
10228 access by a service to /sys/fs/cgroup.
10230 * Various systemd services have been hardened with
10231 ProtectKernelTunables=yes, ProtectControlGroups=yes,
10232 RestrictAddressFamilies=.
10234 * Support for dynamically creating users for the lifetime of a service
10235 has been added. If DynamicUser=yes is specified, user and group IDs
10236 will be allocated from the range 61184…65519 for the lifetime of the
10237 service. They can be resolved using the new nss-systemd.so NSS
10238 module. The module must be enabled in /etc/nsswitch.conf. Services
10239 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
10240 any resources allocated by the service will be cleaned up when the
10241 service exits. They also have ProtectHome=read-only and
10242 ProtectSystem=strict enabled, so they are not able to make any
10243 permanent modifications to the system.
10245 * The nss-systemd module also always resolves root and nobody, making
10246 it possible to have no /etc/passwd or /etc/group files in minimal
10247 container or chroot environments.
10249 * Services may be started with their own user namespace using the new
10250 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
10251 under which the service is running are mapped. All other users are
10254 * Support for the cgroup namespace has been added to systemd-nspawn. If
10255 supported by kernel, the container system started by systemd-nspawn
10256 will have its own view of the cgroup hierarchy. This new behaviour
10257 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
10259 * The new MemorySwapMax= option can be used to limit the maximum swap
10260 usage under the unified cgroup hierarchy.
10262 * Support for the CPU controller in the unified cgroup hierarchy has
10263 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
10264 options. This controller requires out-of-tree patches for the kernel
10265 and the support is provisional.
10267 * Mount and automount units may now be created transiently
10268 (i.e. dynamically at runtime via the bus API, instead of requiring
10269 unit files in the file system).
10271 * systemd-mount is a new tool which may mount file systems – much like
10272 mount(8), optionally pulling in additional dependencies through
10273 transient .mount and .automount units. For example, this tool
10274 automatically runs fsck on a backing block device before mounting,
10275 and allows the automount logic to be used dynamically from the
10276 command line for establishing mount points. This tool is particularly
10277 useful when dealing with removable media, as it will ensure fsck is
10278 run – if necessary – before the first access and that the file system
10279 is quickly unmounted after each access by utilizing the automount
10280 logic. This maximizes the chance that the file system on the
10281 removable media stays in a clean state, and if it isn't in a clean
10282 state is fixed automatically.
10284 * LazyUnmount=yes option for mount units has been added to expose the
10285 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
10288 * /efi will be used as the mount point of the EFI boot partition, if
10289 the directory is present, and the mount point was not configured
10290 through other means (e.g. fstab). If /efi directory does not exist,
10291 /boot will be used as before. This makes it easier to automatically
10292 mount the EFI partition on systems where /boot is used for something
10295 * When operating on GPT disk images for containers, systemd-nspawn will
10296 now mount the ESP to /boot or /efi according to the same rules as PID
10297 1 running on a host. This allows tools like "bootctl" to operate
10298 correctly within such containers, in order to make container images
10299 bootable on physical systems.
10301 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
10303 * Two new user session targets have been added to support running
10304 graphical sessions under the systemd --user instance:
10305 graphical-session.target and graphical-session-pre.target. See
10306 systemd.special(7) for a description of how those targets should be
10309 * The vconsole initialization code has been significantly reworked to
10310 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
10311 support unicode keymaps. Font and keymap configuration will now be
10312 copied to all allocated virtual consoles.
10314 * FreeBSD's bhyve virtualization is now detected.
10316 * Information recorded in the journal for core dumps now includes the
10317 contents of /proc/mountinfo and the command line of the process at
10318 the top of the process hierarchy (which is usually the init process
10321 * systemd-journal-gatewayd learned the --directory= option to serve
10322 files from the specified location.
10324 * journalctl --root=… can be used to peruse the journal in the
10325 /var/log/ directories inside of a container tree. This is similar to
10326 the existing --machine= option, but does not require the container to
10329 * The hardware database has been extended to support
10330 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
10333 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
10334 specify the click rate for mice which include a horizontal wheel with
10335 a click rate that is different than the one for the vertical wheel.
10337 * systemd-run gained a new --wait option that makes service execution
10338 synchronous. (Specifically, the command will not return until the
10339 specified service binary exited.)
10341 * systemctl gained a new --wait option that causes the start command to
10342 wait until the units being started have terminated again.
10344 * A new journal output mode "short-full" has been added which displays
10345 timestamps with abbreviated English day names and adds a timezone
10346 suffix. Those timestamps include more information than the default
10347 "short" output mode, and can be passed directly to journalctl's
10348 --since= and --until= options.
10350 * /etc/resolv.conf will be bind-mounted into containers started by
10351 systemd-nspawn, if possible, so any changes to resolv.conf contents
10352 are automatically propagated to the container.
10354 * The number of instances for socket-activated services originating
10355 from a single IP address can be limited with
10356 MaxConnectionsPerSource=, extending the existing setting of
10359 * systemd-networkd gained support for vcan ("Virtual CAN") interface
10362 * .netdev and .network configuration can now be extended through
10365 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
10366 Segmentation Offload, Generic Receive Offload, Large Receive Offload
10367 can be enabled and disabled using the new UDPSegmentationOffload=,
10368 TCPSegmentationOffload=, GenericSegmentationOffload=,
10369 GenericReceiveOffload=, LargeReceiveOffload= options in the
10370 [Link] section of .link files.
10372 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
10373 Port VLAN ID can be configured for bridge devices using the new STP=,
10374 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
10375 section of .netdev files.
10377 * The route table to which routes received over DHCP or RA should be
10378 added can be configured with the new RouteTable= option in the [DHCP]
10379 and [IPv6AcceptRA] sections of .network files.
10381 * The Address Resolution Protocol can be disabled on links managed by
10382 systemd-networkd using the ARP=no setting in the [Link] section of
10385 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
10386 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
10387 encode information about the result and exit codes of the current
10388 service runtime cycle.
10390 * systemd-sysctl will now configure kernel parameters in the order
10391 they occur in the configuration files. This matches what sysctl
10392 has been traditionally doing.
10394 * kernel-install "plugins" that are executed to perform various
10395 tasks after a new kernel is added and before an old one is removed
10396 can now return a special value to terminate the procedure and
10397 prevent any later plugins from running.
10399 * Journald's SplitMode=login setting has been deprecated. It has been
10400 removed from documentation, and its use is discouraged. In a future
10401 release it will be completely removed, and made equivalent to current
10402 default of SplitMode=uid.
10404 * Storage=both option setting in /etc/systemd/coredump.conf has been
10405 removed. With fast LZ4 compression storing the core dump twice is not
10408 * The --share-system systemd-nspawn option has been replaced with an
10409 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
10410 this functionality is discouraged. In addition the variables
10411 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
10412 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
10413 individual namespaces.
10415 * "machinectl list" now shows the IP address of running containers in
10416 the output, as well as OS release information.
10418 * "loginctl list" now shows the TTY of each session in the output.
10420 * sd-bus gained new API calls sd_bus_track_set_recursive(),
10421 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
10422 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
10423 tracking objects in a "recursive" mode, where a single client can be
10424 counted multiple times, if it takes multiple references.
10426 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
10427 sd_bus_get_exit_on_disconnect(). They may be used to make a
10428 process using sd-bus automatically exit if the bus connection is
10431 * Bus clients of the service manager may now "pin" loaded units into
10432 memory, by taking an explicit reference on them. This is useful to
10433 ensure the client can retrieve runtime data about the service even
10434 after the service completed execution. Taking such a reference is
10435 available only for privileged clients and should be helpful to watch
10436 running services in a race-free manner, and in particular collect
10437 information about exit statuses and results.
10439 * The nss-resolve module has been changed to strictly return UNAVAIL
10440 when communication via D-Bus with resolved failed, and NOTFOUND when
10441 a lookup completed but was negative. This means it is now possible to
10442 neatly configure fallbacks using nsswitch.conf result checking
10443 expressions. Taking benefit of this, the new recommended
10444 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
10446 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
10448 * A new setting CtrlAltDelBurstAction= has been added to
10449 /etc/systemd/system.conf which may be used to configure the precise
10450 behaviour if the user on the console presses Ctrl-Alt-Del more often
10451 than 7 times in 2s. Previously this would unconditionally result in
10452 an expedited, immediate reboot. With this new setting the precise
10453 operation may be configured in more detail, and also turned off
10456 * In .netdev files two new settings RemoteChecksumTx= and
10457 RemoteChecksumRx= are now understood that permit configuring the
10458 remote checksumming logic for VXLAN networks.
10460 * The service manager learnt a new "invocation ID" concept for invoked
10461 services. Each runtime cycle of a service will get a new invocation
10462 ID (a 128-bit random UUID) assigned that identifies the current
10463 run of the service uniquely and globally. A new invocation ID
10464 is generated each time a service starts up. The journal will store
10465 the invocation ID of a service along with any logged messages, thus
10466 making the invocation ID useful for matching the online runtime of a
10467 service with the offline log data it generated in a safe way without
10468 relying on synchronized timestamps. In many ways this new service
10469 invocation ID concept is similar to the kernel's boot ID concept that
10470 uniquely and globally identifies the runtime of each boot. The
10471 invocation ID of a service is passed to the service itself via an
10472 environment variable ($INVOCATION_ID). A new bus call
10473 GetUnitByInvocationID() has been added that is similar to GetUnit()
10474 but instead of retrieving the bus path for a unit by its name
10475 retrieves it by its invocation ID. The returned path is valid only as
10476 long as the passed invocation ID is current.
10478 * systemd-resolved gained a new "DNSStubListener" setting in
10479 resolved.conf. It either takes a boolean value or the special values
10480 "udp" and "tcp", and configures whether to enable the stub DNS
10481 listener on 127.0.0.53:53.
10483 * IP addresses configured via networkd may now carry additional
10484 configuration settings supported by the kernel. New options include:
10485 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
10486 PrefixRoute=, AutoJoin=.
10488 * The PAM configuration fragment file for "user@.service" shipped with
10489 systemd (i.e. the --user instance of systemd) has been stripped to
10490 the minimum necessary to make the system boot. Previously, it
10491 contained Fedora-specific stanzas that did not apply to other
10492 distributions. It is expected that downstream distributions add
10493 additional configuration lines, matching their needs to this file,
10494 using it only as rough template of what systemd itself needs. Note
10495 that this reduced fragment does not even include an invocation of
10496 pam_limits which most distributions probably want to add, even though
10497 systemd itself does not need it. (There's also the new build time
10498 option --with-pamconfdir=no to disable installation of the PAM
10499 fragment entirely.)
10501 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
10502 capability is now also dropped from its set (in addition to
10503 CAP_SYS_MKNOD as before).
10505 * In service unit files it is now possible to connect a specific named
10506 file descriptor with stdin/stdout/stdout of an executed service. The
10507 name may be specified in matching .socket units using the
10508 FileDescriptorName= setting.
10510 * A number of journal settings may now be configured on the kernel
10511 command line. Specifically, the following options are now understood:
10512 systemd.journald.max_level_console=,
10513 systemd.journald.max_level_store=,
10514 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
10515 systemd.journald.max_level_wall=.
10517 * "systemctl is-enabled --full" will now show by which symlinks a unit
10518 file is enabled in the unit dependency tree.
10520 * Support for VeraCrypt encrypted partitions has been added to the
10521 "cryptsetup" logic and /etc/crypttab.
10523 * systemd-detect-virt gained support for a new --private-users switch
10524 that checks whether the invoking processes are running inside a user
10525 namespace. Similar, a new special value "private-users" for the
10526 existing ConditionVirtualization= setting has been added, permitting
10527 skipping of specific units in user namespace environments.
10529 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
10530 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
10531 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
10532 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
10533 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
10534 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
10535 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
10536 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
10537 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
10538 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
10539 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
10540 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
10541 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
10542 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
10543 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
10544 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
10545 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
10546 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
10547 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
10548 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
10549 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
10550 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
10551 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
10552 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
10553 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
10554 Jędrzejewski-Szmek, Zeal Jagannatha
10556 — Santa Fe, 2016-11-03
10560 * In service units the various ExecXYZ= settings have been extended
10561 with an additional special character as first argument of the
10562 assigned value: if the character '+' is used the specified command
10563 line it will be run with full privileges, regardless of User=,
10564 Group=, CapabilityBoundingSet= and similar options. The effect is
10565 similar to the existing PermissionsStartOnly= option, but allows
10566 configuration of this concept for each executed command line
10569 * Services may now alter the service watchdog timeout at runtime by
10570 sending a WATCHDOG_USEC= message via sd_notify().
10572 * MemoryLimit= and related unit settings now optionally take percentage
10573 specifications. The percentage is taken relative to the amount of
10574 physical memory in the system (or in case of containers, the assigned
10575 amount of memory). This allows scaling service resources neatly with
10576 the amount of RAM available on the system. Similarly, systemd-logind's
10577 RuntimeDirectorySize= option now also optionally takes percentage
10580 * In similar fashion TasksMax= takes percentage values now, too. The
10581 value is taken relative to the configured maximum number of processes
10582 on the system. The per-service task maximum has been changed to 15%
10583 using this functionality. (Effectively this is an increase of 512 →
10584 4915 for service units, given the kernel's default pid_max setting.)
10586 * Calendar time specifications in .timer units now understand a ".."
10587 syntax for time ranges. Example: "4..7:10" may now be used for
10588 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
10591 * The InaccessableDirectories=, ReadOnlyDirectories= and
10592 ReadWriteDirectories= unit file settings have been renamed to
10593 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
10594 applied to all kinds of file nodes, and not just directories, with
10595 the exception of symlinks. Specifically these settings may now be
10596 used on block and character device nodes, UNIX sockets and FIFOS as
10597 well as regular files. The old names of these settings remain
10598 available for compatibility.
10600 * systemd will now log about all service processes it kills forcibly
10601 (using SIGKILL) because they remained after the clean shutdown phase
10602 of the service completed. This should help identifying services that
10603 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
10604 systemd-logind's configuration a similar log message is generated for
10605 processes killed at the end of each session due to this setting.
10607 * systemd will now set the $JOURNAL_STREAM environment variable for all
10608 services whose stdout/stderr are connected to the Journal (which
10609 effectively means by default: all services). The variable contains
10610 the device and inode number of the file descriptor used for
10611 stdout/stderr. This may be used by invoked programs to detect whether
10612 their stdout/stderr is connected to the Journal, in which case they
10613 can switch over to direct Journal communication, thus being able to
10614 pass extended, structured metadata along with their log messages. As
10615 one example, this is now used by glib's logging primitives.
10617 * When using systemd's default tmp.mount unit for /tmp, the mount point
10618 will now be established with the "nosuid" and "nodev" options. This
10619 avoids privilege escalation attacks that put traps and exploits into
10620 /tmp. However, this might cause problems if you e.g. put container
10621 images or overlays into /tmp; if you need this, override tmp.mount's
10622 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
10625 * systemd now supports the "memory" cgroup controller also on
10628 * The systemd-cgtop tool now optionally takes a control group path as
10629 command line argument. If specified, the control group list shown is
10630 limited to subgroups of that group.
10632 * The SystemCallFilter= unit file setting gained support for
10633 pre-defined, named system call filter sets. For example
10634 SystemCallFilter=@clock is now an effective way to make all clock
10635 changing-related system calls unavailable to a service. A number of
10636 similar pre-defined groups are defined. Writing system call filters
10637 for system services is simplified substantially with this new
10638 concept. Accordingly, all of systemd's own, long-running services now
10639 enable system call filtering based on this, by default.
10641 * A new service setting MemoryDenyWriteExecute= has been added, taking
10642 a boolean value. If turned on, a service may no longer create memory
10643 mappings that are writable and executable at the same time. This
10644 enhances security for services where this is enabled as it becomes
10645 harder to dynamically write and then execute memory in exploited
10646 service processes. This option has been enabled for all of systemd's
10647 own long-running services.
10649 * A new RestrictRealtime= service setting has been added, taking a
10650 boolean argument. If set the service's processes may no longer
10651 acquire realtime scheduling. This improves security as realtime
10652 scheduling may otherwise be used to easily freeze the system.
10654 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
10655 value. This may be used for requesting that the system manager inside
10656 of the container reports start-up completion to nspawn which then
10657 propagates this notification further to the service manager
10658 supervising nspawn itself. A related option NotifyReady= in .nspawn
10659 files has been added too. This functionality allows ordering of the
10660 start-up of multiple containers using the usual systemd ordering
10663 * machinectl gained a new command "stop" that is an alias for
10666 * systemd-resolved gained support for contacting DNS servers on
10667 link-local IPv6 addresses.
10669 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
10670 its caches. A method call for requesting the same operation has been
10671 added to the bus API too, and is made available via "systemd-resolve
10674 * systemd-resolve gained a new --status switch. If passed a brief
10675 summary of the used DNS configuration with per-interface information
10678 * resolved.conf gained a new Cache= boolean option, defaulting to
10679 on. If turned off local DNS caching is disabled. This comes with a
10680 performance penalty in particular when DNSSEC is enabled. Note that
10681 resolved disables its internal caching implicitly anyway, when the
10682 configured DNS server is on a host-local IP address such as ::1 or
10683 127.0.0.1, thus automatically avoiding double local caching.
10685 * systemd-resolved now listens on the local IP address 127.0.0.53:53
10686 for DNS requests. This improves compatibility with local programs
10687 that do not use the libc NSS or systemd-resolved's bus APIs for name
10688 resolution. This minimal DNS service is only available to local
10689 programs and does not implement the full DNS protocol, but enough to
10690 cover local DNS clients. A new, static resolv.conf file, listing just
10691 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
10692 now recommended to make /etc/resolv.conf a symlink to this file in
10693 order to route all DNS lookups to systemd-resolved, regardless if
10694 done via NSS, the bus API or raw DNS packets. Note that this local
10695 DNS service is not as fully featured as the libc NSS or
10696 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
10697 used to deliver link-local address information (as this implies
10698 sending a local interface index along), LLMNR/mDNS support via this
10699 interface is severely restricted. It is thus strongly recommended for
10700 all applications to use the libc NSS API or native systemd-resolved
10703 * systemd-networkd's bridge support learned a new setting
10704 VLANFiltering= for controlling VLAN filtering. Moreover a new section
10705 in .network files has been added for configuring VLAN bridging in
10706 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
10708 * systemd-networkd's IPv6 Router Advertisement code now makes use of
10709 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
10710 now be acquired without relying on DHCPv6. Two new options
10711 UseDomains= and UseDNS= have been added to configure this behaviour.
10713 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
10714 renamed IPv6AcceptRA=, without altering its behaviour. The old
10715 setting name remains available for compatibility reasons.
10717 * The systemd-networkd VTI/VTI6 tunneling support gained new options
10718 Key=, InputKey= and OutputKey=.
10720 * systemd-networkd gained support for VRF ("Virtual Routing Function")
10721 interface configuration.
10723 * "systemctl edit" may now be used to create new unit files by
10724 specifying the --force switch.
10726 * sd-event gained a new function sd_event_get_iteration() for
10727 requesting the current iteration counter of the event loop. It starts
10728 at zero and is increased by one with each event loop iteration.
10730 * A new rpm macro %systemd_ordering is provided by the macros.systemd
10731 file. It can be used in lieu of %systemd_requires in packages which
10732 don't use any systemd functionality and are intended to be installed
10733 in minimal containers without systemd present. This macro provides
10734 ordering dependencies to ensure that if the package is installed in
10735 the same rpm transaction as systemd, systemd will be installed before
10736 the scriptlets for the package are executed, allowing unit presets
10739 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
10740 been added to simplify packaging of generators.
10742 * The os-release file gained VERSION_CODENAME field for the
10743 distribution nickname (e.g. VERSION_CODENAME=woody).
10745 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
10746 can be set to disable parsing of metadata and the creation
10747 of persistent symlinks for that device.
10749 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
10750 to make them available to logged-in users has been reverted.
10752 * Much of the common code of the various systemd components is now
10753 built into an internal shared library libsystemd-shared-231.so
10754 (incorporating the systemd version number in the name, to be updated
10755 with future releases) that the components link to. This should
10756 decrease systemd footprint both in memory during runtime and on
10757 disk. Note that the shared library is not for public use, and is
10758 neither API nor ABI stable, but is likely to change with every new
10759 released update. Packagers need to make sure that binaries
10760 linking to libsystemd-shared.so are updated in step with the
10763 * Configuration for "mkosi" is now part of the systemd
10764 repository. mkosi is a tool to easily build legacy-free OS images,
10765 and is available on github: https://github.com/systemd/mkosi. If
10766 "mkosi" is invoked in the build tree a new raw OS image is generated
10767 incorporating the systemd sources currently being worked on and a
10768 clean, fresh distribution installation. The generated OS image may be
10769 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
10770 UEFI PC. This functionality is particularly useful to easily test
10771 local changes made to systemd in a pristine, defined environment. See
10772 doc/HACKING for details.
10774 * configure learned the --with-support-url= option to specify the
10775 distribution's bugtracker.
10777 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
10778 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
10779 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
10780 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
10781 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
10782 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
10783 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
10784 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
10785 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
10786 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
10787 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
10788 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
10789 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
10790 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
10791 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
10792 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
10793 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
10794 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
10795 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
10797 — Berlin, 2016-07-25
10801 * DNSSEC is now turned on by default in systemd-resolved (in
10802 "allow-downgrade" mode), but may be turned off during compile time by
10803 passing "--with-default-dnssec=no" to "configure" (and of course,
10804 during runtime with DNSSEC= in resolved.conf). We recommend
10805 downstreams to leave this on at least during development cycles and
10806 report any issues with the DNSSEC logic upstream. We are very
10807 interested in collecting feedback about the DNSSEC validator and its
10808 limitations in the wild. Note however, that DNSSEC support is
10809 probably nothing downstreams should turn on in stable distros just
10810 yet, as it might create incompatibilities with a few DNS servers and
10811 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
10812 automatically whenever we detect such incompatible setups, but there
10813 might be systems we do not cover yet. Hence: please help us testing
10814 the DNSSEC code, leave this on where you can, report back, but then
10815 again don't consider turning this on in your stable, LTS or
10816 production release just yet. (Note that you have to enable
10817 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
10818 and its DNSSEC mode for hostname resolution from local
10821 * systemd-resolve conveniently resolves DANE records with the --tlsa
10822 option and OPENPGPKEY records with the --openpgp option. It also
10823 supports dumping raw DNS record data via the new --raw= switch.
10825 * systemd-logind will now by default terminate user processes that are
10826 part of the user session scope unit (session-XX.scope) when the user
10827 logs out. This behavior is controlled by the KillUserProcesses=
10828 setting in logind.conf, and the previous default of "no" is now
10829 changed to "yes". This means that user sessions will be properly
10830 cleaned up after, but additional steps are necessary to allow
10831 intentionally long-running processes to survive logout.
10833 While the user is logged in at least once, user@.service is running,
10834 and any service that should survive the end of any individual login
10835 session can be started at a user service or scope using systemd-run.
10836 systemd-run(1) man page has been extended with an example which shows
10837 how to run screen in a scope unit underneath user@.service. The same
10838 command works for tmux.
10840 After the user logs out of all sessions, user@.service will be
10841 terminated too, by default, unless the user has "lingering" enabled.
10842 To effectively allow users to run long-term tasks even if they are
10843 logged out, lingering must be enabled for them. See loginctl(1) for
10844 details. The default polkit policy was modified to allow users to
10845 set lingering for themselves without authentication.
10847 Previous defaults can be restored at compile time by the
10848 --without-kill-user-processes option to "configure".
10850 * systemd-logind gained new configuration settings SessionsMax= and
10851 InhibitorsMax=, both with a default of 8192. It will not register new
10852 user sessions or inhibitors above this limit.
10854 * systemd-logind will now reload configuration on SIGHUP.
10856 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
10857 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
10858 enable. Also, support for the "io" cgroup controller in the unified
10859 hierarchy has been added, so that the "memory", "pids" and "io" are
10860 now the controllers that are supported on the unified hierarchy.
10862 WARNING: it is not possible to use previous systemd versions with
10863 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
10864 is necessary to also update systemd in the initramfs if using the
10865 unified hierarchy. An updated SELinux policy is also required.
10867 * LLDP support has been extended, and both passive (receive-only) and
10868 active (sender) modes are supported. Passive mode ("routers-only") is
10869 enabled by default in systemd-networkd. Active LLDP mode is enabled
10870 by default for containers on the internal network. The "networkctl
10871 lldp" command may be used to list information gathered. "networkctl
10872 status" will also show basic LLDP information on connected peers now.
10874 * The IAID and DUID unique identifier sent in DHCP requests may now be
10875 configured for the system and each .network file managed by
10876 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
10878 * systemd-networkd gained support for configuring proxy ARP support for
10879 each interface, via the ProxyArp= setting in .network files. It also
10880 gained support for configuring the multicast querier feature of
10881 bridge devices, via the new MulticastQuerier= setting in .netdev
10882 files. Similarly, snooping on the IGMP traffic can be controlled
10883 via the new setting MulticastSnooping=.
10885 A new setting PreferredLifetime= has been added for addresses
10886 configured in .network file to configure the lifetime intended for an
10889 The systemd-networkd DHCP server gained the option EmitRouter=, which
10890 defaults to yes, to configure whether the DHCP Option 3 (Router)
10893 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
10894 systemd-socket-activate and installed into /usr/bin. It is now fully
10897 * systemd-journald now uses separate threads to flush changes to disk
10898 when closing journal files, thus reducing impact of slow disk I/O on
10899 logging performance.
10901 * The sd-journal API gained two new calls
10902 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
10903 can be used to open journal files using file descriptors instead of
10904 file or directory paths. sd_journal_open_container() has been
10905 deprecated, sd_journal_open_directory_fd() should be used instead
10906 with the flag SD_JOURNAL_OS_ROOT.
10908 * journalctl learned a new output mode "-o short-unix" that outputs log
10909 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
10910 UTC). It also gained support for a new --no-hostname setting to
10911 suppress the hostname column in the family of "short" output modes.
10913 * systemd-ask-password now optionally skips printing of the password to
10914 stdout with --no-output which can be useful in scripts.
10916 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
10917 (devices tagged with ID_MAKER_TOOL) are now tagged with
10918 "uaccess" and are available to logged in users.
10920 * The DeviceAllow= unit setting now supports specifiers (with "%").
10922 * "systemctl show" gained a new --value switch, which allows print a
10923 only the contents of a specific unit property, without also printing
10924 the property's name. Similar support was added to "show*" verbs
10925 of loginctl and machinectl that output "key=value" lists.
10927 * A new unit type "generated" was added for files dynamically generated
10928 by generator tools. Similarly, a new unit type "transient" is used
10929 for unit files created using the runtime API. "systemctl enable" will
10930 refuse to operate on such files.
10932 * A new command "systemctl revert" has been added that may be used to
10933 revert to the vendor version of a unit file, in case local changes
10934 have been made by adding drop-ins or overriding the unit file.
10936 * "machinectl clean" gained a new verb to automatically remove all or
10937 just hidden container images.
10939 * systemd-tmpfiles gained support for a new line type "e" for emptying
10940 directories, if they exist, without creating them if they don't.
10942 * systemd-nspawn gained support for automatically patching the UID/GIDs
10943 of the owners and the ACLs of all files and directories in a
10944 container tree to match the UID/GID user namespacing range selected
10945 for the container invocation. This mode is enabled via the new
10946 --private-users-chown switch. It also gained support for
10947 automatically choosing a free, previously unused UID/GID range when
10948 starting a container, via the new --private-users=pick setting (which
10949 implies --private-users-chown). Together, these options for the first
10950 time make user namespacing for nspawn containers fully automatic and
10951 thus deployable. The systemd-nspawn@.service template unit file has
10952 been changed to use this functionality by default.
10954 * systemd-nspawn gained a new --network-zone= switch, that allows
10955 creating ad-hoc virtual Ethernet links between multiple containers,
10956 that only exist as long as at least one container referencing them is
10957 running. This allows easy connecting of multiple containers with a
10958 common link that implements an Ethernet broadcast domain. Each of
10959 these network "zones" may be named relatively freely by the user, and
10960 may be referenced by any number of containers, but each container may
10961 only reference one of these "zones". On the lower level, this is
10962 implemented by an automatically managed bridge network interface for
10963 each zone, that is created when the first container referencing its
10964 zone is created and removed when the last one referencing its zone
10967 * The default start timeout may now be configured on the kernel command
10968 line via systemd.default_timeout_start_sec=. It was already
10969 configurable via the DefaultTimeoutStartSec= option in
10970 /etc/systemd/system.conf.
10972 * Socket units gained a new TriggerLimitIntervalSec= and
10973 TriggerLimitBurst= setting to configure a limit on the activation
10974 rate of the socket unit.
10976 * The LimitNICE= setting now optionally takes normal UNIX nice values
10977 in addition to the raw integer limit value. If the specified
10978 parameter is prefixed with "+" or "-" and is in the range -20…19 the
10979 value is understood as UNIX nice value. If not prefixed like this it
10980 is understood as raw RLIMIT_NICE limit.
10982 * Note that the effect of the PrivateDevices= unit file setting changed
10983 slightly with this release: the per-device /dev file system will be
10984 mounted read-only from this version on, and will have "noexec"
10985 set. This (minor) change of behavior might cause some (exceptional)
10986 legacy software to break, when PrivateDevices=yes is set for its
10987 service. Please leave PrivateDevices= off if you run into problems
10990 * systemd-bootchart has been split out to a separate repository:
10991 https://github.com/systemd/systemd-bootchart
10993 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
10994 merged into the kernel in its current form.
10996 * The compatibility libraries libsystemd-daemon.so,
10997 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
10998 which have been deprecated since systemd-209 have been removed along
10999 with the corresponding pkg-config files. All symbols provided by
11000 those libraries are provided by libsystemd.so.
11002 * The Capabilities= unit file setting has been removed (it is ignored
11003 for backwards compatibility). AmbientCapabilities= and
11004 CapabilityBoundingSet= should be used instead.
11006 * A new special target has been added, initrd-root-device.target,
11007 which creates a synchronization point for dependencies of the root
11008 device in early userspace. Initramfs builders must ensure that this
11009 target is now included in early userspace.
11011 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
11012 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
11013 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
11014 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
11015 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
11016 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
11017 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
11018 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
11019 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
11020 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
11021 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
11022 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
11023 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
11024 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
11025 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
11026 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
11027 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
11028 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
11029 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
11030 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
11031 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
11032 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
11033 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
11034 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
11035 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
11038 — Fairfax, 2016-05-21
11042 * The systemd-resolved DNS resolver service has gained a substantial
11043 set of new features, most prominently it may now act as a DNSSEC
11044 validating stub resolver. DNSSEC mode is currently turned off by
11045 default, but is expected to be turned on by default in one of the
11046 next releases. For now, we invite everybody to test the DNSSEC logic
11047 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
11048 service also gained a full set of D-Bus interfaces, including calls
11049 to configure DNS and DNSSEC settings per link (for use by external
11050 network management software). systemd-resolved and systemd-networkd
11051 now distinguish between "search" and "routing" domains. The former
11052 are used to qualify single-label names, the latter are used purely
11053 for routing lookups within certain domains to specific links.
11054 resolved now also synthesizes RRs for all entries from /etc/hosts.
11056 * The systemd-resolve tool (which is a client utility for
11057 systemd-resolved) has been improved considerably and is now fully
11058 supported and documented. Hence it has moved from /usr/lib/systemd to
11061 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
11064 * The coredump collection logic has been reworked: when a coredump is
11065 collected it is now written to disk, compressed and processed
11066 (including stacktrace extraction) from a new instantiated service
11067 systemd-coredump@.service, instead of directly from the
11068 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
11069 processing large coredumps can take up a substantial amount of
11070 resources and time, and this previously happened entirely outside of
11071 systemd's service supervision. With the new logic the core_pattern
11072 hook only does minimal metadata collection before passing off control
11073 to the new instantiated service, which is configured with a time
11074 limit, a nice level and other settings to minimize negative impact on
11075 the rest of the system. Also note that the new logic will honour the
11076 RLIMIT_CORE setting of the crashed process, which now allows users
11077 and processes to turn off coredumping for their processes by setting
11080 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
11081 and all forked processes by default. Previously, PID 1 would leave
11082 the setting at "0" for all processes, as set by the kernel. Note that
11083 the resource limit traditionally has no effect on the generated
11084 coredumps on the system if the /proc/sys/kernel/core_pattern hook
11085 logic is used. Since the limit is now honoured (see above) its
11086 default has been changed so that the coredumping logic is enabled by
11087 default for all processes, while allowing specific opt-out.
11089 * When the stacktrace is extracted from processes of system users, this
11090 is now done as "systemd-coredump" user, in order to sandbox this
11091 potentially security sensitive parsing operation. (Note that when
11092 processing coredumps of normal users this is done under the user ID
11093 of process that crashed, as before.) Packagers should take notice
11094 that it is now necessary to create the "systemd-coredump" system user
11095 and group at package installation time.
11097 * The systemd-activate socket activation testing tool gained support
11098 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
11099 and --seqpacket switches. It also has been extended to support both
11100 new-style and inetd-style file descriptor passing. Use the new
11101 --inetd switch to request inetd-style file descriptor passing.
11103 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
11104 variable, which takes a boolean value. If set to false, ANSI color
11105 output is disabled in the tools even when run on a terminal that
11108 * The VXLAN support in networkd now supports two new settings
11109 DestinationPort= and PortRange=.
11111 * A new systemd.machine_id= kernel command line switch has been added,
11112 that may be used to set the machine ID in /etc/machine-id if it is
11113 not initialized yet. This command line option has no effect if the
11114 file is already initialized.
11116 * systemd-nspawn gained a new --as-pid2 switch that invokes any
11117 specified command line as PID 2 rather than PID 1 in the
11118 container. In this mode PID 1 is a minimal stub init process that
11119 implements the special POSIX and Linux semantics of PID 1 regarding
11120 signal and child process management. Note that this stub init process
11121 is implemented in nspawn itself and requires no support from the
11122 container image. This new logic is useful to support running
11123 arbitrary commands in the container, as normal processes are
11124 generally not prepared to run as PID 1.
11126 * systemd-nspawn gained a new --chdir= switch for setting the current
11127 working directory for the process started in the container.
11129 * "journalctl /dev/sda" will now output all kernel log messages for
11130 specified device from the current boot, in addition to all devices
11131 that are parents of it. This should make log output about devices
11132 pretty useful, as long as kernel drivers attach enough metadata to
11133 the log messages. (The usual SATA drivers do.)
11135 * The sd-journal API gained two new calls
11136 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
11137 that report whether log data from /run or /var has been found.
11139 * journalctl gained a new switch "--fields" that prints all journal
11140 record field names currently in use in the journal. This is backed
11141 by two new sd-journal API calls sd_journal_enumerate_fields() and
11142 sd_journal_restart_fields().
11144 * Most configurable timeouts in systemd now expect an argument of
11145 "infinity" to turn them off, instead of "0" as before. The semantics
11146 from now on is that a timeout of "0" means "now", and "infinity"
11147 means "never". To maintain backwards compatibility, "0" continues to
11148 turn off previously existing timeout settings.
11150 * "systemctl reload-or-try-restart" has been renamed to "systemctl
11151 try-reload-or-restart" to clarify what it actually does: the "try"
11152 logic applies to both reloading and restarting, not just restarting.
11153 The old name continues to be accepted for compatibility.
11155 * On boot-up, when PID 1 detects that the system clock is behind the
11156 release date of the systemd version in use, the clock is now set
11157 to the latter. Previously, this was already done in timesyncd, in order
11158 to avoid running with clocks set to the various clock epochs such as
11159 1902, 1938 or 1970. With this change the logic is now done in PID 1
11160 in addition to timesyncd during early boot-up, so that it is enforced
11161 before the first process is spawned by systemd. Note that the logic
11162 in timesyncd remains, as it is more comprehensive and ensures
11163 clock monotonicity by maintaining a persistent timestamp file in
11164 /var. Since /var is generally not available in earliest boot or the
11165 initrd, this part of the logic remains in timesyncd, and is not done
11168 * Support for tweaking details in net_cls.class_id through the
11169 NetClass= configuration directive has been removed, as the kernel
11170 people have decided to deprecate that controller in cgroup v2.
11171 Userspace tools such as nftables are moving over to setting rules
11172 that are specific to the full cgroup path of a task, which obsoletes
11173 these controllers anyway. The NetClass= directive is kept around for
11174 legacy compatibility reasons. For a more in-depth description of the
11175 kernel change, please refer to the respective upstream commit:
11177 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
11179 * A new service setting RuntimeMaxSec= has been added that may be used
11180 to specify a maximum runtime for a service. If the timeout is hit, the
11181 service is terminated and put into a failure state.
11183 * A new service setting AmbientCapabilities= has been added. It allows
11184 configuration of additional Linux process capabilities that are
11185 passed to the activated processes. This is only available on very
11188 * The process resource limit settings in service units may now be used
11189 to configure hard and soft limits individually.
11191 * The various libsystemd APIs such as sd-bus or sd-event now publicly
11192 expose support for gcc's __attribute__((cleanup())) C extension.
11193 Specifically, for many object destructor functions alternative
11194 versions have been added that have names suffixed with "p" and take a
11195 pointer to a pointer to the object to destroy, instead of just a
11196 pointer to the object itself. This is useful because these destructor
11197 functions may be used directly as parameters to the cleanup
11198 construct. Internally, systemd has been a heavy user of this GCC
11199 extension for a long time, and with this change similar support is
11200 now available to consumers of the library outside of systemd. Note
11201 that by using this extension in your sources compatibility with old
11202 and strictly ANSI compatible C compilers is lost. However, all gcc or
11203 LLVM versions of recent years support this extension.
11205 * Timer units gained support for a new setting RandomizedDelaySec= that
11206 allows configuring some additional randomized delay to the configured
11207 time. This is useful to spread out timer events to avoid load peaks in
11208 clusters or larger setups.
11210 * Calendar time specifications now support sub-second accuracy.
11212 * Socket units now support listening on SCTP and UDP-lite protocol
11215 * The sd-event API now comes with a full set of man pages.
11217 * Older versions of systemd contained experimental support for
11218 compressing journal files and coredumps with the LZ4 compressor that
11219 was not compatible with the lz4 binary (due to API limitations of the
11220 lz4 library). This support has been removed; only support for files
11221 compatible with the lz4 binary remains. This LZ4 logic is now
11222 officially supported and no longer considered experimental.
11224 * The dkr image import logic has been removed again from importd. dkr's
11225 micro-services focus doesn't fit into the machine image focus of
11226 importd, and quickly got out of date with the upstream dkr API.
11228 * Creation of the /run/lock/lockdev/ directory was dropped from
11229 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
11230 been available for many years. If you still need this, you need to
11231 create your own tmpfiles.d config file with:
11233 d /run/lock/lockdev 0775 root lock -
11235 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
11236 and RebootArgument= have been moved from the [Service] section of
11237 unit files to [Unit], and they are now supported on all unit types,
11238 not just service units. Of course, systemd will continue to
11239 understand these settings also at the old location, in order to
11240 maintain compatibility.
11242 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
11243 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
11244 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
11245 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
11246 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
11247 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
11248 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
11249 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
11250 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
11251 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
11252 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
11253 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
11254 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
11255 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
11256 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
11257 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
11258 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
11259 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
11260 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
11262 — Berlin, 2016-02-11
11266 * A number of properties previously only settable in unit
11267 files are now also available as properties to set when
11268 creating transient units programmatically via the bus, as it
11269 is exposed with systemd-run's --property=
11270 setting. Specifically, these are: SyslogIdentifier=,
11271 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
11272 EnvironmentFile=, ReadWriteDirectories=,
11273 ReadOnlyDirectories=, InaccessibleDirectories=,
11274 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
11276 * When creating transient services via the bus API it is now
11277 possible to pass in a set of file descriptors to use as
11278 STDIN/STDOUT/STDERR for the invoked process.
11280 * Slice units may now be created transiently via the bus APIs,
11281 similar to the way service and scope units may already be
11282 created transiently.
11284 * Wherever systemd expects a calendar timestamp specification
11285 (like in journalctl's --since= and --until= switches) UTC
11286 timestamps are now supported. Timestamps suffixed with "UTC"
11287 are now considered to be in Universal Time Coordinated
11288 instead of the local timezone. Also, timestamps may now
11289 optionally be specified with sub-second accuracy. Both of
11290 these additions also apply to recurring calendar event
11291 specification, such as OnCalendar= in timer units.
11293 * journalctl gained a new "--sync" switch that asks the
11294 journal daemon to write all so far unwritten log messages to
11295 disk and sync the files, before returning.
11297 * systemd-tmpfiles learned two new line types "q" and "Q" that
11298 operate like "v", but also set up a basic btrfs quota
11299 hierarchy when used on a btrfs file system with quota
11302 * tmpfiles' "v", "q" and "Q" will now create a plain directory
11303 instead of a subvolume (even on a btrfs file system) if the
11304 root directory is a plain directory, and not a
11305 subvolume. This should simplify things with certain chroot()
11306 environments which are not aware of the concept of btrfs
11309 * systemd-detect-virt gained a new --chroot switch to detect
11310 whether execution takes place in a chroot() environment.
11312 * CPUAffinity= now takes CPU index ranges in addition to
11313 individual indexes.
11315 * The various memory-related resource limit settings (such as
11316 LimitAS=) now understand the usual K, M, G, … suffixes to
11317 the base of 1024 (IEC). Similar, the time-related resource
11318 limit settings understand the usual min, h, day, … suffixes
11321 * There's a new system.conf setting DefaultTasksMax= to
11322 control the default TasksMax= setting for services and
11323 scopes running on the system. (TasksMax= is the primary
11324 setting that exposes the "pids" cgroup controller on systemd
11325 and was introduced in the previous systemd release.) The
11326 setting now defaults to 512, which means services that are
11327 not explicitly configured otherwise will only be able to
11328 create 512 processes or threads at maximum, from this
11329 version on. Note that this means that thread- or
11330 process-heavy services might need to be reconfigured to set
11331 TasksMax= to a higher value. It is sufficient to set
11332 TasksMax= in these specific unit files to a higher value, or
11333 even "infinity". Similar, there's now a logind.conf setting
11334 UserTasksMax= that defaults to 4096 and limits the total
11335 number of processes or tasks each user may own
11336 concurrently. nspawn containers also have the TasksMax=
11337 value set by default now, to 8192. Note that all of this
11338 only has an effect if the "pids" cgroup controller is
11339 enabled in the kernel. The general benefit of these changes
11340 should be a more robust and safer system, that provides a
11341 certain amount of per-service fork() bomb protection.
11343 * systemd-nspawn gained the new --network-veth-extra= switch
11344 to define additional and arbitrarily-named virtual Ethernet
11345 links between the host and the container.
11347 * A new service execution setting PassEnvironment= has been
11348 added that allows importing select environment variables
11349 from PID1's environment block into the environment block of
11352 * Timer units gained support for a new RemainAfterElapse=
11353 setting which takes a boolean argument. It defaults to on,
11354 exposing behaviour unchanged to previous releases. If set to
11355 off, timer units are unloaded after they elapsed if they
11356 cannot elapse again. This is particularly useful for
11357 transient timer units, which shall not stay around longer
11358 than until they first elapse.
11360 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
11361 default now (the kernel default is 16). This is beneficial
11362 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
11363 allows substantially larger numbers of queued
11364 datagrams. This should increase the capability of systemd to
11365 parallelize boot-up, as logging and sd_notify() are unlikely
11366 to stall execution anymore. If you need to change the value
11367 from the new defaults, use the usual sysctl.d/ snippets.
11369 * The compression framing format used by the journal or
11370 coredump processing has changed to be in line with what the
11371 official LZ4 tools generate. LZ4 compression support in
11372 systemd was considered unsupported previously, as the format
11373 was not compatible with the normal tools. With this release
11374 this has changed now, and it is hence safe for downstream
11375 distributions to turn it on. While not compressing as well
11376 as the XZ, LZ4 is substantially faster, which makes
11377 it a good default choice for the compression logic in the
11378 journal and in coredump handling.
11380 * Any reference to /etc/mtab has been dropped from
11381 systemd. The file has been obsolete since a while, but
11382 systemd refused to work on systems where it was incorrectly
11383 set up (it should be a symlink or non-existent). Please make
11384 sure to update to util-linux 2.27.1 or newer in conjunction
11385 with this systemd release, which also drops any reference to
11386 /etc/mtab. If you maintain a distribution make sure that no
11387 software you package still references it, as this is a
11388 likely source of bugs. There's also a glibc bug pending,
11389 asking for removal of any reference to this obsolete file:
11391 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
11393 Note that only util-linux versions built with
11394 --enable-libmount-force-mountinfo are supported.
11396 * Support for the ".snapshot" unit type has been removed. This
11397 feature turned out to be little useful and little used, and
11398 has now been removed from the core and from systemctl.
11400 * The dependency types RequiresOverridable= and
11401 RequisiteOverridable= have been removed from systemd. They
11402 have been used only very sparingly to our knowledge and
11403 other options that provide a similar effect (such as
11404 systemctl --mode=ignore-dependencies) are much more useful
11405 and commonly used. Moreover, they were only half-way
11406 implemented as the option to control behaviour regarding
11407 these dependencies was never added to systemctl. By removing
11408 these dependency types the execution engine becomes a bit
11409 simpler. Unit files that use these dependencies should be
11410 changed to use the non-Overridable dependency types
11411 instead. In fact, when parsing unit files with these
11412 options, that's what systemd will automatically convert them
11413 too, but it will also warn, asking users to fix the unit
11414 files accordingly. Removal of these dependency types should
11415 only affect a negligible number of unit files in the wild.
11417 * Behaviour of networkd's IPForward= option changed
11418 (again). It will no longer maintain a per-interface setting,
11419 but propagate one way from interfaces where this is enabled
11420 to the global kernel setting. The global setting will be
11421 enabled when requested by a network that is set up, but
11422 never be disabled again. This change was made to make sure
11423 IPv4 and IPv6 behaviour regarding packet forwarding is
11424 similar (as the Linux IPv6 stack does not support
11425 per-interface control of this setting) and to minimize
11428 * In unit files the behaviour of %u, %U, %h, %s has
11429 changed. These specifiers will now unconditionally resolve
11430 to the various user database fields of the user that the
11431 systemd instance is running as, instead of the user
11432 configured in the specific unit via User=. Note that this
11433 effectively doesn't change much, as resolving of these
11434 specifiers was already turned off in the --system instance
11435 of systemd, as we cannot do NSS lookups from PID 1. In the
11436 --user instance of systemd these specifiers where correctly
11437 resolved, but hardly made any sense, since the user instance
11438 lacks privileges to do user switches anyway, and User= is
11439 hence useless. Moreover, even in the --user instance of
11440 systemd behaviour was awkward as it would only take settings
11441 from User= assignment placed before the specifier into
11442 account. In order to unify and simplify the logic around
11443 this the specifiers will now always resolve to the
11444 credentials of the user invoking the manager (which in case
11445 of PID 1 is the root user).
11447 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
11448 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
11449 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
11450 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
11451 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
11452 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
11453 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
11454 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
11455 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
11456 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
11457 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
11458 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
11459 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
11460 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
11463 — Berlin, 2015-11-18
11467 * systemd now depends on util-linux v2.27. More specifically,
11468 the newly added mount monitor feature in libmount now
11469 replaces systemd's former own implementation.
11471 * libmount mandates /etc/mtab not to be regular file, and
11472 systemd now enforces this condition at early boot.
11473 /etc/mtab has been deprecated and warned about for a very
11474 long time, so systems running systemd should already have
11475 stopped having this file around as anything else than a
11476 symlink to /proc/self/mounts.
11478 * Support for the "pids" cgroup controller has been added. It
11479 allows accounting the number of tasks in a cgroup and
11480 enforcing limits on it. This adds two new setting
11481 TasksAccounting= and TasksMax= to each unit, as well as a
11482 global option DefaultTasksAccounting=.
11484 * Support for the "net_cls" cgroup controller has been added.
11485 It allows assigning a net class ID to each task in the
11486 cgroup, which can then be used in firewall rules and traffic
11487 shaping configurations. Note that the kernel netfilter net
11488 class code does not currently work reliably for ingress
11489 packets on unestablished sockets.
11491 This adds a new config directive called NetClass= to CGroup
11492 enabled units. Allowed values are positive numbers for fixed
11493 assignments and "auto" for picking a free value
11496 * 'systemctl is-system-running' now returns 'offline' if the
11497 system is not booted with systemd. This command can now be
11498 used as a substitute for 'systemd-notify --booted'.
11500 * Watchdog timeouts have been increased to 3 minutes for all
11501 in-tree service files. Apparently, disk IO issues are more
11502 frequent than we hoped, and user reported >1 minute waiting
11505 * 'machine-id-commit' functionality has been merged into
11506 'machine-id-setup --commit'. The separate binary has been
11509 * The WorkingDirectory= directive in unit files may now be set
11510 to the special value '~'. In this case, the working
11511 directory is set to the home directory of the user
11512 configured in User=.
11514 * "machinectl shell" will now open the shell in the home
11515 directory of the selected user by default.
11517 * The CrashChVT= configuration file setting is renamed to
11518 CrashChangeVT=, following our usual logic of not
11519 abbreviating unnecessarily. The old directive is still
11520 supported for compat reasons. Also, this directive now takes
11521 an integer value between 1 and 63, or a boolean value. The
11522 formerly supported '-1' value for disabling stays around for
11525 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
11526 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
11527 RootDirectory= properties can now be set for transient
11530 * The systemd-analyze tool gained a new "set-log-target" verb
11531 to change the logging target the system manager logs to
11532 dynamically during runtime. This is similar to how
11533 "systemd-analyze set-log-level" already changes the log
11536 * In nspawn /sys is now mounted as tmpfs, with only a selected
11537 set of subdirectories mounted in from the real sysfs. This
11538 enhances security slightly, and is useful for ensuring user
11539 namespaces work correctly.
11541 * Support for USB FunctionFS activation has been added. This
11542 allows implementation of USB gadget services that are
11543 activated as soon as they are requested, so that they don't
11544 have to run continuously, similar to classic socket
11547 * The "systemctl exit" command now optionally takes an
11548 additional parameter that sets the exit code to return from
11549 the systemd manager when exiting. This is only relevant when
11550 running the systemd user instance, or when running the
11551 system instance in a container.
11553 * sd-bus gained the new API calls sd_bus_path_encode_many()
11554 and sd_bus_path_decode_many() that allow easy encoding and
11555 decoding of multiple identifier strings inside a D-Bus
11556 object path. Another new call sd_bus_default_flush_close()
11557 has been added to flush and close per-thread default
11560 * systemd-cgtop gained support for a -M/--machine= switch to
11561 show the control groups within a certain container only.
11563 * "systemctl kill" gained support for an optional --fail
11564 switch. If specified the requested operation will fail of no
11565 processes have been killed, because the unit had no
11566 processes attached, or similar.
11568 * A new systemd.crash_reboot=1 kernel command line option has
11569 been added that triggers a reboot after crashing. This can
11570 also be set through CrashReboot= in systemd.conf.
11572 * The RuntimeDirectory= setting now understands unit
11573 specifiers like %i or %f.
11575 * A new (still internal) library API sd-ipv4acd has been added,
11576 that implements address conflict detection for IPv4. It's
11577 based on code from sd-ipv4ll, and will be useful for
11578 detecting DHCP address conflicts.
11580 * File descriptors passed during socket activation may now be
11581 named. A new API sd_listen_fds_with_names() is added to
11582 access the names. The default names may be overridden,
11583 either in the .socket file using the FileDescriptorName=
11584 parameter, or by passing FDNAME= when storing the file
11585 descriptors using sd_notify().
11587 * systemd-networkd gained support for:
11589 - Setting the IPv6 Router Advertisement settings via
11590 IPv6AcceptRouterAdvertisements= in .network files.
11592 - Configuring the HelloTimeSec=, MaxAgeSec= and
11593 ForwardDelaySec= bridge parameters in .netdev files.
11595 - Configuring PreferredSource= for static routes in
11598 * The "ask-password" framework used to query for LUKS harddisk
11599 passwords or SSL passwords during boot gained support for
11600 caching passwords in the kernel keyring, if it is
11601 available. This makes sure that the user only has to type in
11602 a passphrase once if there are multiple objects to unlock
11603 with the same one. Previously, such password caching was
11604 available only when Plymouth was used; this moves the
11605 caching logic into the systemd codebase itself. The
11606 "systemd-ask-password" utility gained a new --keyname=
11607 switch to control which kernel keyring key to use for
11608 caching a password in. This functionality is also useful for
11609 enabling display managers such as gdm to automatically
11610 unlock the user's GNOME keyring if its passphrase, the
11611 user's password and the harddisk password are the same, if
11612 gdm-autologin is used.
11614 * When downloading tar or raw images using "machinectl
11615 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
11616 file is now also downloaded, if it is available and stored
11617 next to the image file.
11619 * Units of type ".socket" gained a new boolean setting
11620 Writable= which is only useful in conjunction with
11621 ListenSpecial=. If true, enables opening the specified
11622 special file in O_RDWR mode rather than O_RDONLY mode.
11624 * systemd-rfkill has been reworked to become a singleton
11625 service that is activated through /dev/rfkill on each rfkill
11626 state change and saves the settings to disk. This way,
11627 systemd-rfkill is now compatible with devices that exist
11628 only intermittendly, and even restores state if the previous
11629 system shutdown was abrupt rather than clean.
11631 * The journal daemon gained support for vacuuming old journal
11632 files controlled by the number of files that shall remain,
11633 in addition to the already existing control by size and by
11634 date. This is useful as journal interleaving performance
11635 degrades with too many separate journal files, and allows
11636 putting an effective limit on them. The new setting defaults
11637 to 100, but this may be changed by setting SystemMaxFiles=
11638 and RuntimeMaxFiles= in journald.conf. Also, the
11639 "journalctl" tool gained the new --vacuum-files= switch to
11640 manually vacuum journal files to leave only the specified
11641 number of files in place.
11643 * udev will now create /dev/disk/by-path links for ATA devices
11644 on kernels where that is supported.
11646 * Galician, Serbian, Turkish and Korean translations were added.
11648 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
11649 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
11650 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
11651 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
11652 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
11653 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
11654 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
11655 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
11656 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
11657 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
11658 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
11659 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
11660 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
11661 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
11662 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
11663 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
11664 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
11665 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
11667 — Berlin, 2015-10-07
11671 * The DHCP implementation of systemd-networkd gained a set of
11674 - The DHCP server now supports emitting DNS and NTP
11675 information. It may be enabled and configured via
11676 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
11677 and NTP information is enabled, but no servers are
11678 configured, the corresponding uplink information (if there
11679 is any) is propagated.
11681 - Server and client now support transmission and reception
11682 of timezone information. It can be configured via the
11683 newly introduced network options UseTimezone=,
11684 EmitTimezone=, and Timezone=. Transmission of timezone
11685 information is enabled between host and containers by
11686 default now: the container will change its local timezone
11687 to what the host has set.
11689 - Lease timeouts can now be configured via
11690 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
11692 - The DHCP server improved on the stability of
11693 leases. Clients are more likely to get the same lease
11694 information back, even if the server loses state.
11696 - The DHCP server supports two new configuration options to
11697 control the lease address pool metrics, PoolOffset= and
11700 * The encapsulation limit of tunnels in systemd-networkd may
11701 now be configured via 'EncapsulationLimit='. It allows
11702 modifying the maximum additional levels of encapsulation
11703 that are permitted to be prepended to a packet.
11705 * systemd now supports the concept of user buses replacing
11706 session buses, if used with dbus-1.10 (and enabled via dbus
11707 --enable-user-session). It previously only supported this on
11708 kdbus-enabled systems, and this release expands this to
11709 'dbus-daemon' systems.
11711 * systemd-networkd now supports predictable interface names
11712 for virtio devices.
11714 * systemd now optionally supports the new Linux kernel
11715 "unified" control group hierarchy. If enabled via the kernel
11716 command-line option 'systemd.unified_cgroup_hierarchy=1',
11717 systemd will try to mount the unified cgroup hierarchy
11718 directly on /sys/fs/cgroup. If not enabled, or not
11719 available, systemd will fall back to the legacy cgroup
11720 hierarchy setup, as before. Host system and containers can
11721 mix and match legacy and unified hierarchies as they
11722 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
11723 environment variable to individually select the hierarchy to
11724 use for executed containers. By default, nspawn will use the
11725 unified hierarchy for the containers if the host uses the
11726 unified hierarchy, and the legacy hierarchy otherwise.
11727 Please note that at this point the unified hierarchy is an
11728 experimental kernel feature and is likely to change in one
11729 of the next kernel releases. Therefore, it should not be
11730 enabled by default in downstream distributions yet. The
11731 minimum required kernel version for the unified hierarchy to
11732 work is 4.2. Note that when the unified hierarchy is used
11733 for the first time delegated access to controllers is
11734 safe. Because of this systemd-nspawn containers will get
11735 access to controllers now, as will systemd user
11736 sessions. This means containers and user sessions may now
11737 manage their own resources, partitioning up what the system
11740 * A new special scope unit "init.scope" has been introduced
11741 that encapsulates PID 1 of the system. It may be used to
11742 determine resource usage and enforce resource limits on PID
11743 1 itself. PID 1 hence moved out of the root of the control
11746 * The cgtop tool gained support for filtering out kernel
11747 threads when counting tasks in a control group. Also, the
11748 count of processes is now recursively summed up by
11749 default. Two options -k and --recursive= have been added to
11750 revert to old behaviour. The tool has also been updated to
11751 work correctly in containers now.
11753 * systemd-nspawn's --bind= and --bind-ro= options have been
11754 extended to allow creation of non-recursive bind mounts.
11756 * libsystemd gained two new calls sd_pid_get_cgroup() and
11757 sd_peer_get_cgroup() which return the control group path of
11758 a process or peer of a connected AF_UNIX socket. This
11759 function call is particularly useful when implementing
11760 delegated subtrees support in the control group hierarchy.
11762 * The "sd-event" event loop API of libsystemd now supports
11763 correct dequeuing of real-time signals, without losing
11766 * When systemd requests a polkit decision when managing units it
11767 will now add additional fields to the request, including unit
11768 name and desired operation. This enables more powerful polkit
11769 policies, that make decisions depending on these parameters.
11771 * nspawn learnt support for .nspawn settings files, that may
11772 accompany the image files or directories of containers, and
11773 may contain additional settings for the container. This is
11774 an alternative to configuring container parameters via the
11775 nspawn command line.
11777 Contributions from: Cristian Rodríguez, Daniel Mack, David
11778 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
11779 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
11780 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
11781 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
11782 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
11783 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
11784 Andersen, Tom Gundersen, Torstein Husebø
11786 — Berlin, 2015-09-08
11790 * machinectl gained a new verb 'shell' which opens a fresh
11791 shell on the target container or the host. It is similar to
11792 the existing 'login' command of machinectl, but spawns the
11793 shell directly without prompting for username or
11794 password. The pseudo machine '.host' now refers to the local
11795 host and is used by default. Hence, 'machinectl shell' can
11796 be used as replacement for 'su -' which spawns a session as
11797 a fresh systemd unit in a way that is fully isolated from
11798 the originating session.
11800 * systemd-networkd learned to cope with private-zone DHCP
11801 options and allows other programs to query the values.
11803 * SELinux access control when enabling/disabling units is no
11804 longer enforced with this release. The previous implementation
11805 was incorrect, and a new corrected implementation is not yet
11806 available. As unit file operations are still protected via
11807 polkit and D-Bus policy this is not a security problem. Yet,
11808 distributions which care about optimal SELinux support should
11809 probably not stabilize on this release.
11811 * sd-bus gained support for matches of type "arg0has=", that
11812 test for membership of strings in string arrays sent in bus
11815 * systemd-resolved now dumps the contents of its DNS and LLMNR
11816 caches to the logs on reception of the SIGUSR1 signal. This
11817 is useful to debug DNS behaviour.
11819 * The coredumpctl tool gained a new --directory= option to
11820 operate on journal files in a specific directory.
11822 * "systemctl reboot" and related commands gained a new
11823 "--message=" option which may be used to set a free-text
11824 wall message when shutting down or rebooting the
11825 system. This message is also logged, which is useful for
11826 figuring out the reason for a reboot or shutdown a
11829 * The "systemd-resolve-host" tool's -i switch now takes
11830 network interface numbers as alternative to interface names.
11832 * A new unit file setting for services has been introduced:
11833 UtmpMode= allows configuration of how precisely systemd
11834 handles utmp and wtmp entries for the service if this is
11835 enabled. This allows writing services that appear similar to
11836 user sessions in the output of the "w", "who", "last" and
11839 * systemd-resolved will now locally synthesize DNS resource
11840 records for the "localhost" and "gateway" domains as well as
11841 the local hostname. This should ensure that clients querying
11842 RRs via resolved will get similar results as those going via
11843 NSS, if nss-myhostname is enabled.
11845 Contributions from: Alastair Hughes, Alex Crawford, Daniel
11846 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
11847 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
11848 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
11849 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
11850 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
11851 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
11852 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
11853 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
11854 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
11855 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
11856 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11858 — Berlin, 2015-08-27
11862 * The systemd-efi-boot-generator functionality was merged into
11863 systemd-gpt-auto-generator.
11865 * systemd-networkd now supports Group Policy for vxlan
11866 devices. It can be enabled via the new boolean configuration
11867 option called 'GroupPolicyExtension='.
11869 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
11870 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
11871 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
11873 — Berlin, 2015-07-31
11877 * The python-systemd code has been removed from the systemd repository.
11878 A new repository has been created which accommodates the code from
11879 now on, and we kindly ask distributions to create a separate package
11880 for this: https://github.com/systemd/python-systemd
11882 * The systemd daemon will now reload its main configuration
11883 (/etc/systemd/system.conf) on daemon-reload.
11885 * sd-dhcp now exposes vendor specific extensions via
11886 sd_dhcp_lease_get_vendor_specific().
11888 * systemd-networkd gained a number of new configuration options.
11890 - A new boolean configuration option for TAP devices called
11891 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
11892 device, thus allowing to send and receive GSO packets.
11894 - A new tunnel configuration option called 'CopyDSCP='.
11895 If enabled, the DSCP field of ip6 tunnels is copied into the
11896 decapsulated packet.
11898 - A set of boolean bridge configuration options were added.
11899 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
11900 and 'UnicastFlood=' are now parsed by networkd and applied to the
11901 respective bridge link device via the respective IFLA_BRPORT_*
11904 - A new string configuration option to override the hostname sent
11905 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
11906 is true, networkd will use the configured hostname instead of the
11907 system hostname when sending DHCP requests.
11909 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
11910 networkd will configure the IPv6 flow-label of the tunnel device
11911 according to RFC2460.
11913 - The 'macvtap' virtual network devices are now supported, similar to
11914 the already supported 'macvlan' devices.
11916 * systemd-resolved now implements RFC5452 to improve resilience against
11917 cache poisoning. Additionally, source port randomization is enabled
11918 by default to further protect against DNS spoofing attacks.
11920 * nss-mymachines now supports translating UIDs and GIDs of running
11921 containers with user-namespaces enabled. If a container 'foo'
11922 translates a host uid 'UID' to the container uid 'TUID', then
11923 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
11924 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
11925 mapped as 'vg-foo-TGID'.
11927 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
11928 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
11929 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
11930 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
11931 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
11932 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
11933 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
11934 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
11935 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
11936 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
11938 — Berlin, 2015-07-29
11942 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
11943 There are no known issues with current sysfs, and udev does not need
11944 or should be used to work around such bugs.
11946 * udev does no longer enable USB HID power management. Several reports
11947 indicate, that some devices cannot handle that setting.
11949 * The udev accelerometer helper was removed. The functionality
11950 is now fully included in iio-sensor-proxy. But this means,
11951 older iio-sensor-proxy versions will no longer provide
11952 accelerometer/orientation data with this systemd version.
11953 Please upgrade iio-sensor-proxy to version 1.0.
11955 * networkd gained a new configuration option IPv6PrivacyExtensions=
11956 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
11957 for Stateless Address") on selected networks.
11959 * For the sake of fewer build-time dependencies and less code in the
11960 main repository, the python bindings are about to be removed in the
11961 next release. A new repository has been created which accommodates
11962 the code from now on, and we kindly ask distributions to create a
11963 separate package for this. The removal will take place in v223.
11965 https://github.com/systemd/python-systemd
11967 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
11968 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
11969 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
11970 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
11971 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
11972 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
11973 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
11974 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
11975 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
11976 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11978 — Berlin, 2015-07-07
11982 * The sd-bus.h and sd-event.h APIs have now been declared
11983 stable and have been added to the official interface of
11984 libsystemd.so. sd-bus implements an alternative D-Bus client
11985 library, that is relatively easy to use, very efficient and
11986 supports both classic D-Bus as well as kdbus as transport
11987 backend. sd-event is a generic event loop abstraction that
11988 is built around Linux epoll, but adds features such as event
11989 prioritization or efficient timer handling. Both APIs are good
11990 choices for C programs looking for a bus and/or event loop
11991 implementation that is minimal and does not have to be
11992 portable to other kernels.
11994 * kdbus support is no longer compile-time optional. It is now
11995 always built-in. However, it can still be disabled at
11996 runtime using the kdbus=0 kernel command line setting, and
11997 that setting may be changed to default to off, by specifying
11998 --disable-kdbus at build-time. Note though that the kernel
11999 command line setting has no effect if the kdbus.ko kernel
12000 module is not installed, in which case kdbus is (obviously)
12001 also disabled. We encourage all downstream distributions to
12002 begin testing kdbus by adding it to the kernel images in the
12003 development distributions, and leaving kdbus support in
12006 * The minimal required util-linux version has been bumped to
12009 * Support for chkconfig (--enable-chkconfig) was removed in
12010 favor of calling an abstraction tool
12011 /lib/systemd/systemd-sysv-install. This needs to be
12012 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
12013 in README for details.
12015 * If there's a systemd unit and a SysV init script for the
12016 same service name, and the user executes "systemctl enable"
12017 for it (or a related call), then this will now enable both
12018 (or execute the related operation on both), not just the
12021 * The libudev API documentation has been converted from gtkdoc
12024 * gudev has been removed from the systemd tree, it is now an
12027 * The systemd-cgtop tool learnt a new --raw switch to generate
12028 "raw" (machine parsable) output.
12030 * networkd's IPForwarding= .network file setting learnt the
12031 new setting "kernel", which ensures that networkd does not
12032 change the IP forwarding sysctl from the default kernel
12035 * The systemd-logind bus API now exposes a new boolean
12036 property "Docked" that reports whether logind considers the
12037 system "docked", i.e. connected to a docking station or not.
12039 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
12040 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
12041 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
12042 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
12043 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
12044 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
12045 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
12046 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
12047 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
12048 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
12049 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
12050 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
12051 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
12052 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12053 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
12054 Fink, Zbigniew Jędrzejewski-Szmek
12056 — Berlin, 2015-06-19
12060 * The gudev library has been extracted into a separate repository
12061 available at: https://git.gnome.org/browse/libgudev/
12062 It is now managed as part of the Gnome project. Distributions
12063 are recommended to pass --disable-gudev to systemd and use
12064 gudev from the Gnome project instead. gudev is still included
12065 in systemd, for now. It will be removed soon, though. Please
12066 also see the announcement-thread on systemd-devel:
12067 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
12069 * systemd now exposes a CPUUsageNSec= property for each
12070 service unit on the bus, that contains the overall consumed
12071 CPU time of a service (the sum of what each process of the
12072 service consumed). This value is only available if
12073 CPUAccounting= is turned on for a service, and is then shown
12074 in the "systemctl status" output.
12076 * Support for configuring alternative mappings of the old SysV
12077 runlevels to systemd targets has been removed. They are now
12078 hardcoded in a way that runlevels 2, 3, 4 all map to
12079 multi-user.target and 5 to graphical.target (which
12080 previously was already the default behaviour).
12082 * The auto-mounter logic gained support for mount point
12083 expiry, using a new TimeoutIdleSec= setting in .automount
12084 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
12086 * The EFI System Partition (ESP) as mounted to /boot by
12087 systemd-efi-boot-generator will now be unmounted
12088 automatically after 2 minutes of not being used. This should
12089 minimize the risk of ESP corruptions.
12091 * New /etc/fstab options x-systemd.requires= and
12092 x-systemd.requires-mounts-for= are now supported to express
12093 additional dependencies for mounts. This is useful for
12094 journaling file systems that support external journal
12095 devices or overlay file systems that require underlying file
12096 systems to be mounted.
12098 * systemd does not support direct live-upgrades (via systemctl
12099 daemon-reexec) from versions older than v44 anymore. As no
12100 distribution we are aware of shipped such old versions in a
12101 stable release this should not be problematic.
12103 * When systemd forks off a new per-connection service instance
12104 it will now set the $REMOTE_ADDR environment variable to the
12105 remote IP address, and $REMOTE_PORT environment variable to
12106 the remote IP port. This behaviour is similar to the
12107 corresponding environment variables defined by CGI.
12109 * systemd-networkd gained support for uplink failure
12110 detection. The BindCarrier= option allows binding interface
12111 configuration dynamically to the link sense of other
12112 interfaces. This is useful to achieve behaviour like in
12115 * systemd-networkd gained support for configuring the DHCP
12116 client identifier to use when requesting leases.
12118 * systemd-networkd now has a per-network UseNTP= option to
12119 configure whether NTP server information acquired via DHCP
12120 is passed on to services like systemd-timesyncd.
12122 * systemd-networkd gained support for vti6 tunnels.
12124 * Note that systemd-networkd manages the sysctl variable
12125 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
12126 it is configured for since v219. The variable controls IP
12127 forwarding, and is a per-interface alternative to the global
12128 /proc/sys/net/ipv[46]/ip_forward. This setting is
12129 configurable in the IPForward= option, which defaults to
12130 "no". This means if networkd is used for an interface it is
12131 no longer sufficient to set the global sysctl option to turn
12132 on IP forwarding! Instead, the .network file option
12133 IPForward= needs to be turned on! Note that the
12134 implementation of this behaviour was broken in v219 and has
12135 been fixed in v220.
12137 * Many bonding and vxlan options are now configurable in
12140 * systemd-nspawn gained a new --property= setting to set unit
12141 properties for the container scope. This is useful for
12142 setting resource parameters (e.g. "CPUShares=500") on
12143 containers started from the command line.
12145 * systemd-nspawn gained a new --private-users= switch to make
12146 use of user namespacing available on recent Linux kernels.
12148 * systemd-nspawn may now be called as part of a shell pipeline
12149 in which case the pipes used for stdin and stdout are passed
12150 directly to the process invoked in the container, without
12151 indirection via a pseudo tty.
12153 * systemd-nspawn gained a new switch to control the UNIX
12154 signal to use when killing the init process of the container
12155 when shutting down.
12157 * systemd-nspawn gained a new --overlay= switch for mounting
12158 overlay file systems into the container using the new kernel
12161 * When a container image is imported via systemd-importd and
12162 the host file system is not btrfs, a loopback block device
12163 file is created in /var/lib/machines.raw with a btrfs file
12164 system inside. It is then mounted to /var/lib/machines to
12165 enable btrfs features for container management. The loopback
12166 file and btrfs file system is grown as needed when container
12167 images are imported via systemd-importd.
12169 * systemd-machined/systemd-importd gained support for btrfs
12170 quota, to enforce container disk space limits on disk. This
12171 is exposed in "machinectl set-limit".
12173 * systemd-importd now can import containers from local .tar,
12174 .raw and .qcow2 images, and export them to .tar and .raw. It
12175 can also import dkr v2 images now from the network (on top
12178 * systemd-importd gained support for verifying downloaded
12179 images with gpg2 (previously only gpg1 was supported).
12181 * systemd-machined, systemd-logind, systemd: most bus calls are
12182 now accessible to unprivileged processes via polkit. Also,
12183 systemd-logind will now allow users to kill their own sessions
12184 without further privileges or authorization.
12186 * systemd-shutdownd has been removed. This service was
12187 previously responsible for implementing scheduled shutdowns
12188 as exposed in /usr/bin/shutdown's time parameter. This
12189 functionality has now been moved into systemd-logind and is
12190 accessible via a bus interface.
12192 * "systemctl reboot" gained a new switch --firmware-setup that
12193 can be used to reboot into the EFI firmware setup, if that
12194 is available. systemd-logind now exposes an API on the bus
12195 to trigger such reboots, in case graphical desktop UIs want
12196 to cover this functionality.
12198 * "systemctl enable", "systemctl disable" and "systemctl mask"
12199 now support a new "--now" switch. If specified the units
12200 that are enabled will also be started, and the ones
12201 disabled/masked also stopped.
12203 * The Gummiboot EFI boot loader tool has been merged into
12204 systemd, and renamed to "systemd-boot". The bootctl tool has been
12205 updated to support systemd-boot.
12207 * An EFI kernel stub has been added that may be used to create
12208 kernel EFI binaries that contain not only the actual kernel,
12209 but also an initrd, boot splash, command line and OS release
12210 information. This combined binary can then be signed as a
12211 single image, so that the firmware can verify it all in one
12212 step. systemd-boot has special support for EFI binaries created
12213 like this and can extract OS release information from them
12214 and show them in the boot menu. This functionality is useful
12215 to implement cryptographically verified boot schemes.
12217 * Optional support has been added to systemd-fsck to pass
12218 fsck's progress report to an AF_UNIX socket in the file
12221 * udev will no longer create device symlinks for all block devices by
12222 default. A deny list for excluding special block devices from this
12223 logic has been turned into an allow list that requires picking block
12224 devices explicitly that require device symlinks.
12226 * A new (currently still internal) API sd-device.h has been
12227 added to libsystemd. This modernized API is supposed to
12228 replace libudev eventually. In fact, already much of libudev
12229 is now just a wrapper around sd-device.h.
12231 * A new hwdb database for storing metadata about pointing
12232 stick devices has been added.
12234 * systemd-tmpfiles gained support for setting file attributes
12235 similar to the "chattr" tool with new 'h' and 'H' lines.
12237 * systemd-journald will no longer unconditionally set the
12238 btrfs NOCOW flag on new journal files. This is instead done
12239 with tmpfiles snippet using the new 'h' line type. This
12240 allows easy disabling of this logic, by masking the
12241 journal-nocow.conf tmpfiles file.
12243 * systemd-journald will now translate audit message types to
12244 human readable identifiers when writing them to the
12245 journal. This should improve readability of audit messages.
12247 * The LUKS logic gained support for the offset= and skip=
12248 options in /etc/crypttab, as previously implemented by
12251 * /usr/lib/os-release gained a new optional field VARIANT= for
12252 distributions that support multiple variants (such as a
12253 desktop edition, a server edition, …)
12255 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
12256 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
12257 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
12258 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
12259 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
12260 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
12261 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
12262 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
12263 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
12264 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
12265 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
12266 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
12267 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
12268 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
12269 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
12270 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
12271 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
12272 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
12273 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
12274 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
12275 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
12276 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
12277 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
12278 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
12279 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
12280 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
12281 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
12283 — Berlin, 2015-05-22
12287 * Introduce a new API "sd-hwdb.h" for querying the hardware
12288 metadata database. With this minimal interface one can query
12289 and enumerate the udev hwdb, decoupled from the old libudev
12290 library. libudev's interface for this is now only a wrapper
12291 around sd-hwdb. A new tool systemd-hwdb has been added to
12292 interface with and update the database.
12294 * When any of systemd's tools copies files (for example due to
12295 tmpfiles' C lines) a btrfs reflink will attempted first,
12296 before bytewise copying is done.
12298 * systemd-nspawn gained a new --ephemeral switch. When
12299 specified a btrfs snapshot is taken of the container's root
12300 directory, and immediately removed when the container
12301 terminates again. Thus, a container can be started whose
12302 changes never alter the container's root directory, and are
12303 lost on container termination. This switch can also be used
12304 for starting a container off the root file system of the
12305 host without affecting the host OS. This switch is only
12306 available on btrfs file systems.
12308 * systemd-nspawn gained a new --template= switch. It takes the
12309 path to a container tree to use as template for the tree
12310 specified via --directory=, should that directory be
12311 missing. This allows instantiating containers dynamically,
12312 on first run. This switch is only available on btrfs file
12315 * When a .mount unit refers to a mount point on which multiple
12316 mounts are stacked, and the .mount unit is stopped all of
12317 the stacked mount points will now be unmounted until no
12318 mount point remains.
12320 * systemd now has an explicit notion of supported and
12321 unsupported unit types. Jobs enqueued for unsupported unit
12322 types will now fail with an "unsupported" error code. More
12323 specifically .swap, .automount and .device units are not
12324 supported in containers, .busname units are not supported on
12325 non-kdbus systems. .swap and .automount are also not
12326 supported if their respective kernel compile time options
12329 * machinectl gained support for two new "copy-from" and
12330 "copy-to" commands for copying files from a running
12331 container to the host or vice versa.
12333 * machinectl gained support for a new "bind" command to bind
12334 mount host directories into local containers. This is
12335 currently only supported for nspawn containers.
12337 * networkd gained support for configuring bridge forwarding
12338 database entries (fdb) from .network files.
12340 * A new tiny daemon "systemd-importd" has been added that can
12341 download container images in tar, raw, qcow2 or dkr formats,
12342 and make them available locally in /var/lib/machines, so
12343 that they can run as nspawn containers. The daemon can GPG
12344 verify the downloads (not supported for dkr, since it has no
12345 provisions for verifying downloads). It will transparently
12346 decompress bz2, xz, gzip compressed downloads if necessary,
12347 and restore sparse files on disk. The daemon uses privilege
12348 separation to ensure the actual download logic runs with
12349 fewer privileges than the daemon itself. machinectl has
12350 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
12351 make the functionality of importd available to the
12352 user. With this in place the Fedora and Ubuntu "Cloud"
12353 images can be downloaded and booted as containers unmodified
12354 (the Fedora images lack the appropriate GPG signature files
12355 currently, so they cannot be verified, but this will change
12356 soon, hopefully). Note that downloading images is currently
12357 only fully supported on btrfs.
12359 * machinectl is now able to list container images found in
12360 /var/lib/machines, along with some metadata about sizes of
12361 disk and similar. If the directory is located on btrfs and
12362 quota is enabled, this includes quota display. A new command
12363 "image-status" has been added that shows additional
12364 information about images.
12366 * machinectl is now able to clone container images
12367 efficiently, if the underlying file system (btrfs) supports
12368 it, with the new "machinectl clone" command. It also
12369 gained commands for renaming and removing images, as well as
12370 marking them read-only or read-write (supported also on
12371 legacy file systems).
12373 * networkd gained support for collecting LLDP network
12374 announcements, from hardware that supports this. This is
12375 shown in networkctl output.
12377 * systemd-run gained support for a new -t (--pty) switch for
12378 invoking a binary on a pty whose input and output is
12379 connected to the invoking terminal. This allows executing
12380 processes as system services while interactively
12381 communicating with them via the terminal. Most interestingly
12382 this is supported across container boundaries. Invoking
12383 "systemd-run -t /bin/bash" is an alternative to running a
12384 full login session, the difference being that the former
12385 will not register a session, nor go through the PAM session
12388 * tmpfiles gained support for a new "v" line type for creating
12389 btrfs subvolumes. If the underlying file system is a legacy
12390 file system, this automatically degrades to creating a
12391 normal directory. Among others /var/lib/machines is now
12392 created like this at boot, should it be missing.
12394 * The directory /var/lib/containers/ has been deprecated and
12395 been replaced by /var/lib/machines. The term "machines" has
12396 been used in the systemd context as generic term for both
12397 VMs and containers, and hence appears more appropriate for
12398 this, as the directory can also contain raw images bootable
12401 * systemd-nspawn when invoked with -M but without --directory=
12402 or --image= is now capable of searching for the container
12403 root directory, subvolume or disk image automatically, in
12404 /var/lib/machines. systemd-nspawn@.service has been updated
12405 to make use of this, thus allowing it to be used for raw
12408 * A new machines.target unit has been introduced that is
12409 supposed to group all containers/VMs invoked as services on
12410 the system. systemd-nspawn@.service has been updated to
12411 integrate with that.
12413 * machinectl gained a new "start" command, for invoking a
12414 container as a service. "machinectl start foo" is mostly
12415 equivalent to "systemctl start systemd-nspawn@foo.service",
12416 but handles escaping in a nicer way.
12418 * systemd-nspawn will now mount most of the cgroupfs tree
12419 read-only into each container, with the exception of the
12420 container's own subtree in the name=systemd hierarchy.
12422 * journald now sets the special FS_NOCOW file flag for its
12423 journal files. This should improve performance on btrfs, by
12424 avoiding heavy fragmentation when journald's write-pattern
12425 is used on COW file systems. It degrades btrfs' data
12426 integrity guarantees for the files to the same levels as for
12427 ext3/ext4 however. This should be OK though as journald does
12428 its own data integrity checks and all its objects are
12429 checksummed on disk. Also, journald should handle btrfs disk
12430 full events a lot more gracefully now, by processing SIGBUS
12431 errors, and not relying on fallocate() anymore.
12433 * When journald detects that journal files it is writing to
12434 have been deleted it will immediately start new journal
12437 * systemd now provides a way to store file descriptors
12438 per-service in PID 1. This is useful for daemons to ensure
12439 that fds they require are not lost during a daemon
12440 restart. The fds are passed to the daemon on the next
12441 invocation in the same way socket activation fds are
12442 passed. This is now used by journald to ensure that the
12443 various sockets connected to all the system's stdout/stderr
12444 are not lost when journald is restarted. File descriptors
12445 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
12446 an extension to sd_notify(). Note that a limit is enforced
12447 on the number of fds a service can store in PID 1, and it
12448 defaults to 0, so that no fds may be stored, unless this is
12449 explicitly turned on.
12451 * The default TERM variable to use for units connected to a
12452 terminal, when no other value is explicitly is set is now
12453 vt220 rather than vt102. This should be fairly safe still,
12454 but allows PgUp/PgDn work.
12456 * The /etc/crypttab option header= as known from Debian is now
12459 * "loginctl user-status" and "loginctl session-status" will
12460 now show the last 10 lines of log messages of the
12461 user/session following the status output. Similar,
12462 "machinectl status" will show the last 10 log lines
12463 associated with a virtual machine or container
12464 service. (Note that this is usually not the log messages
12465 done in the VM/container itself, but simply what the
12466 container manager logs. For nspawn this includes all console
12469 * "loginctl session-status" without further argument will now
12470 show the status of the session of the caller. Similar,
12471 "lock-session", "unlock-session", "activate",
12472 "enable-linger", "disable-linger" may now be called without
12473 session/user parameter in which case they apply to the
12474 caller's session/user.
12476 * An X11 session scriptlet is now shipped that uploads
12477 $DISPLAY and $XAUTHORITY into the environment of the systemd
12478 --user daemon if a session begins. This should improve
12479 compatibility with X11 enabled applications run as systemd
12482 * Generators are now subject to masking via /etc and /run, the
12483 same way as unit files.
12485 * networkd .network files gained support for configuring
12486 per-link IPv4/IPv6 packet forwarding as well as IPv4
12487 masquerading. This is by default turned on for veth links to
12488 containers, as registered by systemd-nspawn. This means that
12489 nspawn containers run with --network-veth will now get
12490 automatic routed access to the host's networks without any
12491 further configuration or setup, as long as networkd runs on
12494 * systemd-nspawn gained the --port= (-p) switch to expose TCP
12495 or UDP posts of a container on the host. With this in place
12496 it is possible to run containers with private veth links
12497 (--network-veth), and have their functionality exposed on
12498 the host as if their services were running directly on the
12501 * systemd-nspawn's --network-veth switch now gained a short
12502 version "-n", since with the changes above it is now truly
12503 useful out-of-the-box. The systemd-nspawn@.service has been
12504 updated to make use of it too by default.
12506 * systemd-nspawn will now maintain a per-image R/W lock, to
12507 ensure that the same image is not started more than once
12508 writable. (It's OK to run an image multiple times
12509 simultaneously in read-only mode.)
12511 * systemd-nspawn's --image= option is now capable of
12512 dissecting and booting MBR and GPT disk images that contain
12513 only a single active Linux partition. Previously it
12514 supported only GPT disk images with proper GPT type
12515 IDs. This allows running cloud images from major
12516 distributions directly with systemd-nspawn, without
12519 * In addition to collecting mouse dpi data in the udev
12520 hardware database, there's now support for collecting angle
12521 information for mouse scroll wheels. The database is
12522 supposed to guarantee similar scrolling behavior on mice
12523 that it knows about. There's also support for collecting
12524 information about Touchpad types.
12526 * udev's input_id built-in will now also collect touch screen
12527 dimension data and attach it to probed devices.
12529 * /etc/os-release gained support for a Distribution Privacy
12532 * networkd gained support for creating "ipvlan", "gretap",
12533 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
12535 * systemd-tmpfiles gained support for "a" lines for setting
12538 * systemd-nspawn will now mount /tmp in the container to
12539 tmpfs, automatically.
12541 * systemd now exposes the memory.usage_in_bytes cgroup
12542 attribute and shows it for each service in the "systemctl
12543 status" output, if available.
12545 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
12546 immediate reboot is triggered. This useful if shutdown is
12547 hung and is unable to complete, to expedite the
12548 operation. Note that this kind of reboot will still unmount
12549 all file systems, and hence should not result in fsck being
12550 run on next reboot.
12552 * A .device unit for an optical block device will now be
12553 considered active only when a medium is in the drive. Also,
12554 mount units are now bound to their backing devices thus
12555 triggering automatic unmounting when devices become
12556 unavailable. With this in place systemd will now
12557 automatically unmount left-over mounts when a CD-ROM is
12558 ejected or a USB stick is yanked from the system.
12560 * networkd-wait-online now has support for waiting for
12561 specific interfaces only (with globbing), and for giving up
12562 after a configurable timeout.
12564 * networkd now exits when idle. It will be automatically
12565 restarted as soon as interfaces show up, are removed or
12566 change state. networkd will stay around as long as there is
12567 at least one DHCP state machine or similar around, that keep
12570 * networkd may now configure IPv6 link-local addressing in
12571 addition to IPv4 link-local addressing.
12573 * The IPv6 "token" for use in SLAAC may now be configured for
12574 each .network interface in networkd.
12576 * Routes configured with networkd may now be assigned a scope
12579 * networkd's [Match] sections now support globbing and lists
12580 of multiple space-separated matches per item.
12582 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
12583 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
12584 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
12585 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
12586 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
12587 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
12588 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
12589 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
12590 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
12591 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
12592 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
12593 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
12594 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
12595 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
12596 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
12597 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
12598 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
12599 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
12600 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
12601 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
12602 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
12603 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
12604 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
12605 Hoffmann, Zbigniew Jędrzejewski-Szmek
12607 — Berlin, 2015-02-16
12611 * When querying unit file enablement status (for example via
12612 "systemctl is-enabled"), a new state "indirect" is now known
12613 which indicates that a unit might not be enabled itself, but
12614 another unit listed in its Also= setting might be.
12616 * Similar to the various existing ConditionXYZ= settings for
12617 units, there are now matching AssertXYZ= settings. While
12618 failing conditions cause a unit to be skipped, but its job
12619 to succeed, failing assertions declared like this will cause
12620 a unit start operation and its job to fail.
12622 * hostnamed now knows a new chassis type "embedded".
12624 * systemctl gained a new "edit" command. When used on a unit
12625 file, this allows extending unit files with .d/ drop-in
12626 configuration snippets or editing the full file (after
12627 copying it from /usr/lib to /etc). This will invoke the
12628 user's editor (as configured with $EDITOR), and reload the
12629 modified configuration after editing.
12631 * "systemctl status" now shows the suggested enablement state
12632 for a unit, as declared in the (usually vendor-supplied)
12633 system preset files.
12635 * nss-myhostname will now resolve the single-label hostname
12636 "gateway" to the locally configured default IP routing
12637 gateways, ordered by their metrics. This assigns a stable
12638 name to the used gateways, regardless which ones are
12639 currently configured. Note that the name will only be
12640 resolved after all other name sources (if nss-myhostname is
12641 configured properly) and should hence not negatively impact
12642 systems that use the single-label hostname "gateway" in
12645 * systemd-inhibit now allows filtering by mode when listing
12648 * Scope and service units gained a new "Delegate" boolean
12649 property, which, when set, allows processes running inside the
12650 unit to further partition resources. This is primarily
12651 useful for systemd user instances as well as container
12654 * journald will now pick up audit messages directly from
12655 the kernel, and log them like any other log message. The
12656 audit fields are split up and fully indexed. This means that
12657 journalctl in many ways is now a (nicer!) alternative to
12658 ausearch, the traditional audit client. Note that this
12659 implements only a minimal audit client. If you want the
12660 special audit modes like reboot-on-log-overflow, please use
12661 the traditional auditd instead, which can be used in
12662 parallel to journald.
12664 * The ConditionSecurity= unit file option now understands the
12665 special string "audit" to check whether auditing is
12668 * journalctl gained two new commands --vacuum-size= and
12669 --vacuum-time= to delete old journal files until the
12670 remaining ones take up no more than the specified size on disk,
12671 or are not older than the specified time.
12673 * A new, native PPPoE library has been added to sd-network,
12674 systemd's library of light-weight networking protocols. This
12675 library will be used in a future version of networkd to
12676 enable PPPoE communication without an external pppd daemon.
12678 * The busctl tool now understands a new "capture" verb that
12679 works similar to "monitor", but writes a packet capture
12680 trace to STDOUT that can be redirected to a file which is
12681 compatible with libcap's capture file format. This can then
12682 be loaded in Wireshark and similar tools to inspect bus
12685 * The busctl tool now understands a new "tree" verb that shows
12686 the object trees of a specific service on the bus, or of all
12689 * The busctl tool now understands a new "introspect" verb that
12690 shows all interfaces and members of objects on the bus,
12691 including their signature and values. This is particularly
12692 useful to get more information about bus objects shown by
12693 the new "busctl tree" command.
12695 * The busctl tool now understands new verbs "call",
12696 "set-property" and "get-property" for invoking bus method
12697 calls, setting and getting bus object properties in a
12700 * busctl gained a new --augment-creds= argument that controls
12701 whether the tool shall augment credential information it
12702 gets from the bus with data from /proc, in a possibly
12705 * nspawn's --link-journal= switch gained two new values
12706 "try-guest" and "try-host" that work like "guest" and
12707 "host", but do not fail if the host has no persistent
12708 journaling enabled. -j is now equivalent to
12709 --link-journal=try-guest.
12711 * macvlan network devices created by nspawn will now have
12712 stable MAC addresses.
12714 * A new SmackProcessLabel= unit setting has been added, which
12715 controls the SMACK security label processes forked off by
12716 the respective unit shall use.
12718 * If compiled with --enable-xkbcommon, systemd-localed will
12719 verify x11 keymap settings by compiling the given keymap. It
12720 will spew out warnings if the compilation fails. This
12721 requires libxkbcommon to be installed.
12723 * When a coredump is collected, a larger number of metadata
12724 fields is now collected and included in the journal records
12725 created for it. More specifically, control group membership,
12726 environment variables, memory maps, working directory,
12727 chroot directory, /proc/$PID/status, and a list of open file
12728 descriptors is now stored in the log entry.
12730 * The udev hwdb now contains DPI information for mice. For
12733 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
12735 * All systemd programs that read standalone configuration
12736 files in /etc now also support a corresponding series of
12737 .conf.d configuration directories in /etc/, /run/,
12738 /usr/local/lib/, /usr/lib/, and (if configured with
12739 --enable-split-usr) /lib/. In particular, the following
12740 configuration files now have corresponding configuration
12741 directories: system.conf user.conf, logind.conf,
12742 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
12743 resolved.conf, timesyncd.conf, journal-remote.conf, and
12744 journal-upload.conf. Note that distributions should use the
12745 configuration directories in /usr/lib/; the directories in
12746 /etc/ are reserved for the system administrator.
12748 * systemd-rfkill will no longer take the rfkill device name
12749 into account when storing rfkill state on disk, as the name
12750 might be dynamically assigned and not stable. Instead, the
12751 ID_PATH udev variable combined with the rfkill type (wlan,
12752 bluetooth, …) is used.
12754 * A new service systemd-machine-id-commit.service has been
12755 added. When used on systems where /etc is read-only during
12756 boot, and /etc/machine-id is not initialized (but an empty
12757 file), this service will copy the temporary machine ID
12758 created as replacement into /etc after the system is fully
12759 booted up. This is useful for systems that are freshly
12760 installed with a non-initialized machine ID, but should get
12761 a fixed machine ID for subsequent boots.
12763 * networkd's .netdev files now provide a large set of
12764 configuration parameters for VXLAN devices. Similarly, the
12765 bridge port cost parameter is now configurable in .network
12766 files. There's also new support for configuring IP source
12767 routing. networkd .link files gained support for a new
12768 OriginalName= match that is useful to match against the
12769 original interface name the kernel assigned. .network files
12770 may include MTU= and MACAddress= fields for altering the MTU
12771 and MAC address while being connected to a specific network
12774 * The LUKS logic gained supported for configuring
12775 UUID-specific key files. There's also new support for naming
12776 LUKS device from the kernel command line, using the new
12777 luks.name= argument.
12779 * Timer units may now be transiently created via the bus API
12780 (this was previously already available for scope and service
12781 units). In addition it is now possible to create multiple
12782 transient units at the same time with a single bus call. The
12783 "systemd-run" tool has been updated to make use of this for
12784 running commands on a specified time, in at(1)-style.
12786 * tmpfiles gained support for "t" lines, for assigning
12787 extended attributes to files. Among other uses this may be
12788 used to assign SMACK labels to files.
12790 Contributions from: Alin Rauta, Alison Chaiken, Andrej
12791 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
12792 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
12793 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
12794 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
12795 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
12796 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
12797 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
12798 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
12799 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
12800 Michael Biebl, Michael Chapman, Michael Marineau, Michal
12801 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
12802 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
12803 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
12804 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
12805 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
12806 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
12807 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
12809 — Berlin, 2014-12-10
12813 * journalctl gained the new options -t/--identifier= to match
12814 on the syslog identifier (aka "tag"), as well as --utc to
12815 show log timestamps in the UTC timezone. journalctl now also
12816 accepts -n/--lines=all to disable line capping in a pager.
12818 * journalctl gained a new switch, --flush, that synchronously
12819 flushes logs from /run/log/journal to /var/log/journal if
12820 persistent storage is enabled. systemd-journal-flush.service
12821 now waits until the operation is complete.
12823 * Services can notify the manager before they start a reload
12824 (by sending RELOADING=1) or shutdown (by sending
12825 STOPPING=1). This allows the manager to track and show the
12826 internal state of daemons and closes a race condition when
12827 the process is still running but has closed its D-Bus
12830 * Services with Type=oneshot do not have to have any ExecStart
12833 * User units are now loaded also from
12834 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
12835 /run/systemd/user directory that was already previously
12836 supported, but is under the control of the user.
12838 * Job timeouts (i.e. timeouts on the time a job that is
12839 queued stays in the run queue) can now optionally result in
12840 immediate reboot or power-off actions (JobTimeoutAction= and
12841 JobTimeoutRebootArgument=). This is useful on ".target"
12842 units, to limit the maximum time a target remains
12843 undispatched in the run queue, and to trigger an emergency
12844 operation in such a case. This is now used by default to
12845 turn off the system if boot-up (as defined by everything in
12846 basic.target) hangs and does not complete for at least
12847 15min. Also, if power-off or reboot hang for at least 30min
12848 an immediate power-off/reboot operation is triggered. This
12849 functionality is particularly useful to increase reliability
12850 on embedded devices, but also on laptops which might
12851 accidentally get powered on when carried in a backpack and
12852 whose boot stays stuck in a hard disk encryption passphrase
12855 * systemd-logind can be configured to also handle lid switch
12856 events even when the machine is docked or multiple displays
12857 are attached (HandleLidSwitchDocked= option).
12859 * A helper binary and a service have been added which can be
12860 used to resume from hibernation in the initramfs. A
12861 generator will parse the resume= option on the kernel
12862 command line to trigger resume.
12864 * A user console daemon systemd-consoled has been
12865 added. Currently, it is a preview, and will so far open a
12866 single terminal on each session of the user marked as
12867 Desktop=systemd-console.
12869 * Route metrics can be specified for DHCP routes added by
12872 * The SELinux context of socket-activated services can be set
12873 from the information provided by the networking stack
12874 (SELinuxContextFromNet= option).
12876 * Userspace firmware loading support has been removed and
12877 the minimum supported kernel version is thus bumped to 3.7.
12879 * Timeout for udev workers has been increased from 1 to 3
12880 minutes, but a warning will be printed after 1 minute to
12881 help diagnose kernel modules that take a long time to load.
12883 * Udev rules can now remove tags on devices with TAG-="foobar".
12885 * systemd's readahead implementation has been removed. In many
12886 circumstances it didn't give expected benefits even for
12887 rotational disk drives and was becoming less relevant in the
12888 age of SSDs. As none of the developers has been using
12889 rotating media anymore, and nobody stepped up to actively
12890 maintain this component of systemd it has now been removed.
12892 * Swap units can use Options= to specify discard options.
12893 Discard options specified for swaps in /etc/fstab are now
12896 * Docker containers are now detected as a separate type of
12899 * The Password Agent protocol gained support for queries where
12900 the user input is shown, useful e.g. for user names.
12901 systemd-ask-password gained a new --echo option to turn that
12904 * The default sysctl.d/ snippets will now set:
12906 net.core.default_qdisc = fq_codel
12908 This selects Fair Queuing Controlled Delay as the default
12909 queuing discipline for network interfaces. fq_codel helps
12910 fight the network bufferbloat problem. It is believed to be
12911 a good default with no tuning required for most workloads.
12912 Downstream distributions may override this choice. On 10Gbit
12913 servers that do not do forwarding, "fq" may perform better.
12914 Systems without a good clocksource should use "pfifo_fast".
12916 * If kdbus is enabled during build a new option BusPolicy= is
12917 available for service units, that allows locking all service
12918 processes into a stricter bus policy, in order to limit
12919 access to various bus services, or even hide most of them
12920 from the service's view entirely.
12922 * networkctl will now show the .network and .link file
12923 networkd has applied to a specific interface.
12925 * sd-login gained a new API call sd_session_get_desktop() to
12926 query which desktop environment has been selected for a
12929 * UNIX utmp support is now compile-time optional to support
12930 legacy-free systems.
12932 * systemctl gained two new commands "add-wants" and
12933 "add-requires" for pulling in units from specific targets
12936 * If the word "rescue" is specified on the kernel command line
12937 the system will now boot into rescue mode (aka
12938 rescue.target), which was previously available only by
12939 specifying "1" or "systemd.unit=rescue.target" on the kernel
12940 command line. This new kernel command line option nicely
12941 mirrors the already existing "emergency" kernel command line
12944 * New kernel command line options mount.usr=, mount.usrflags=,
12945 mount.usrfstype= have been added that match root=, rootflags=,
12946 rootfstype= but allow mounting a specific file system to
12949 * The $NOTIFY_SOCKET is now also passed to control processes of
12950 services, not only the main process.
12952 * This version reenables support for fsck's -l switch. This
12953 means at least version v2.25 of util-linux is required for
12954 operation, otherwise dead-locks on device nodes may
12955 occur. Again: you need to update util-linux to at least
12956 v2.25 when updating systemd to v217.
12958 * The "multi-seat-x" tool has been removed from systemd, as
12959 its functionality has been integrated into X servers 1.16,
12960 and the tool is hence redundant. It is recommended to update
12961 display managers invoking this tool to simply invoke X
12962 directly from now on, again.
12964 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
12965 message flag has been added for all of systemd's polkit
12966 authenticated method calls has been added. In particular this
12967 now allows optional interactive authorization via polkit for
12968 many of PID1's privileged operations such as unit file
12969 enabling and disabling.
12971 * "udevadm hwdb --update" learnt a new switch "--usr" for
12972 placing the rebuilt hardware database in /usr instead of
12973 /etc. When used only hardware database entries stored in
12974 /usr will be used, and any user database entries in /etc are
12975 ignored. This functionality is useful for vendors to ship a
12976 pre-built database on systems where local configuration is
12977 unnecessary or unlikely.
12979 * Calendar time specifications in .timer units now also
12980 understand the strings "semi-annually", "quarterly" and
12981 "minutely" as shortcuts (in addition to the preexisting
12982 "annually", "hourly", …).
12984 * systemd-tmpfiles will now correctly create files in /dev
12985 at boot which are marked for creation only at boot. It is
12986 recommended to always create static device nodes with 'c!'
12987 and 'b!', so that they are created only at boot and not
12988 overwritten at runtime.
12990 * When the watchdog logic is used for a service (WatchdogSec=)
12991 and the watchdog timeout is hit the service will now be
12992 terminated with SIGABRT (instead of just SIGTERM), in order
12993 to make sure a proper coredump and backtrace is
12994 generated. This ensures that hanging services will result in
12995 similar coredump/backtrace behaviour as services that hit a
12996 segmentation fault.
12998 Contributions from: Andreas Henriksson, Andrei Borzenkov,
12999 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
13000 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
13001 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
13002 Herrmann, David Sommerseth, David Strauss, Emil Renner
13003 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
13004 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
13005 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
13006 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
13007 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
13008 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
13009 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
13010 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
13011 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
13012 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
13013 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
13014 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
13015 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
13016 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
13017 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
13018 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
13021 — Berlin, 2014-10-28
13025 * timedated no longer reads NTP implementation unit names from
13026 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
13027 implementations should add a
13029 Conflicts=systemd-timesyncd.service
13031 to their unit files to take over and replace systemd's NTP
13032 default functionality.
13034 * systemd-sysusers gained a new line type "r" for configuring
13035 which UID/GID ranges to allocate system users/groups
13036 from. Lines of type "u" may now add an additional column
13037 that specifies the home directory for the system user to be
13038 created. Also, systemd-sysusers may now optionally read user
13039 information from STDIN instead of a file. This is useful for
13040 invoking it from RPM preinst scriptlets that need to create
13041 users before the first RPM file is installed since these
13042 files might need to be owned by them. A new
13043 %sysusers_create_inline RPM macro has been introduced to do
13044 just that. systemd-sysusers now updates the shadow files as
13045 well as the user/group databases, which should enhance
13046 compatibility with certain tools like grpck.
13048 * A number of bus APIs of PID 1 now optionally consult polkit to
13049 permit access for otherwise unprivileged clients under certain
13050 conditions. Note that this currently doesn't support
13051 interactive authentication yet, but this is expected to be
13052 added eventually, too.
13054 * /etc/machine-info now has new fields for configuring the
13055 deployment environment of the machine, as well as the
13056 location of the machine. hostnamectl has been updated with
13057 new command to update these fields.
13059 * systemd-timesyncd has been updated to automatically acquire
13060 NTP server information from systemd-networkd, which might
13061 have been discovered via DHCP.
13063 * systemd-resolved now includes a caching DNS stub resolver
13064 and a complete LLMNR name resolution implementation. A new
13065 NSS module "nss-resolve" has been added which can be used
13066 instead of glibc's own "nss-dns" to resolve hostnames via
13067 systemd-resolved. Hostnames, addresses and arbitrary RRs may
13068 be resolved via systemd-resolved D-Bus APIs. In contrast to
13069 the glibc internal resolver systemd-resolved is aware of
13070 multi-homed system, and keeps DNS server and caches separate
13071 and per-interface. Queries are sent simultaneously on all
13072 interfaces that have DNS servers configured, in order to
13073 properly handle VPNs and local LANs which might resolve
13074 separate sets of domain names. systemd-resolved may acquire
13075 DNS server information from systemd-networkd automatically,
13076 which in turn might have discovered them via DHCP. A tool
13077 "systemd-resolve-host" has been added that may be used to
13078 query the DNS logic in resolved. systemd-resolved implements
13079 IDNA and automatically uses IDNA or UTF-8 encoding depending
13080 on whether classic DNS or LLMNR is used as transport. In the
13081 next releases we intend to add a DNSSEC and mDNS/DNS-SD
13082 implementation to systemd-resolved.
13084 * A new NSS module nss-mymachines has been added, that
13085 automatically resolves the names of all local registered
13086 containers to their respective IP addresses.
13088 * A new client tool "networkctl" for systemd-networkd has been
13089 added. It currently is entirely passive and will query
13090 networking configuration from udev, rtnetlink and networkd,
13091 and present it to the user in a very friendly
13092 way. Eventually, we hope to extend it to become a full
13093 control utility for networkd.
13095 * .socket units gained a new DeferAcceptSec= setting that
13096 controls the kernels' TCP_DEFER_ACCEPT sockopt for
13097 TCP. Similarly, support for controlling TCP keep-alive
13098 settings has been added (KeepAliveTimeSec=,
13099 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
13100 turning off Nagle's algorithm on TCP has been added
13103 * logind learned a new session type "web", for use in projects
13104 like Cockpit which register web clients as PAM sessions.
13106 * timer units with at least one OnCalendar= setting will now
13107 be started only after time-sync.target has been
13108 reached. This way they will not elapse before the system
13109 clock has been corrected by a local NTP client or
13110 similar. This is particular useful on RTC-less embedded
13111 machines, that come up with an invalid system clock.
13113 * systemd-nspawn's --network-veth= switch should now result in
13114 stable MAC addresses for both the outer and the inner side
13117 * systemd-nspawn gained a new --volatile= switch for running
13118 container instances with /etc or /var unpopulated.
13120 * The kdbus client code has been updated to use the new Linux
13121 3.17 memfd subsystem instead of the old kdbus-specific one.
13123 * systemd-networkd's DHCP client and server now support
13124 FORCERENEW. There are also new configuration options to
13125 configure the vendor client identifier and broadcast mode
13128 * systemd will no longer inform the kernel about the current
13129 timezone, as this is necessarily incorrect and racy as the
13130 kernel has no understanding of DST and similar
13131 concepts. This hence means FAT timestamps will be always
13132 considered UTC, similar to what Android is already
13133 doing. Also, when the RTC is configured to the local time
13134 (rather than UTC) systemd will never synchronize back to it,
13135 as this might confuse Windows at a later boot.
13137 * systemd-analyze gained a new command "verify" for offline
13138 validation of unit files.
13140 * systemd-networkd gained support for a couple of additional
13141 settings for bonding networking setups. Also, the metric for
13142 statically configured routes may now be configured. For
13143 network interfaces where this is appropriate the peer IP
13144 address may now be configured.
13146 * systemd-networkd's DHCP client will no longer request
13147 broadcasting by default, as this tripped up some networks.
13148 For hardware where broadcast is required the feature should
13149 be switched back on using RequestBroadcast=yes.
13151 * systemd-networkd will now set up IPv4LL addresses (when
13152 enabled) even if DHCP is configured successfully.
13154 * udev will now default to respect network device names given
13155 by the kernel when the kernel indicates that these are
13156 predictable. This behavior can be tweaked by changing
13157 NamePolicy= in the relevant .link file.
13159 * A new library systemd-terminal has been added that
13160 implements full TTY stream parsing and rendering. This
13161 library is supposed to be used later on for implementing a
13162 full userspace VT subsystem, replacing the current kernel
13165 * A new tool systemd-journal-upload has been added to push
13166 journal data to a remote system running
13167 systemd-journal-remote.
13169 * journald will no longer forward all local data to another
13170 running syslog daemon. This change has been made because
13171 rsyslog (which appears to be the most commonly used syslog
13172 implementation these days) no longer makes use of this, and
13173 instead pulls the data out of the journal on its own. Since
13174 forwarding the messages to a non-existent syslog server is
13175 more expensive than we assumed we have now turned this
13176 off. If you run a syslog server that is not a recent rsyslog
13177 version, you have to turn this option on again
13178 (ForwardToSyslog= in journald.conf).
13180 * journald now optionally supports the LZ4 compressor for
13181 larger journal fields. This compressor should perform much
13182 better than XZ which was the previous default.
13184 * machinectl now shows the IP addresses of local containers,
13185 if it knows them, plus the interface name of the container.
13187 * A new tool "systemd-escape" has been added that makes it
13188 easy to escape strings to build unit names and similar.
13190 * sd_notify() messages may now include a new ERRNO= field
13191 which is parsed and collected by systemd and shown among the
13192 "systemctl status" output for a service.
13194 * A new component "systemd-firstboot" has been added that
13195 queries the most basic systemd information (timezone,
13196 hostname, root password) interactively on first
13197 boot. Alternatively it may also be used to provision these
13198 things offline on OS images installed into directories.
13200 * The default sysctl.d/ snippets will now set
13202 net.ipv4.conf.default.promote_secondaries=1
13204 This has the benefit of no flushing secondary IP addresses
13205 when primary addresses are removed.
13207 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
13208 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
13209 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
13210 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
13211 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
13212 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
13213 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
13214 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
13215 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
13216 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
13217 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
13218 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
13219 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
13220 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
13221 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
13223 — Berlin, 2014-08-19
13227 * A new tool systemd-sysusers has been added. This tool
13228 creates system users and groups in /etc/passwd and
13229 /etc/group, based on static declarative system user/group
13230 definitions in /usr/lib/sysusers.d/. This is useful to
13231 enable factory resets and volatile systems that boot up with
13232 an empty /etc directory, and thus need system users and
13233 groups created during early boot. systemd now also ships
13234 with two default sysusers.d/ files for the most basic
13235 users and groups systemd and the core operating system
13238 * A new tmpfiles snippet has been added that rebuilds the
13239 essential files in /etc on boot, should they be missing.
13241 * A directive for ensuring automatic clean-up of
13242 /var/cache/man/ has been removed from the default
13243 configuration. This line should now be shipped by the man
13244 implementation. The necessary change has been made to the
13245 man-db implementation. Note that you need to update your man
13246 implementation to one that ships this line, otherwise no
13247 automatic clean-up of /var/cache/man will take place.
13249 * A new condition ConditionNeedsUpdate= has been added that
13250 may conditionalize services to only run when /etc or /var
13251 are "older" than the vendor operating system resources in
13252 /usr. This is useful for reconstructing or updating /etc
13253 after an offline update of /usr or a factory reset, on the
13254 next reboot. Services that want to run once after such an
13255 update or reset should use this condition and order
13256 themselves before the new systemd-update-done.service, which
13257 will mark the two directories as fully updated. A number of
13258 service files have been added making use of this, to rebuild
13259 the udev hardware database, the journald message catalog and
13260 dynamic loader cache (ldconfig). The systemd-sysusers tool
13261 described above also makes use of this now. With this in
13262 place it is now possible to start up a minimal operating
13263 system with /etc empty cleanly. For more information on the
13264 concepts involved see this recent blog story:
13266 https://0pointer.de/blog/projects/stateless.html
13268 * A new system group "input" has been introduced, and all
13269 input device nodes get this group assigned. This is useful
13270 for system-level software to get access to input devices. It
13271 complements what is already done for "audio" and "video".
13273 * systemd-networkd learnt minimal DHCPv4 server support in
13274 addition to the existing DHCPv4 client support. It also
13275 learnt DHCPv6 client and IPv6 Router Solicitation client
13276 support. The DHCPv4 client gained support for static routes
13277 passed in from the server. Note that the [DHCPv4] section
13278 known in older systemd-networkd versions has been renamed to
13279 [DHCP] and is now also used by the DHCPv6 client. Existing
13280 .network files using settings of this section should be
13281 updated, though compatibility is maintained. Optionally, the
13282 client hostname may now be sent to the DHCP server.
13284 * networkd gained support for vxlan virtual networks as well
13285 as tun/tap and dummy devices.
13287 * networkd gained support for automatic allocation of address
13288 ranges for interfaces from a system-wide pool of
13289 addresses. This is useful for dynamically managing a large
13290 number of interfaces with a single network configuration
13291 file. In particular this is useful to easily assign
13292 appropriate IP addresses to the veth links of a large number
13293 of nspawn instances.
13295 * RPM macros for processing sysusers, sysctl and binfmt
13296 drop-in snippets at package installation time have been
13299 * The /etc/os-release file should now be placed in
13300 /usr/lib/os-release. The old location is automatically
13301 created as symlink. /usr/lib is the more appropriate
13302 location of this file, since it shall actually describe the
13303 vendor operating system shipped in /usr, and not the
13304 configuration stored in /etc.
13306 * .mount units gained a new boolean SloppyOptions= setting
13307 that maps to mount(8)'s -s option which enables permissive
13308 parsing of unknown mount options.
13310 * tmpfiles learnt a new "L+" directive which creates a symlink
13311 but (unlike "L") deletes a pre-existing file first, should
13312 it already exist and not already be the correct
13313 symlink. Similarly, "b+", "c+" and "p+" directives have been
13314 added as well, which create block and character devices, as
13315 well as fifos in the filesystem, possibly removing any
13316 pre-existing files of different types.
13318 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
13319 'argument' field (which so far specified the source to
13320 symlink/copy the files from) is now optional. If omitted the
13321 same file os copied from /usr/share/factory/ suffixed by the
13322 full destination path. This is useful for populating /etc
13323 with essential files, by copying them from vendor defaults
13324 shipped in /usr/share/factory/etc.
13326 * A new command "systemctl preset-all" has been added that
13327 applies the service preset settings to all installed unit
13328 files. A new switch --preset-mode= has been added that
13329 controls whether only enable or only disable operations
13332 * A new command "systemctl is-system-running" has been added
13333 that allows checking the overall state of the system, for
13334 example whether it is fully up and running.
13336 * When the system boots up with an empty /etc, the equivalent
13337 to "systemctl preset-all" is executed during early boot, to
13338 make sure all default services are enabled after a factory
13341 * systemd now contains a minimal preset file that enables the
13342 most basic services systemd ships by default.
13344 * Unit files' [Install] section gained a new DefaultInstance=
13345 field for defining the default instance to create if a
13346 template unit is enabled with no instance specified.
13348 * A new passive target cryptsetup-pre.target has been added
13349 that may be used by services that need to make they run and
13350 finish before the first LUKS cryptographic device is set up.
13352 * The /dev/loop-control and /dev/btrfs-control device nodes
13353 are now owned by the "disk" group by default, opening up
13354 access to this group.
13356 * systemd-coredump will now automatically generate a
13357 stack trace of all core dumps taking place on the system,
13358 based on elfutils' libdw library. This stack trace is logged
13361 * systemd-coredump may now optionally store coredumps directly
13362 on disk (in /var/lib/systemd/coredump, possibly compressed),
13363 instead of storing them unconditionally in the journal. This
13364 mode is the new default. A new configuration file
13365 /etc/systemd/coredump.conf has been added to configure this
13366 and other parameters of systemd-coredump.
13368 * coredumpctl gained a new "info" verb to show details about a
13369 specific coredump. A new switch "-1" has also been added
13370 that makes sure to only show information about the most
13371 recent entry instead of all entries. Also, as the tool is
13372 generally useful now the "systemd-" prefix of the binary
13373 name has been removed. Distributions that want to maintain
13374 compatibility with the old name should add a symlink from
13375 the old name to the new name.
13377 * journald's SplitMode= now defaults to "uid". This makes sure
13378 that unprivileged users can access their own coredumps with
13379 coredumpctl without restrictions.
13381 * New kernel command line options "systemd.wants=" (for
13382 pulling an additional unit during boot), "systemd.mask="
13383 (for masking a specific unit for the boot), and
13384 "systemd.debug-shell" (for enabling the debug shell on tty9)
13385 have been added. This is implemented in the new generator
13386 "systemd-debug-generator".
13388 * systemd-nspawn will now by default filter a couple of
13389 syscalls for containers, among them those required for
13390 kernel module loading, direct x86 IO port access, swap
13391 management, and kexec. Most importantly though
13392 open_by_handle_at() is now prohibited for containers,
13393 closing a hole similar to a recently discussed vulnerability
13394 in docker regarding access to files on file hierarchies the
13395 container should normally not have access to. Note that, for
13396 nspawn, we generally make no security claims anyway (and
13397 this is explicitly documented in the man page), so this is
13398 just a fix for one of the most obvious problems.
13400 * A new man page file-hierarchy(7) has been added that
13401 contains a minimized, modernized version of the file system
13402 layout systemd expects, similar in style to the FHS
13403 specification or hier(5). A new tool systemd-path(1) has
13404 been added to query many of these paths for the local
13407 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
13408 longer done. Since the directory now has a per-user size
13409 limit, and is cleaned on logout this appears unnecessary,
13410 in particular since this now brings the lifecycle of this
13411 directory closer in line with how IPC objects are handled.
13413 * systemd.pc now exports a number of additional directories,
13414 including $libdir (which is useful to identify the library
13415 path for the primary architecture of the system), and a
13416 couple of drop-in directories.
13418 * udev's predictable network interface names now use the dev_port
13419 sysfs attribute, introduced in linux 3.15 instead of dev_id to
13420 distinguish between ports of the same PCI function. dev_id should
13421 only be used for ports using the same HW address, hence the need
13424 * machined has been updated to export the OS version of a
13425 container (read from /etc/os-release and
13426 /usr/lib/os-release) on the bus. This is now shown in
13427 "machinectl status" for a machine.
13429 * A new service setting RestartForceExitStatus= has been
13430 added. If configured to a set of exit signals or process
13431 return values, the service will be restarted when the main
13432 daemon process exits with any of them, regardless of the
13435 * systemctl's -H switch for connecting to remote systemd
13436 machines has been extended so that it may be used to
13437 directly connect to a specific container on the
13438 host. "systemctl -H root@foobar:waldi" will now connect as
13439 user "root" to host "foobar", and then proceed directly to
13440 the container named "waldi". Note that currently you have to
13441 authenticate as user "root" for this to work, as entering
13442 containers is a privileged operation.
13444 Contributions from: Andreas Henriksson, Benjamin Steinwender,
13445 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
13446 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
13447 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
13448 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
13449 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
13450 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
13451 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
13452 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
13453 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
13454 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
13455 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
13457 — Berlin, 2014-07-03
13461 * As an experimental feature, udev now tries to lock the
13462 disk device node (flock(LOCK_SH|LOCK_NB)) while it
13463 executes events for the disk or any of its partitions.
13464 Applications like partitioning programs can lock the
13465 disk device node (flock(LOCK_EX)) and claim temporary
13466 device ownership that way; udev will entirely skip all event
13467 handling for this disk and its partitions. If the disk
13468 was opened for writing, the close will trigger a partition
13469 table rescan in udev's "watch" facility, and if needed
13470 synthesize "change" events for the disk and all its partitions.
13471 This is now unconditionally enabled, and if it turns out to
13472 cause major problems, we might turn it on only for specific
13473 devices, or might need to disable it entirely. Device Mapper
13474 devices are excluded from this logic.
13476 * We temporarily dropped the "-l" switch for fsck invocations,
13477 since they collide with the flock() logic above. util-linux
13478 upstream has been changed already to avoid this conflict,
13479 and we will re-add "-l" as soon as util-linux with this
13480 change has been released.
13482 * The dependency on libattr has been removed. Since a long
13483 time, the extended attribute calls have moved to glibc, and
13484 libattr is thus unnecessary.
13486 * Virtualization detection works without privileges now. This
13487 means the systemd-detect-virt binary no longer requires
13488 CAP_SYS_PTRACE file capabilities, and our daemons can run
13489 with fewer privileges.
13491 * systemd-networkd now runs under its own "systemd-network"
13492 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
13493 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
13494 loses the ability to write to files owned by root this way.
13496 * Similarly, systemd-resolved now runs under its own
13497 "systemd-resolve" user with no capabilities remaining.
13499 * Similarly, systemd-bus-proxyd now runs under its own
13500 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
13502 * systemd-networkd gained support for setting up "veth"
13503 virtual Ethernet devices for container connectivity, as well
13504 as GRE and VTI tunnels.
13506 * systemd-networkd will no longer automatically attempt to
13507 manually load kernel modules necessary for certain tunnel
13508 transports. Instead, it is assumed the kernel loads them
13509 automatically when required. This only works correctly on
13510 very new kernels. On older kernels, please consider adding
13511 the kernel modules to /etc/modules-load.d/ as a work-around.
13513 * The resolv.conf file systemd-resolved generates has been
13514 moved to /run/systemd/resolve/. If you have a symlink from
13515 /etc/resolv.conf, it might be necessary to correct it.
13517 * Two new service settings, ProtectHome= and ProtectSystem=,
13518 have been added. When enabled, they will make the user data
13519 (such as /home) inaccessible or read-only and the system
13520 (such as /usr) read-only, for specific services. This allows
13521 very light-weight per-service sandboxing to avoid
13522 modifications of user data or system files from
13523 services. These two new switches have been enabled for all
13524 of systemd's long-running services, where appropriate.
13526 * Socket units gained new SocketUser= and SocketGroup=
13527 settings to set the owner user and group of AF_UNIX sockets
13528 and FIFOs in the file system.
13530 * Socket units gained a new RemoveOnStop= setting. If enabled,
13531 all FIFOS and sockets in the file system will be removed
13532 when the specific socket unit is stopped.
13534 * Socket units gained a new Symlinks= setting. It takes a list
13535 of symlinks to create to file system sockets or FIFOs
13536 created by the specific Unix sockets. This is useful to
13537 manage symlinks to socket nodes with the same lifecycle as
13540 * The /dev/log socket and /dev/initctl FIFO have been moved to
13541 /run, and have been replaced by symlinks. This allows
13542 connecting to these facilities even if PrivateDevices=yes is
13543 used for a service (which makes /dev/log itself unavailable,
13544 but /run is left). This also has the benefit of ensuring
13545 that /dev only contains device nodes, directories and
13546 symlinks, and nothing else.
13548 * sd-daemon gained two new calls sd_pid_notify() and
13549 sd_pid_notifyf(). They are similar to sd_notify() and
13550 sd_notifyf(), but allow overriding of the source PID of
13551 notification messages if permissions permit this. This is
13552 useful to send notify messages on behalf of a different
13553 process (for example, the parent process). The
13554 systemd-notify tool has been updated to make use of this
13555 when sending messages (so that notification messages now
13556 originate from the shell script invoking systemd-notify and
13557 not the systemd-notify process itself. This should minimize
13558 a race where systemd fails to associate notification
13559 messages to services when the originating process already
13562 * A new "on-abnormal" setting for Restart= has been added. If
13563 set, it will result in automatic restarts on all "abnormal"
13564 reasons for a process to exit, which includes unclean
13565 signals, core dumps, timeouts and watchdog timeouts, but
13566 does not include clean and unclean exit codes or clean
13567 signals. Restart=on-abnormal is an alternative for
13568 Restart=on-failure for services that shall be able to
13569 terminate and avoid restarts on certain errors, by
13570 indicating so with an unclean exit code. Restart=on-failure
13571 or Restart=on-abnormal is now the recommended setting for
13572 all long-running services.
13574 * If the InaccessibleDirectories= service setting points to a
13575 mount point (or if there are any submounts contained within
13576 it), it is now attempted to completely unmount it, to make
13577 the file systems truly unavailable for the respective
13580 * The ReadOnlyDirectories= service setting and
13581 systemd-nspawn's --read-only parameter are now recursively
13582 applied to all submounts, too.
13584 * Mount units may now be created transiently via the bus APIs.
13586 * The support for SysV and LSB init scripts has been removed
13587 from the systemd daemon itself. Instead, it is now
13588 implemented as a generator that creates native systemd units
13589 from these scripts when needed. This enables us to remove a
13590 substantial amount of legacy code from PID 1, following the
13591 fact that many distributions only ship a very small number
13592 of LSB/SysV init scripts nowadays.
13594 * Privileged Xen (dom0) domains are not considered
13595 virtualization anymore by the virtualization detection
13596 logic. After all, they generally have unrestricted access to
13597 the hardware and usually are used to manage the unprivileged
13600 * systemd-tmpfiles gained a new "C" line type, for copying
13601 files or entire directories.
13603 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
13604 lines. So far, they have been non-globbing versions of the
13605 latter, and have thus been redundant. In future, it is
13606 recommended to only use "z". "m" has hence been removed
13607 from the documentation, even though it stays supported.
13609 * A tmpfiles snippet to recreate the most basic structure in
13610 /var has been added. This is enough to create the /var/run →
13611 /run symlink and create a couple of structural
13612 directories. This allows systems to boot up with an empty or
13613 volatile /var. Of course, while with this change, the core OS
13614 now is capable with dealing with a volatile /var, not all
13615 user services are ready for it. However, we hope that sooner
13616 or later, many service daemons will be changed upstream so
13617 that they are able to automatically create their necessary
13618 directories in /var at boot, should they be missing. This is
13619 the first step to allow state-less systems that only require
13620 the vendor image for /usr to boot.
13622 * systemd-nspawn has gained a new --tmpfs= switch to mount an
13623 empty tmpfs instance to a specific directory. This is
13624 particularly useful for making use of the automatic
13625 reconstruction of /var (see above), by passing --tmpfs=/var.
13627 * Access modes specified in tmpfiles snippets may now be
13628 prefixed with "~", which indicates that they shall be masked
13629 by whether the existing file or directory is currently
13630 writable, readable or executable at all. Also, if specified,
13631 the sgid/suid/sticky bits will be masked for all
13634 * A new passive target unit "network-pre.target" has been
13635 added which is useful for services that shall run before any
13636 network is configured, for example firewall scripts.
13638 * The "floppy" group that previously owned the /dev/fd*
13639 devices is no longer used. The "disk" group is now used
13640 instead. Distributions should probably deprecate usage of
13643 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
13644 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
13645 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
13646 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
13647 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
13648 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
13651 — Berlin, 2014-06-11
13655 * A new "systemd-timesyncd" daemon has been added for
13656 synchronizing the system clock across the network. It
13657 implements an SNTP client. In contrast to NTP
13658 implementations such as chrony or the NTP reference server,
13659 this only implements a client side, and does not bother with
13660 the full NTP complexity, focusing only on querying time from
13661 one remote server and synchronizing the local clock to
13662 it. Unless you intend to serve NTP to networked clients or
13663 want to connect to local hardware clocks, this simple NTP
13664 client should be more than appropriate for most
13665 installations. The daemon runs with minimal privileges, and
13666 has been hooked up with networkd to only operate when
13667 network connectivity is available. The daemon saves the
13668 current clock to disk every time a new NTP sync has been
13669 acquired, and uses this to possibly correct the system clock
13670 early at bootup, in order to accommodate for systems that
13671 lack an RTC such as the Raspberry Pi and embedded devices,
13672 and to make sure that time monotonically progresses on these
13673 systems, even if it is not always correct. To make use of
13674 this daemon, a new system user and group "systemd-timesync"
13675 needs to be created on installation of systemd.
13677 * The queue "seqnum" interface of libudev has been disabled, as
13678 it was generally incompatible with device namespacing as
13679 sequence numbers of devices go "missing" if the devices are
13680 part of a different namespace.
13682 * "systemctl list-timers" and "systemctl list-sockets" gained
13683 a --recursive switch for showing units of these types also
13684 for all local containers, similar in style to the already
13685 supported --recursive switch for "systemctl list-units".
13687 * A new RebootArgument= setting has been added for service
13688 units, which may be used to specify a kernel reboot argument
13689 to use when triggering reboots with StartLimitAction=.
13691 * A new FailureAction= setting has been added for service
13692 units which may be used to specify an operation to trigger
13693 when a service fails. This works similarly to
13694 StartLimitAction=, but unlike it, controls what is done
13695 immediately rather than only after several attempts to
13696 restart the service in question.
13698 * hostnamed got updated to also expose the kernel name,
13699 release, and version on the bus. This is useful for
13700 executing commands like hostnamectl with the -H switch.
13701 systemd-analyze makes use of this to properly display
13702 details when running non-locally.
13704 * The bootchart tool can now show cgroup information in the
13705 graphs it generates.
13707 * The CFS CPU quota cgroup attribute is now exposed for
13708 services. The new CPUQuota= switch has been added for this
13709 which takes a percentage value. Setting this will have the
13710 result that a service may never get more CPU time than the
13711 specified percentage, even if the machine is otherwise idle.
13713 * systemd-networkd learned IPIP and SIT tunnel support.
13715 * LSB init scripts exposing a dependency on $network will now
13716 get a dependency on network-online.target rather than simply
13717 network.target. This should bring LSB handling closer to
13718 what it was on SysV systems.
13720 * A new fsck.repair= kernel option has been added to control
13721 how fsck shall deal with unclean file systems at boot.
13723 * The (.ini) configuration file parser will now silently ignore
13724 sections whose names begin with "X-". This may be used to maintain
13725 application-specific extension sections in unit files.
13727 * machined gained a new API to query the IP addresses of
13728 registered containers. "machinectl status" has been updated
13729 to show these addresses in its output.
13731 * A new call sd_uid_get_display() has been added to the
13732 sd-login APIs for querying the "primary" session of a
13733 user. The "primary" session of the user is elected from the
13734 user's sessions and generally a graphical session is
13735 preferred over a text one.
13737 * A minimal systemd-resolved daemon has been added. It
13738 currently simply acts as a companion to systemd-networkd and
13739 manages resolv.conf based on per-interface DNS
13740 configuration, possibly supplied via DHCP. In the long run
13741 we hope to extend this into a local DNSSEC enabled DNS and
13744 * The systemd-networkd-wait-online tool is now enabled by
13745 default. It will delay network-online.target until a network
13746 connection has been configured. The tool primarily integrates
13747 with networkd, but will also make a best effort to make sense
13748 of network configuration performed in some other way.
13750 * Two new service options StartupCPUShares= and
13751 StartupBlockIOWeight= have been added that work similarly to
13752 CPUShares= and BlockIOWeight= however only apply during
13753 system startup. This is useful to prioritize certain services
13754 differently during bootup than during normal runtime.
13756 * hostnamed has been changed to prefer the statically
13757 configured hostname in /etc/hostname (unless set to
13758 'localhost' or empty) over any dynamic one supplied by
13759 dhcp. With this change, the rules for picking the hostname
13760 match more closely the rules of other configuration settings
13761 where the local administrator's configuration in /etc always
13762 overrides any other settings.
13764 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
13765 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
13766 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
13767 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
13768 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
13769 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
13770 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
13771 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
13772 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
13773 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
13774 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
13775 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
13776 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
13777 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
13778 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
13779 Lindskog, WaLyong Cho, Will Woods, Zbigniew
13782 — Beijing, 2014-05-28
13786 * When restoring the screen brightness at boot, stay away from
13787 the darkest setting or from the lowest 5% of the available
13788 range, depending on which is the larger value of both. This
13789 should effectively protect the user from rebooting into a
13790 black screen, should the brightness have been set to minimum
13793 * sd-login gained a new sd_machine_get_class() call to
13794 determine the class ("vm" or "container") of a machine
13795 registered with machined.
13797 * sd-login gained new calls
13798 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
13799 to query the identity of the peer of a local AF_UNIX
13800 connection. They operate similarly to their sd_pid_get_xyz()
13803 * PID 1 will now maintain a system-wide system state engine
13804 with the states "starting", "running", "degraded",
13805 "maintenance", "stopping". These states are bound to system
13806 startup, normal runtime, runtime with at least one failed
13807 service, rescue/emergency mode and system shutdown. This
13808 state is shown in the "systemctl status" output when no unit
13809 name is passed. It is useful to determine system state, in
13810 particularly when doing so for many systems or containers at
13813 * A new command "list-machines" has been added to "systemctl"
13814 that lists all local OS containers and shows their system
13815 state (see above), if systemd runs inside of them.
13817 * systemctl gained a new "-r" switch to recursively enumerate
13818 units on all local containers, when used with the
13819 "list-unit" command (which is the default one that is
13820 executed when no parameters are specified).
13822 * The GPT automatic partition discovery logic will now honour
13823 two GPT partition flags: one may be set on a partition to
13824 cause it to be mounted read-only, and the other may be set
13825 on a partition to ignore it during automatic discovery.
13827 * Two new GPT type UUIDs have been added for automatic root
13828 partition discovery, for 32-bit and 64-bit ARM. This is not
13829 particularly useful for discovering the root directory on
13830 these architectures during bare-metal boots (since UEFI is
13831 not common there), but still very useful to allow booting of
13832 ARM disk images in nspawn with the -i option.
13834 * MAC addresses of interfaces created with nspawn's
13835 --network-interface= switch will now be generated from the
13836 machine name, and thus be stable between multiple invocations
13839 * logind will now automatically remove all IPC objects owned
13840 by a user if she or he fully logs out. This makes sure that
13841 users who are logged out cannot continue to consume IPC
13842 resources. This covers SysV memory, semaphores and message
13843 queues as well as POSIX shared memory and message
13844 queues. Traditionally, SysV and POSIX IPC had no lifecycle
13845 limits. With this functionality, that is corrected. This may
13846 be turned off by using the RemoveIPC= switch of logind.conf.
13848 * The systemd-machine-id-setup and tmpfiles tools gained a
13849 --root= switch to operate on a specific root directory,
13852 * journald can now forward logged messages to the TTYs of all
13853 logged in users ("wall"). This is the default for all
13854 emergency messages now.
13856 * A new tool systemd-journal-remote has been added to stream
13857 journal log messages across the network.
13859 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
13860 controller trees are mounted into it. Note that the
13861 directories mounted beneath it are not read-only. This is a
13862 security measure and is particularly useful because glibc
13863 actually includes a search logic to pick any tmpfs it can
13864 find to implement shm_open() if /dev/shm is not available
13865 (which it might very well be in namespaced setups).
13867 * machinectl gained a new "poweroff" command to cleanly power
13868 down a local OS container.
13870 * The PrivateDevices= unit file setting will now also drop the
13871 CAP_MKNOD capability from the capability bound set, and
13872 imply DevicePolicy=closed.
13874 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
13875 comprehensively on all long-running systemd services where
13876 this is appropriate.
13878 * systemd-udevd will now run in a disassociated mount
13879 namespace. To mount directories from udev rules, make sure to
13880 pull in mount units via SYSTEMD_WANTS properties.
13882 * The kdbus support gained support for uploading policy into
13883 the kernel. sd-bus gained support for creating "monitoring"
13884 connections that can eavesdrop into all bus communication
13885 for debugging purposes.
13887 * Timestamps may now be specified in seconds since the UNIX
13888 epoch Jan 1st, 1970 by specifying "@" followed by the value
13891 * Native tcpwrap support in systemd has been removed. tcpwrap
13892 is old code, not really maintained anymore and has serious
13893 shortcomings, and better options such as firewalls
13894 exist. For setups that require tcpwrap usage, please
13895 consider invoking your socket-activated service via tcpd,
13896 like on traditional inetd.
13898 * A new system.conf configuration option
13899 DefaultTimerAccuracySec= has been added that controls the
13900 default AccuracySec= setting of .timer units.
13902 * Timer units gained a new WakeSystem= switch. If enabled,
13903 timers configured this way will cause the system to resume
13904 from system suspend (if the system supports that, which most
13907 * Timer units gained a new Persistent= switch. If enabled,
13908 timers configured this way will save to disk when they have
13909 been last triggered. This information is then used on next
13910 reboot to possible execute overdue timer events, that
13911 could not take place because the system was powered off.
13912 This enables simple anacron-like behaviour for timer units.
13914 * systemctl's "list-timers" will now also list the time a
13915 timer unit was last triggered in addition to the next time
13916 it will be triggered.
13918 * systemd-networkd will now assign predictable IPv4LL
13919 addresses to its local interfaces.
13921 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
13922 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
13923 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
13924 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
13925 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
13926 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
13927 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
13928 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
13931 — Berlin, 2014-03-25
13935 * A new unit file setting RestrictAddressFamilies= has been
13936 added to restrict which socket address families unit
13937 processes gain access to. This takes address family names
13938 like "AF_INET" or "AF_UNIX", and is useful to minimize the
13939 attack surface of services via exotic protocol stacks. This
13940 is built on seccomp system call filters.
13942 * Two new unit file settings RuntimeDirectory= and
13943 RuntimeDirectoryMode= have been added that may be used to
13944 manage a per-daemon runtime directories below /run. This is
13945 an alternative for setting up directory permissions with
13946 tmpfiles snippets, and has the advantage that the runtime
13947 directory's lifetime is bound to the daemon runtime and that
13948 the daemon starts up with an empty directory each time. This
13949 is particularly useful when writing services that drop
13950 privileges using the User= or Group= setting.
13952 * The DeviceAllow= unit setting now supports globbing for
13953 matching against device group names.
13955 * The systemd configuration file system.conf gained new
13956 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
13957 DefaultMemoryAccounting= to globally turn on/off accounting
13958 for specific resources (cgroups) for all units. These
13959 settings may still be overridden individually in each unit
13962 * systemd-gpt-auto-generator is now able to discover /srv and
13963 root partitions in addition to /home and swap partitions. It
13964 also supports LUKS-encrypted partitions now. With this in
13965 place, automatic discovery of partitions to mount following
13966 the Discoverable Partitions Specification
13967 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
13968 is now a lot more complete. This allows booting without
13969 /etc/fstab and without root= on the kernel command line on
13970 systems prepared appropriately.
13972 * systemd-nspawn gained a new --image= switch which allows
13973 booting up disk images and Linux installations on any block
13974 device that follow the Discoverable Partitions Specification
13975 (see above). This means that installations made with
13976 appropriately updated installers may now be started and
13977 deployed using container managers, completely
13978 unmodified. (We hope that libvirt-lxc will add support for
13979 this feature soon, too.)
13981 * systemd-nspawn gained a new --network-macvlan= setting to
13982 set up a private macvlan interface for the
13983 container. Similarly, systemd-networkd gained a new
13984 Kind=macvlan setting in .netdev files.
13986 * systemd-networkd now supports configuring local addresses
13989 * A new tool systemd-network-wait-online has been added to
13990 synchronously wait for network connectivity using
13993 * The sd-bus.h bus API gained a new sd_bus_track object for
13994 tracking the lifecycle of bus peers. Note that sd-bus.h is
13995 still not a public API though (unless you specify
13996 --enable-kdbus on the configure command line, which however
13997 voids your warranty and you get no API stability guarantee).
13999 * The $XDG_RUNTIME_DIR runtime directories for each user are
14000 now individual tmpfs instances, which has the benefit of
14001 introducing separate pools for each user, with individual
14002 size limits, and thus making sure that unprivileged clients
14003 can no longer negatively impact the system or other users by
14004 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
14005 RuntimeDirectorySize= has been introduced that allows
14006 controlling the default size limit for all users. It
14007 defaults to 10% of the available physical memory. This is no
14008 replacement for quotas on tmpfs though (which the kernel
14009 still does not support), as /dev/shm and /tmp are still
14010 shared resources used by both the system and unprivileged
14013 * logind will now automatically turn off automatic suspending
14014 on laptop lid close when more than one display is
14015 connected. This was previously expected to be implemented
14016 individually in desktop environments (such as GNOME),
14017 however has been added to logind now, in order to fix a
14018 boot-time race where a desktop environment might not have
14019 been started yet and thus not been able to take an inhibitor
14020 lock at the time where logind already suspends the system
14021 due to a closed lid.
14023 * logind will now wait at least 30s after each system
14024 suspend/resume cycle, and 3min after system boot before
14025 suspending the system due to a closed laptop lid. This
14026 should give USB docking stations and similar enough time to
14027 be probed and configured after system resume and boot in
14028 order to then act as suspend blocker.
14030 * systemd-run gained a new --property= setting which allows
14031 initialization of resource control properties (and others)
14032 for the created scope or service unit. Example: "systemd-run
14033 --property=BlockIOWeight=10 updatedb" may be used to run
14034 updatedb at a low block IO scheduling weight.
14036 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
14037 now also work in --scope mode.
14039 * When systemd is compiled with kdbus support, basic support
14040 for enforced policies is now in place. (Note that enabling
14041 kdbus still voids your warranty and no API compatibility
14042 promises are made.)
14044 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
14045 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14046 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
14047 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
14048 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
14049 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
14050 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
14051 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
14052 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
14053 Zbigniew Jędrzejewski-Szmek
14055 — Berlin, 2014-03-12
14059 * systemd will now relabel /dev after loading the SMACK policy
14060 according to SMACK rules.
14062 * A new unit file option AppArmorProfile= has been added to
14063 set the AppArmor profile for the processes of a unit.
14065 * A new condition check ConditionArchitecture= has been added
14066 to conditionalize units based on the system architecture, as
14067 reported by uname()'s "machine" field.
14069 * systemd-networkd now supports matching on the system
14070 virtualization, architecture, kernel command line, hostname
14073 * logind is now a lot more aggressive when suspending the
14074 machine due to a closed laptop lid. Instead of acting only
14075 on the lid close action, it will continuously watch the lid
14076 status and act on it. This is useful for laptops where the
14077 power button is on the outside of the chassis so that it can
14078 be reached without opening the lid (such as the Lenovo
14079 Yoga). On those machines, logind will now immediately
14080 re-suspend the machine if the power button has been
14081 accidentally pressed while the laptop was suspended and in a
14082 backpack or similar.
14084 * logind will now watch SW_DOCK switches and inhibit reaction
14085 to the lid switch if it is pressed. This means that logind
14086 will not suspend the machine anymore if the lid is closed
14087 and the system is docked, if the laptop supports SW_DOCK
14088 notifications via the input layer. Note that ACPI docking
14089 stations do not generate this currently. Also note that this
14090 logic is usually not fully sufficient and Desktop
14091 Environments should take a lid switch inhibitor lock when an
14092 external display is connected, as systemd will not watch
14095 * nspawn will now make use of the devices cgroup controller by
14096 default, and only permit creation of and access to the usual
14097 API device nodes like /dev/null or /dev/random, as well as
14098 access to (but not creation of) the pty devices.
14100 * We will now ship a default .network file for
14101 systemd-networkd that automatically configures DHCP for
14102 network interfaces created by nspawn's --network-veth or
14103 --network-bridge= switches.
14105 * systemd will now understand the usual M, K, G, T suffixes
14106 according to SI conventions (i.e. to the base 1000) when
14107 referring to throughput and hardware metrics. It will stay
14108 with IEC conventions (i.e. to the base 1024) for software
14109 metrics, according to what is customary according to
14110 Wikipedia. We explicitly document which base applies for
14111 each configuration option.
14113 * The DeviceAllow= setting in unit files now supports a syntax to
14114 allow-list an entire group of devices node majors at once, based on
14115 the /proc/devices listing. For example, with the string "char-pts",
14116 it is now possible to allow-list all current and future pseudo-TTYs
14119 * sd-event learned a new "post" event source. Event sources of
14120 this type are triggered by the dispatching of any event
14121 source of a type that is not "post". This is useful for
14122 implementing clean-up and check event sources that are
14123 triggered by other work being done in the program.
14125 * systemd-networkd is no longer statically enabled, but uses
14126 the usual [Install] sections so that it can be
14127 enabled/disabled using systemctl. It still is enabled by
14130 * When creating a veth interface pair with systemd-nspawn, the
14131 host side will now be prefixed with "vb-" if
14132 --network-bridge= is used, and with "ve-" if --network-veth
14133 is used. This way, it is easy to distinguish these cases on
14134 the host, for example to apply different configuration to
14135 them with systemd-networkd.
14137 * The compatibility libraries for libsystemd-journal.so,
14138 libsystem-id128.so, libsystemd-login.so and
14139 libsystemd-daemon.so do not make use of IFUNC
14140 anymore. Instead, we now build libsystemd.so multiple times
14141 under these alternative names. This means that the footprint
14142 is drastically increased, but given that these are
14143 transitional compatibility libraries, this should not matter
14144 much. This change has been made necessary to support the ARM
14145 platform for these compatibility libraries, as the ARM
14146 toolchain is not really at the same level as the toolchain
14147 for other architectures like x86 and does not support
14148 IFUNC. Please make sure to use --enable-compat-libs only
14149 during a transitional period!
14151 * The .include syntax has been deprecated and is not documented
14152 anymore. Drop-in files in .d directories should be used instead.
14154 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
14155 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14156 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
14157 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
14158 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
14159 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
14160 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
14161 Zbigniew Jędrzejewski-Szmek
14163 — Berlin, 2014-02-24
14167 * A new component "systemd-networkd" has been added that can
14168 be used to configure local network interfaces statically or
14169 via DHCP. It is capable of bringing up bridges, VLANs, and
14170 bonding. Currently, no hook-ups for interactive network
14171 configuration are provided. Use this for your initrd,
14172 container, embedded, or server setup if you need a simple,
14173 yet powerful, network configuration solution. This
14174 configuration subsystem is quite nifty, as it allows wildcard
14175 hotplug matching in interfaces. For example, with a single
14176 configuration snippet, you can configure that all Ethernet
14177 interfaces showing up are automatically added to a bridge,
14178 or similar. It supports link-sensing and more.
14180 * A new tool "systemd-socket-proxyd" has been added which can
14181 act as a bidirectional proxy for TCP sockets. This is
14182 useful for adding socket activation support to services that
14183 do not actually support socket activation, including virtual
14184 machines and the like.
14186 * Add a new tool to save/restore rfkill state on
14189 * Save/restore state of keyboard backlights in addition to
14190 display backlights on shutdown/boot.
14192 * udev learned a new SECLABEL{} construct to label device
14193 nodes with a specific security label when they appear. For
14194 now, only SECLABEL{selinux} is supported, but the syntax is
14195 prepared for additional security frameworks.
14197 * udev gained a new scheme to configure link-level attributes
14198 from files in /etc/systemd/network/*.link. These files can
14199 match against MAC address, device path, driver name and type,
14200 and will apply attributes like the naming policy, link speed,
14201 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
14202 address assignment policy (randomized, …).
14204 * The configuration of network interface naming rules for
14205 "permanent interface names" has changed: a new NamePolicy=
14206 setting in the [Link] section of .link files determines the
14207 priority of possible naming schemes (onboard, slot, MAC,
14208 path). The default value of this setting is determined by
14209 /usr/lib/net/links/99-default.link. Old
14210 80-net-name-slot.rules udev configuration file has been
14211 removed, so local configuration overriding this file should
14212 be adapted to override 99-default.link instead.
14214 * When the User= switch is used in a unit file, also
14215 initialize $SHELL= based on the user database entry.
14217 * systemd no longer depends on libdbus. All communication is
14218 now done with sd-bus, systemd's low-level bus library
14221 * kdbus support has been added to PID 1 itself. When kdbus is
14222 enabled, this causes PID 1 to set up the system bus and
14223 enable support for a new ".busname" unit type that
14224 encapsulates bus name activation on kdbus. It works a little
14225 bit like ".socket" units, except for bus names. A new
14226 generator has been added that converts classic dbus1 service
14227 activation files automatically into native systemd .busname
14228 and .service units.
14230 * sd-bus: add a light-weight vtable implementation that allows
14231 defining objects on the bus with a simple static const
14232 vtable array of its methods, signals and properties.
14234 * systemd will not generate or install static dbus
14235 introspection data anymore to /usr/share/dbus-1/interfaces,
14236 as the precise format of these files is unclear, and
14237 nothing makes use of it.
14239 * A proxy daemon is now provided to proxy clients connecting
14240 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
14241 compatibility with classic D-Bus.
14243 * A bus driver implementation has been added that supports the
14244 classic D-Bus bus driver calls on kdbus, also for
14245 compatibility purposes.
14247 * A new API "sd-event.h" has been added that implements a
14248 minimal event loop API built around epoll. It provides a
14249 couple of features that direct epoll usage is lacking:
14250 prioritization of events, scales to large numbers of timer
14251 events, per-event timer slack (accuracy), system-wide
14252 coalescing of timer events, exit handlers, watchdog
14253 supervision support using systemd's sd_notify() API, child
14256 * A new API "sd-rntl.h" has been added that provides an API
14257 around the route netlink interface of the kernel, similar in
14258 style to "sd-bus.h".
14260 * A new API "sd-dhcp-client.h" has been added that provides a
14261 small DHCPv4 client-side implementation. This is used by
14262 "systemd-networkd".
14264 * There is a new kernel command line option
14265 "systemd.restore_state=0|1". When set to "0", none of the
14266 systemd tools will restore saved runtime state to hardware
14267 devices. More specifically, the rfkill and backlight states
14270 * The FsckPassNo= compatibility option in mount/service units
14271 has been removed. The fstab generator will now add the
14272 necessary dependencies automatically, and does not require
14273 PID1's support for that anymore.
14275 * journalctl gained a new switch, --list-boots, that lists
14276 recent boots with their times and boot IDs.
14278 * The various tools like systemctl, loginctl, timedatectl,
14279 busctl, systemd-run, … have gained a new switch "-M" to
14280 connect to a specific, local OS container (as direct
14281 connection, without requiring SSH). This works on any
14282 container that is registered with machined, such as those
14283 created by libvirt-lxc or nspawn.
14285 * systemd-run and systemd-analyze also gained support for "-H"
14286 to connect to remote hosts via SSH. This is particularly
14287 useful for systemd-run because it enables queuing of jobs
14288 onto remote systems.
14290 * machinectl gained a new command "login" to open a getty
14291 login in any local container. This works with any container
14292 that is registered with machined (such as those created by
14293 libvirt-lxc or nspawn), and which runs systemd inside.
14295 * machinectl gained a new "reboot" command that may be used to
14296 trigger a reboot on a specific container that is registered
14297 with machined. This works on any container that runs an init
14298 system of some kind.
14300 * systemctl gained a new "list-timers" command to print a nice
14301 listing of installed timer units with the times they elapse
14304 * Alternative reboot() parameters may now be specified on the
14305 "systemctl reboot" command line and are passed to the
14306 reboot() system call.
14308 * systemctl gained a new --job-mode= switch to configure the
14309 mode to queue a job with. This is a more generic version of
14310 --fail, --irreversible, and --ignore-dependencies, which are
14311 still available but not advertised anymore.
14313 * /etc/systemd/system.conf gained new settings to configure
14314 various default timeouts of units, as well as the default
14315 start limit interval and burst. These may still be overridden
14318 * PID1 will now export on the bus profile data of the security
14319 policy upload process (such as the SELinux policy upload to
14322 * journald: when forwarding logs to the console, include
14323 timestamps (following the setting in
14324 /sys/module/printk/parameters/time).
14326 * OnCalendar= in timer units now understands the special
14327 strings "yearly" and "annually". (Both are equivalent)
14329 * The accuracy of timer units is now configurable with the new
14330 AccuracySec= setting. It defaults to 1min.
14332 * A new dependency type JoinsNamespaceOf= has been added that
14333 allows running two services within the same /tmp and network
14334 namespace, if PrivateNetwork= or PrivateTmp= are used.
14336 * A new command "cat" has been added to systemctl. It outputs
14337 the original unit file of a unit, and concatenates the
14338 contents of additional "drop-in" unit file snippets, so that
14339 the full configuration is shown.
14341 * systemctl now supports globbing on the various "list-xyz"
14342 commands, like "list-units" or "list-sockets", as well as on
14343 those commands which take multiple unit names.
14345 * journalctl's --unit= switch gained support for globbing.
14347 * All systemd daemons now make use of the watchdog logic so
14348 that systemd automatically notices when they hang.
14350 * If the $container_ttys environment variable is set,
14351 getty-generator will automatically spawn a getty for each
14352 listed tty. This is useful for container managers to request
14353 login gettys to be spawned on as many ttys as needed.
14355 * %h, %s, %U specifier support is not available anymore when
14356 used in unit files for PID 1. This is because NSS calls are
14357 not safe from PID 1. They stay available for --user
14358 instances of systemd, and as special case for the root user.
14360 * loginctl gained a new "--no-legend" switch to turn off output
14361 of the legend text.
14363 * The "sd-login.h" API gained three new calls:
14364 sd_session_is_remote(), sd_session_get_remote_user(),
14365 sd_session_get_remote_host() to query information about
14368 * The udev hardware database now also carries vendor/product
14369 information of SDIO devices.
14371 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
14372 determine whether watchdog notifications are requested by
14373 the system manager.
14375 * Socket-activated per-connection services now include a
14376 short description of the connection parameters in the
14379 * tmpfiles gained a new "--boot" option. When this is not used,
14380 only lines where the command character is not suffixed with
14381 "!" are executed. When this option is specified, those
14382 options are executed too. This partitions tmpfiles
14383 directives into those that can be safely executed at any
14384 time, and those which should be run only at boot (for
14385 example, a line that creates /run/nologin).
14387 * A new API "sd-resolve.h" has been added which provides a simple
14388 asynchronous wrapper around glibc NSS hostname resolution
14389 calls, such as getaddrinfo(). In contrast to glibc's
14390 getaddrinfo_a(), it does not use signals. In contrast to most
14391 other asynchronous name resolution libraries, this one does
14392 not reimplement DNS, but reuses NSS, so that alternate
14393 hostname resolution systems continue to work, such as mDNS,
14394 LDAP, etc. This API is based on libasyncns, but it has been
14395 cleaned up for inclusion in systemd.
14397 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
14398 "sd-daemon.h" are no longer found in individual libraries
14399 libsystemd-journal.so, libsystemd-login.so,
14400 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
14401 merged them into a single library, libsystemd.so, which
14402 provides all symbols. The reason for this is cyclic
14403 dependencies, as these libraries tend to use each other's
14404 symbols. So far, we have managed to workaround that by linking
14405 a copy of a good part of our code into each of these
14406 libraries again and again, which, however, makes certain
14407 things hard to do, like sharing static variables. Also, it
14408 substantially increases footprint. With this change, there
14409 is only one library for the basic APIs systemd
14410 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
14411 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
14412 library as well, however are subject to the --enable-kdbus
14413 switch (see below). Note that "sd-dhcp-client.h" is not part
14414 of this library (this is because it only consumes, never
14415 provides, services of/to other APIs). To make the transition
14416 easy from the separate libraries to the unified one, we
14417 provide the --enable-compat-libs compile-time switch which
14418 will generate stub libraries that are compatible with the
14419 old ones but redirect all calls to the new one.
14421 * All of the kdbus logic and the new APIs "sd-bus.h",
14422 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
14423 and "sd-utf8.h" are compile-time optional via the
14424 "--enable-kdbus" switch, and they are not compiled in by
14425 default. To make use of kdbus, you have to explicitly enable
14426 the switch. Note however, that neither the kernel nor the
14427 userspace API for all of this is considered stable yet. We
14428 want to maintain the freedom to still change the APIs for
14429 now. By specifying this build-time switch, you acknowledge
14430 that you are aware of the instability of the current
14433 * Also, note that while kdbus is pretty much complete,
14434 it lacks one thing: proper policy support. This means you
14435 can build a fully working system with all features; however,
14436 it will be highly insecure. Policy support will be added in
14437 one of the next releases, at the same time that we will
14438 declare the APIs stable.
14440 * When the kernel command line argument "kdbus" is specified,
14441 systemd will automatically load the kdbus.ko kernel module. At
14442 this stage of development, it is only useful for testing kdbus
14443 and should not be used in production. Note: if "--enable-kdbus"
14444 is specified, and the kdbus.ko kernel module is available, and
14445 "kdbus" is added to the kernel command line, the entire system
14446 runs with kdbus instead of dbus-daemon, with the above mentioned
14447 problem of missing the system policy enforcement. Also a future
14448 version of kdbus.ko or a newer systemd will not be compatible with
14449 each other, and will unlikely be able to boot the machine if only
14450 one of them is updated.
14452 * systemctl gained a new "import-environment" command which
14453 uploads the caller's environment (or parts thereof) into the
14454 service manager so that it is inherited by services started
14455 by the manager. This is useful to upload variables like
14456 $DISPLAY into the user service manager.
14458 * A new PrivateDevices= switch has been added to service units
14459 which allows running a service with a namespaced /dev
14460 directory that does not contain any device nodes for
14461 physical devices. More specifically, it only includes devices
14462 such as /dev/null, /dev/urandom, and /dev/zero which are API
14465 * logind has been extended to support behaviour like VT
14466 switching on seats that do not support a VT. This makes
14467 multi-session available on seats that are not the first seat
14468 (seat0), and on systems where kernel support for VTs has
14469 been disabled at compile-time.
14471 * If a process holds a delay lock for system sleep or shutdown
14472 and fails to release it in time, we will now log its
14473 identity. This makes it easier to identify processes that
14474 cause slow suspends or power-offs.
14476 * When parsing /etc/crypttab, support for a new key-slot=
14477 option as supported by Debian is added. It allows indicating
14478 which LUKS slot to use on disk, speeding up key loading.
14480 * The sd_journal_sendv() API call has been checked and
14481 officially declared to be async-signal-safe so that it may
14482 be invoked from signal handlers for logging purposes.
14484 * Boot-time status output is now enabled automatically after a
14485 short timeout if boot does not progress, in order to give
14486 the user an indication what she or he is waiting for.
14488 * The boot-time output has been improved to show how much time
14489 remains until jobs expire.
14491 * The KillMode= switch in service units gained a new possible
14492 value "mixed". If set, and the unit is shut down, then the
14493 initial SIGTERM signal is sent only to the main daemon
14494 process, while the following SIGKILL signal is sent to
14495 all remaining processes of the service.
14497 * When a scope unit is registered, a new property "Controller"
14498 may be set. If set to a valid bus name, systemd will send a
14499 RequestStop() signal to this name when it would like to shut
14500 down the scope. This may be used to hook manager logic into
14501 the shutdown logic of scope units. Also, scope units may now
14502 be put in a special "abandoned" state, in which case the
14503 manager process which created them takes no further
14504 responsibilities for it.
14506 * When reading unit files, systemd will now verify
14507 the access mode of these files, and warn about certain
14508 suspicious combinations. This has been added to make it
14509 easier to track down packaging bugs where unit files are
14510 marked executable or world-writable.
14512 * systemd-nspawn gained a new "--setenv=" switch to set
14513 container-wide environment variables. The similar option in
14514 systemd-activate was renamed from "--environment=" to
14515 "--setenv=" for consistency.
14517 * systemd-nspawn has been updated to create a new kdbus domain
14518 for each container that is invoked, thus allowing each
14519 container to have its own set of system and user buses,
14520 independent of the host.
14522 * systemd-nspawn gained a new --drop-capability= switch to run
14523 the container with less capabilities than the default. Both
14524 --drop-capability= and --capability= now take the special
14525 string "all" for dropping or keeping all capabilities.
14527 * systemd-nspawn gained new switches for executing containers
14528 with specific SELinux labels set.
14530 * systemd-nspawn gained a new --quiet switch to not generate
14531 any additional output but the container's own console
14534 * systemd-nspawn gained a new --share-system switch to run a
14535 container without PID namespacing enabled.
14537 * systemd-nspawn gained a new --register= switch to control
14538 whether the container is registered with systemd-machined or
14539 not. This is useful for containers that do not run full
14540 OS images, but only specific apps.
14542 * systemd-nspawn gained a new --keep-unit which may be used
14543 when invoked as the only program from a service unit, and
14544 results in registration of the unit service itself in
14545 systemd-machined, instead of a newly opened scope unit.
14547 * systemd-nspawn gained a new --network-interface= switch for
14548 moving arbitrary interfaces to the container. The new
14549 --network-veth switch creates a virtual Ethernet connection
14550 between host and container. The new --network-bridge=
14551 switch then allows assigning the host side of this virtual
14552 Ethernet connection to a bridge device.
14554 * systemd-nspawn gained a new --personality= switch for
14555 setting the kernel personality for the container. This is
14556 useful when running a 32-bit container on a 64-bit host. A
14557 similar option Personality= is now also available for service
14560 * logind will now also track a "Desktop" identifier for each
14561 session which encodes the desktop environment of it. This is
14562 useful for desktop environments that want to identify
14563 multiple running sessions of itself easily.
14565 * A new SELinuxContext= setting for service units has been
14566 added that allows setting a specific SELinux execution
14567 context for a service.
14569 * Most systemd client tools will now honour $SYSTEMD_LESS for
14570 settings of the "less" pager. By default, these tools will
14571 override $LESS to allow certain operations to work, such as
14572 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
14573 influence this logic.
14575 * systemd's "seccomp" hook-up has been changed to make use of
14576 the libseccomp library instead of using its own
14577 implementation. This has benefits for portability among
14580 * For usage together with SystemCallFilter=, a new
14581 SystemCallErrorNumber= setting has been introduced that
14582 allows configuration of a system error number to be returned
14583 on filtered system calls, instead of immediately killing the
14584 process. Also, SystemCallArchitectures= has been added to
14585 limit access to system calls of a particular architecture
14586 (in order to turn off support for unused secondary
14587 architectures). There is also a global
14588 SystemCallArchitectures= setting in system.conf now to turn
14589 off support for non-native system calls system-wide.
14591 * systemd requires a kernel with a working name_to_handle_at(),
14592 please see the kernel config requirements in the README file.
14594 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
14595 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
14596 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
14597 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
14598 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
14599 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
14600 Elia Pinto, Florian Weimer, George McCollister, Goffredo
14601 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
14602 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
14603 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
14604 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
14605 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
14606 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
14607 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
14608 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
14609 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
14610 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
14611 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
14612 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
14613 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
14614 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
14615 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
14616 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
14617 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
14619 — Berlin, 2014-02-20
14623 * logind has gained support for facilitating privileged input
14624 and drm device access for unprivileged clients. This work is
14625 useful to allow Wayland display servers (and similar
14626 programs, such as kmscon) to run under the user's ID and
14627 access input and drm devices which are normally
14628 protected. When this is used (and the kernel is new enough)
14629 logind will "mute" IO on the file descriptors passed to
14630 Wayland as long as it is in the background and "unmute" it
14631 if it returns into the foreground. This allows secure
14632 session switching without allowing background sessions to
14633 eavesdrop on input and display data. This also introduces
14634 session switching support if VT support is turned off in the
14635 kernel, and on seats that are not seat0.
14637 * A new kernel command line option luks.options= is understood
14638 now which allows specifying LUKS options for usage for LUKS
14639 encrypted partitions specified with luks.uuid=.
14641 * tmpfiles.d(5) snippets may now use specifier expansion in
14642 path names. More specifically %m, %b, %H, %v, are now
14643 replaced by the local machine id, boot id, hostname, and
14644 kernel version number.
14646 * A new tmpfiles.d(5) command "m" has been introduced which
14647 may be used to change the owner/group/access mode of a file
14648 or directory if it exists, but do nothing if it does not.
14650 * This release removes high-level support for the
14651 MemorySoftLimit= cgroup setting. The underlying kernel
14652 cgroup attribute memory.soft_limit= is currently badly
14653 designed and likely to be removed from the kernel API in its
14654 current form, hence we should not expose it for now.
14656 * The memory.use_hierarchy cgroup attribute is now enabled for
14657 all cgroups systemd creates in the memory cgroup
14658 hierarchy. This option is likely to be come the built-in
14659 default in the kernel anyway, and the non-hierarchical mode
14660 never made much sense in the intrinsically hierarchical
14663 * A new field _SYSTEMD_SLICE= is logged along with all journal
14664 messages containing the slice a message was generated
14665 from. This is useful to allow easy per-customer filtering of
14666 logs among other things.
14668 * systemd-journald will no longer adjust the group of journal
14669 files it creates to the "systemd-journal" group. Instead we
14670 rely on the journal directory to be owned by the
14671 "systemd-journal" group, and its setgid bit set, so that the
14672 kernel file system layer will automatically enforce that
14673 journal files inherit this group assignment. The reason for
14674 this change is that we cannot allow NSS look-ups from
14675 journald which would be necessary to resolve
14676 "systemd-journal" to a numeric GID, because this might
14677 create deadlocks if NSS involves synchronous queries to
14678 other daemons (such as nscd, or sssd) which in turn are
14679 logging clients of journald and might block on it, which
14680 would then dead lock. A tmpfiles.d(5) snippet included in
14681 systemd will make sure the setgid bit and group are
14682 properly set on the journal directory if it exists on every
14683 boot. However, we recommend adjusting it manually after
14684 upgrades too (or from RPM scriptlets), so that the change is
14685 not delayed until next reboot.
14687 * Backlight and random seed files in /var/lib/ have moved into
14688 the /var/lib/systemd/ directory, in order to centralize all
14689 systemd generated files in one directory.
14691 * Boot time performance measurements (as displayed by
14692 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
14693 performance information if that's available to determine how
14694 much time BIOS and boot loader initialization required. With
14695 a sufficiently new BIOS you hence no longer need to boot
14696 with Gummiboot to get access to such information.
14698 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
14699 Cristian Rodríguez, Dave Reisner, David Herrmann, David
14700 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
14701 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
14702 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
14703 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
14704 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
14706 — Berlin, 2013-10-02
14710 * The Restart= option for services now understands a new
14711 on-watchdog setting, which will restart the service
14712 automatically if the service stops sending out watchdog keep
14713 alive messages (as configured with WatchdogSec=).
14715 * The getty generator (which is responsible for bringing up a
14716 getty on configured serial consoles) will no longer only
14717 start a getty on the primary kernel console but on all
14718 others, too. This makes the order in which console= is
14719 specified on the kernel command line less important.
14721 * libsystemd-logind gained a new sd_session_get_vt() call to
14722 retrieve the VT number of a session.
14724 * If the option "tries=0" is set for an entry of /etc/crypttab
14725 its passphrase is queried indefinitely instead of any
14726 maximum number of tries.
14728 * If a service with a configure PID file terminates its PID
14729 file will now be removed automatically if it still exists
14730 afterwards. This should put an end to stale PID files.
14732 * systemd-run will now also take relative binary path names
14733 for execution and no longer insists on absolute paths.
14735 * InaccessibleDirectories= and ReadOnlyDirectories= now take
14736 paths that are optionally prefixed with "-" to indicate that
14737 it should not be considered a failure if they do not exist.
14739 * journalctl -o (and similar commands) now understands a new
14740 output mode "short-precise", it is similar to "short" but
14741 shows timestamps with usec accuracy.
14743 * The option "discard" (as known from Debian) is now
14744 synonymous to "allow-discards" in /etc/crypttab. In fact,
14745 "discard" is preferred now (since it is easier to remember
14748 * Some licensing clean-ups were made, so that more code is now
14749 LGPL-2.1 licensed than before.
14751 * A minimal tool to save/restore the display backlight
14752 brightness across reboots has been added. It will store the
14753 backlight setting as late as possible at shutdown, and
14754 restore it as early as possible during reboot.
14756 * A logic to automatically discover and enable home and swap
14757 partitions on GPT disks has been added. With this in place
14758 /etc/fstab becomes optional for many setups as systemd can
14759 discover certain partitions located on the root disk
14760 automatically. Home partitions are recognized under their
14761 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
14762 partitions are recognized under their GPT type ID
14763 0657fd6da4ab43c484e50933c84b4f4f.
14765 * systemd will no longer pass any environment from the kernel
14766 or initrd to system services. If you want to set an
14767 environment for all services, do so via the kernel command
14768 line systemd.setenv= assignment.
14770 * The systemd-sysctl tool no longer natively reads the file
14771 /etc/sysctl.conf. If desired, the file should be symlinked
14772 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
14773 legacy support by a symlink rather than built-in code, it
14774 also makes the otherwise hidden order of application of the
14775 different files visible. (Note that this partly reverts to a
14776 pre-198 application order of sysctl knobs!)
14778 * The "systemctl set-log-level" and "systemctl dump" commands
14779 have been moved to systemd-analyze.
14781 * systemd-run learned the new --remain-after-exit switch,
14782 which causes the scope unit not to be cleaned up
14783 automatically after the process terminated.
14785 * tmpfiles learned a new --exclude-prefix= switch to exclude
14786 certain paths from operation.
14788 * journald will now automatically flush all messages to disk
14789 as soon as a message at the log level CRIT, ALERT or EMERG
14792 Contributions from: Andrew Cook, Brandon Philips, Christian
14793 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
14794 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
14795 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
14796 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
14797 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
14798 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
14799 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
14800 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
14801 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
14802 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
14803 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
14804 William Giokas, Zbigniew Jędrzejewski-Szmek
14806 — Berlin, 2013-09-13
14810 * The documentation has been updated to cover the various new
14811 concepts introduced with 205.
14813 * Unit files now understand the new %v specifier which
14814 resolves to the kernel version string as returned by "uname
14817 * systemctl now supports filtering the unit list output by
14818 load state, active state and sub state, using the new
14819 --state= parameter.
14821 * "systemctl status" will now show the results of the
14822 condition checks (like ConditionPathExists= and similar) of
14823 the last start attempts of the unit. They are also logged to
14826 * "journalctl -b" may now be used to look for boot output of a
14827 specific boot. Try "journalctl -b -1" for the previous boot,
14828 but the syntax is substantially more powerful.
14830 * "journalctl --show-cursor" has been added which prints the
14831 cursor string the last shown log line. This may then be used
14832 with the new "journalctl --after-cursor=" switch to continue
14833 browsing logs from that point on.
14835 * "journalctl --force" may now be used to force regeneration
14838 * Creation of "dead" device nodes has been moved from udev
14839 into kmod and tmpfiles. Previously, udev would read the kmod
14840 databases to pre-generate dead device nodes based on meta
14841 information contained in kernel modules, so that these would
14842 be auto-loaded on access rather then at boot. As this
14843 does not really have much to do with the exposing actual
14844 kernel devices to userspace this has always been slightly
14845 alien in the udev codebase. Following the new scheme kmod
14846 will now generate a runtime snippet for tmpfiles from the
14847 module meta information and it now is tmpfiles' job to the
14848 create the nodes. This also allows overriding access and
14849 other parameters for the nodes using the usual tmpfiles
14850 facilities. As side effect this allows us to remove the
14851 CAP_SYS_MKNOD capability bit from udevd entirely.
14853 * logind's device ACLs may now be applied to these "dead"
14854 devices nodes too, thus finally allowing managed access to
14855 devices such as /dev/snd/sequencer without loading the
14856 backing module right-away.
14858 * A new RPM macro has been added that may be used to apply
14859 tmpfiles configuration during package installation.
14861 * systemd-detect-virt and ConditionVirtualization= now can
14862 detect User-Mode-Linux machines (UML).
14864 * journald will now implicitly log the effective capabilities
14865 set of processes in the message metadata.
14867 * systemd-cryptsetup has gained support for TrueCrypt volumes.
14869 * The initrd interface has been simplified (more specifically,
14870 support for passing performance data via environment
14871 variables and fsck results via files in /run has been
14872 removed). These features were non-essential, and are
14873 nowadays available in a much nicer way by having systemd in
14874 the initrd serialize its state and have the hosts systemd
14875 deserialize it again.
14877 * The udev "keymap" data files and tools to apply keyboard
14878 specific mappings of scan to key codes, and force-release
14879 scan code lists have been entirely replaced by a udev
14880 "keyboard" builtin and a hwdb data file.
14882 * systemd will now honour the kernel's "quiet" command line
14883 argument also during late shutdown, resulting in a
14884 completely silent shutdown when used.
14886 * There's now an option to control the SO_REUSEPORT socket
14887 option in .socket units.
14889 * Instance units will now automatically get a per-template
14890 subslice of system.slice unless something else is explicitly
14891 configured. For example, instances of sshd@.service will now
14892 implicitly be placed in system-sshd.slice rather than
14893 system.slice as before.
14895 * Test coverage support may now be enabled at build time.
14897 Contributions from: Dave Reisner, Frederic Crozat, Harald
14898 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
14899 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
14900 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
14901 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
14902 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
14903 Giokas, Zbigniew Jędrzejewski-Szmek
14905 — Berlin, 2013-07-23
14909 * Two new unit types have been introduced:
14911 Scope units are very similar to service units, however, are
14912 created out of pre-existing processes — instead of PID 1
14913 forking off the processes. By using scope units it is
14914 possible for system services and applications to group their
14915 own child processes (worker processes) in a powerful way
14916 which then maybe used to organize them, or kill them
14917 together, or apply resource limits on them.
14919 Slice units may be used to partition system resources in an
14920 hierarchical fashion and then assign other units to them. By
14921 default there are now three slices: system.slice (for all
14922 system services), user.slice (for all user sessions),
14923 machine.slice (for VMs and containers).
14925 Slices and scopes have been introduced primarily in
14926 context of the work to move cgroup handling to a
14927 single-writer scheme, where only PID 1
14928 creates/removes/manages cgroups.
14930 * There's a new concept of "transient" units. In contrast to
14931 normal units these units are created via an API at runtime,
14932 not from configuration from disk. More specifically this
14933 means it is now possible to run arbitrary programs as
14934 independent services, with all execution parameters passed
14935 in via bus APIs rather than read from disk. Transient units
14936 make systemd substantially more dynamic then it ever was,
14937 and useful as a general batch manager.
14939 * logind has been updated to make use of scope and slice units
14940 for managing user sessions. As a user logs in he will get
14941 his own private slice unit, to which all sessions are added
14942 as scope units. We also added support for automatically
14943 adding an instance of user@.service for the user into the
14944 slice. Effectively logind will no longer create cgroup
14945 hierarchies on its own now, it will defer entirely to PID 1
14946 for this by means of scope, service and slice units. Since
14947 user sessions this way become entities managed by PID 1
14948 the output of "systemctl" is now a lot more comprehensive.
14950 * A new mini-daemon "systemd-machined" has been added which
14951 may be used by virtualization managers to register local
14952 VMs/containers. nspawn has been updated accordingly, and
14953 libvirt will be updated shortly. machined will collect a bit
14954 of meta information about the VMs/containers, and assign
14955 them their own scope unit (see above). The collected
14956 meta-data is then made available via the "machinectl" tool,
14957 and exposed in "ps" and similar tools. machined/machinectl
14958 is compile-time optional.
14960 * As discussed earlier, the low-level cgroup configuration
14961 options ControlGroup=, ControlGroupModify=,
14962 ControlGroupPersistent=, ControlGroupAttribute= have been
14963 removed. Please use high-level attribute settings instead as
14964 well as slice units.
14966 * A new bus call SetUnitProperties() has been added to alter
14967 various runtime parameters of a unit. This is primarily
14968 useful to alter cgroup parameters dynamically in a nice way,
14969 but will be extended later on to make more properties
14970 modifiable at runtime. systemctl gained a new set-properties
14971 command that wraps this call.
14973 * A new tool "systemd-run" has been added which can be used to
14974 run arbitrary command lines as transient services or scopes,
14975 while configuring a number of settings via the command
14976 line. This tool is currently very basic, however already
14977 very useful. We plan to extend this tool to even allow
14978 queuing of execution jobs with time triggers from the
14979 command line, similar in fashion to "at".
14981 * nspawn will now inform the user explicitly that kernels with
14982 audit enabled break containers, and suggest the user to turn
14985 * Support for detecting the IMA and AppArmor security
14986 frameworks with ConditionSecurity= has been added.
14988 * journalctl gained a new "-k" switch for showing only kernel
14989 messages, mimicking dmesg output; in addition to "--user"
14990 and "--system" switches for showing only user's own logs
14993 * systemd-delta can now show information about drop-in
14994 snippets extending unit files.
14996 * libsystemd-bus has been substantially updated but is still
14997 not available as public API.
14999 * systemd will now look for the "debug" argument on the kernel
15000 command line and enable debug logging, similar to what
15001 "systemd.log_level=debug" already did before.
15003 * "systemctl set-default", "systemctl get-default" has been
15004 added to configure the default.target symlink, which
15005 controls what to boot into by default.
15007 * "systemctl set-log-level" has been added as a convenient
15008 way to raise and lower systemd logging threshold.
15010 * "systemd-analyze plot" will now show the time the various
15011 generators needed for execution, as well as information
15012 about the unit file loading.
15014 * libsystemd-journal gained a new sd_journal_open_files() call
15015 for opening specific journal files. journactl also gained a
15016 new switch to expose this new functionality. Previously we
15017 only supported opening all files from a directory, or all
15018 files from the system, as opening individual files only is
15019 racy due to journal file rotation.
15021 * systemd gained the new DefaultEnvironment= setting in
15022 /etc/systemd/system.conf to set environment variables for
15025 * If a privileged process logs a journal message with the
15026 OBJECT_PID= field set, then journald will automatically
15027 augment this with additional OBJECT_UID=, OBJECT_GID=,
15028 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
15029 system services want to log events about specific client
15030 processes. journactl/systemctl has been updated to make use
15031 of this information if all log messages regarding a specific
15034 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
15035 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
15036 Reisner, David Coppa, David King, David Strauss, Eelco
15037 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
15038 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
15039 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
15040 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
15041 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
15042 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
15043 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
15044 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
15045 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
15046 Łukasz Stelmach, 장동준
15050 * The Python bindings gained some minimal support for the APIs
15051 exposed by libsystemd-logind.
15053 * ConditionSecurity= gained support for detecting SMACK. Since
15054 this condition already supports SELinux and AppArmor we only
15055 miss IMA for this. Patches welcome!
15057 Contributions from: Karol Lewandowski, Lennart Poettering,
15058 Zbigniew Jędrzejewski-Szmek
15062 * systemd-nspawn will now create /etc/resolv.conf if
15063 necessary, before bind-mounting the host's file onto it.
15065 * systemd-nspawn will now store meta information about a
15066 container on the container's cgroup as extended attribute
15067 fields, including the root directory.
15069 * The cgroup hierarchy has been reworked in many ways. All
15070 objects any of the components systemd creates in the cgroup
15071 tree are now suffixed. More specifically, user sessions are
15072 now placed in cgroups suffixed with ".session", users in
15073 cgroups suffixed with ".user", and nspawn containers in
15074 cgroups suffixed with ".nspawn". Furthermore, all cgroup
15075 names are now escaped in a simple scheme to avoid collision
15076 of userspace object names with kernel filenames. This work
15077 is preparation for making these objects relocatable in the
15078 cgroup tree, in order to allow easy resource partitioning of
15079 these objects without causing naming conflicts.
15081 * systemctl list-dependencies gained the new switches
15082 --plain, --reverse, --after and --before.
15084 * systemd-inhibit now shows the process name of processes that
15085 have taken an inhibitor lock.
15087 * nss-myhostname will now also resolve "localhost"
15088 implicitly. This makes /etc/hosts an optional file and
15089 nicely handles that on IPv6 ::1 maps to both "localhost" and
15090 the local hostname.
15092 * libsystemd-logind.so gained a new call
15093 sd_get_machine_names() to enumerate running containers and
15094 VMs (currently only supported by very new libvirt and
15095 nspawn). sd_login_monitor can now be used to watch
15096 VMs/containers coming and going.
15098 * .include is not allowed recursively anymore, and only in
15099 unit files. Usually it is better to use drop-in snippets in
15100 .d/*.conf anyway, as introduced with systemd 198.
15102 * systemd-analyze gained a new "critical-chain" command that
15103 determines the slowest chain of units run during system
15104 boot-up. It is very useful for tracking down where
15105 optimizing boot time is the most beneficial.
15107 * systemd will no longer allow manipulating service paths in
15108 the name=systemd:/system cgroup tree using ControlGroup= in
15109 units. (But is still fine with it in all other dirs.)
15111 * There's a new systemd-nspawn@.service service file that may
15112 be used to easily run nspawn containers as system
15113 services. With the container's root directory in
15114 /var/lib/container/foobar it is now sufficient to run
15115 "systemctl start systemd-nspawn@foobar.service" to boot it.
15117 * systemd-cgls gained a new parameter "--machine" to list only
15118 the processes within a certain container.
15120 * ConditionSecurity= now can check for "apparmor". We still
15121 are lacking checks for SMACK and IMA for this condition
15122 check though. Patches welcome!
15124 * A new configuration file /etc/systemd/sleep.conf has been
15125 added that may be used to configure which kernel operation
15126 systemd is supposed to execute when "suspend", "hibernate"
15127 or "hybrid-sleep" is requested. This makes the new kernel
15128 "freeze" state accessible to the user.
15130 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
15131 the passed argument if applicable.
15133 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
15134 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
15135 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
15136 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
15137 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
15138 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
15139 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
15144 * The output of 'systemctl list-jobs' got some polishing. The
15145 '--type=' argument may now be passed more than once. A new
15146 command 'systemctl list-sockets' has been added which shows
15147 a list of kernel sockets systemd is listening on with the
15148 socket units they belong to, plus the units these socket
15151 * The experimental libsystemd-bus library got substantial
15152 updates to work in conjunction with the (also experimental)
15153 kdbus kernel project. It works well enough to exchange
15154 messages with some sophistication. Note that kdbus is not
15155 ready yet, and the library is mostly an elaborate test case
15156 for now, and not installable.
15158 * systemd gained a new unit 'systemd-static-nodes.service'
15159 that generates static device nodes earlier during boot, and
15160 can run in conjunction with udev.
15162 * libsystemd-login gained a new call sd_pid_get_user_unit()
15163 to retrieve the user systemd unit a process is running
15164 in. This is useful for systems where systemd is used as
15167 * systemd-nspawn now places all containers in the new /machine
15168 top-level cgroup directory in the name=systemd
15169 hierarchy. libvirt will soon do the same, so that we get a
15170 uniform separation of /system, /user and /machine for system
15171 services, user processes and containers/virtual
15172 machines. This new cgroup hierarchy is also useful to stick
15173 stable names to specific container instances, which can be
15174 recognized later this way (this name may be controlled
15175 via systemd-nspawn's new -M switch). libsystemd-login also
15176 gained a new call sd_pid_get_machine_name() to retrieve the
15177 name of the container/VM a specific process belongs to.
15179 * bootchart can now store its data in the journal.
15181 * libsystemd-journal gained a new call
15182 sd_journal_add_conjunction() for AND expressions to the
15183 matching logic. This can be used to express more complex
15184 logical expressions.
15186 * journactl can now take multiple --unit= and --user-unit=
15189 * The cryptsetup logic now understands the "luks.key=" kernel
15190 command line switch for specifying a file to read the
15191 decryption key from. Also, if a configured key file is not
15192 found the tool will now automatically fall back to prompting
15195 * Python systemd.journal module was updated to wrap recently
15196 added functions from libsystemd-journal. The interface was
15197 changed to bring the low level interface in s.j._Reader
15198 closer to the C API, and the high level interface in
15199 s.j.Reader was updated to wrap and convert all data about
15202 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
15203 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
15204 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
15205 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
15206 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
15207 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15211 * journalctl --update-catalog now understands a new --root=
15212 option to operate on catalogs found in a different root
15215 * During shutdown after systemd has terminated all running
15216 services a final killing loop kills all remaining left-over
15217 processes. We will now print the name of these processes
15218 when we send SIGKILL to them, since this usually indicates a
15221 * If /etc/crypttab refers to password files stored on
15222 configured mount points automatic dependencies will now be
15223 generated to ensure the specific mount is established first
15224 before the key file is attempted to be read.
15226 * 'systemctl status' will now show information about the
15227 network sockets a socket unit is listening on.
15229 * 'systemctl status' will also shown information about any
15230 drop-in configuration file for units. (Drop-In configuration
15231 files in this context are files such as
15232 /etc/systemd/system/foobar.service.d/*.conf)
15234 * systemd-cgtop now optionally shows summed up CPU times of
15235 cgroups. Press '%' while running cgtop to switch between
15236 percentage and absolute mode. This is useful to determine
15237 which cgroups use up the most CPU time over the entire
15238 runtime of the system. systemd-cgtop has also been updated
15239 to be 'pipeable' for processing with further shell tools.
15241 * 'hostnamectl set-hostname' will now allow setting of FQDN
15244 * The formatting and parsing of time span values has been
15245 changed. The parser now understands fractional expressions
15246 such as "5.5h". The formatter will now output fractional
15247 expressions for all time spans under 1min, i.e. "5.123456s"
15248 rather than "5s 123ms 456us". For time spans under 1s
15249 millisecond values are shown, for those under 1ms
15250 microsecond values are shown. This should greatly improve
15251 all time-related output of systemd.
15253 * libsystemd-login and libsystemd-journal gained new
15254 functions for querying the poll() events mask and poll()
15255 timeout value for integration into arbitrary event
15258 * localectl gained the ability to list available X11 keymaps
15259 (models, layouts, variants, options).
15261 * 'systemd-analyze dot' gained the ability to filter for
15262 specific units via shell-style globs, to create smaller,
15263 more useful graphs. I.e. it is now possible to create simple
15264 graphs of all the dependencies between only target units, or
15265 of all units that Avahi has dependencies with.
15267 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
15268 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
15269 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
15270 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
15271 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
15272 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
15273 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
15277 * The boot-time readahead implementation for rotating media
15278 will now read the read-ahead data in multiple passes which
15279 consist of all read requests made in equidistant time
15280 intervals. This means instead of strictly reading read-ahead
15281 data in its physical order on disk we now try to find a
15282 middle ground between physical and access time order.
15284 * /etc/os-release files gained a new BUILD_ID= field for usage
15285 on operating systems that provide continuous builds of OS
15288 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
15289 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
15290 William Douglas, Zbigniew Jędrzejewski-Szmek
15294 * systemd-python gained an API exposing libsystemd-daemon.
15296 * The SMACK setup logic gained support for uploading CIPSO
15299 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
15300 ReadOnlyDirectories= and InaccessibleDirectories= has
15301 changed. The private /tmp and /var/tmp directories are now
15302 shared by all processes of a service (which means
15303 ExecStartPre= may now leave data in /tmp that ExecStart= of
15304 the same service can still access). When a service is
15305 stopped its temporary directories are immediately deleted
15306 (normal clean-up with tmpfiles is still done in addition to
15309 * By default, systemd will now set a couple of sysctl
15310 variables in the kernel: the safe sysrq options are turned
15311 on, IP route verification is turned on, and source routing
15312 disabled. The recently added hardlink and softlink
15313 protection of the kernel is turned on. These settings should
15314 be reasonably safe, and good defaults for all new systems.
15316 * The predictable network naming logic may now be turned off
15317 with a new kernel command line switch: net.ifnames=0.
15319 * A new libsystemd-bus module has been added that implements a
15320 pretty complete D-Bus client library. For details see:
15322 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
15324 * journald will now explicitly flush the journal files to disk
15325 at the latest 5min after each write. The file will then also
15326 be marked offline until the next write. This should increase
15327 reliability in case of a crash. The synchronization delay
15328 can be configured via SyncIntervalSec= in journald.conf.
15330 * There's a new remote-fs-setup.target unit that can be used
15331 to pull in specific services when at least one remote file
15332 system is to be mounted.
15334 * There are new targets timers.target and paths.target as
15335 canonical targets to pull user timer and path units in
15336 from. This complements sockets.target with a similar
15337 purpose for socket units.
15339 * libudev gained a new call udev_device_set_attribute_value()
15340 to set sysfs attributes of a device.
15342 * The udev daemon now sets the default number of worker
15343 processes executed in parallel based on the number of available
15344 CPUs instead of the amount of available RAM. This is supposed
15345 to provide a more reliable default and limit a too aggressive
15346 parallelism for setups with 1000s of devices connected.
15348 Contributions from: Auke Kok, Colin Walters, Cristian
15349 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
15350 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
15351 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
15352 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
15353 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
15354 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
15355 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
15356 Zbigniew Jędrzejewski-Szmek
15360 * Configuration of unit files may now be extended via drop-in
15361 files without having to edit/override the unit files
15362 themselves. More specifically, if the administrator wants to
15363 change one value for a service file foobar.service he can
15364 now do so by dropping in a configuration snippet into
15365 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
15366 will load all these snippets and apply them on top of the
15367 main unit configuration file, possibly extending or
15368 overriding its settings. Using these drop-in snippets is
15369 generally nicer than the two earlier options for changing
15370 unit files locally: copying the files from
15371 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
15372 them there; or creating a new file in /etc/systemd/system/
15373 that incorporates the original one via ".include". Drop-in
15374 snippets into these .d/ directories can be placed in any
15375 directory systemd looks for units in, and the usual
15376 overriding semantics between /usr/lib, /etc and /run apply
15379 * Most unit file settings which take lists of items can now be
15380 reset by assigning the empty string to them. For example,
15381 normally, settings such as Environment=FOO=BAR append a new
15382 environment variable assignment to the environment block,
15383 each time they are used. By assigning Environment= the empty
15384 string the environment block can be reset to empty. This is
15385 particularly useful with the .d/*.conf drop-in snippets
15386 mentioned above, since this adds the ability to reset list
15387 settings from vendor unit files via these drop-ins.
15389 * systemctl gained a new "list-dependencies" command for
15390 listing the dependencies of a unit recursively.
15392 * Inhibitors are now honored and listed by "systemctl
15393 suspend", "systemctl poweroff" (and similar) too, not only
15394 GNOME. These commands will also list active sessions by
15397 * Resource limits (as exposed by the various control group
15398 controllers) can now be controlled dynamically at runtime
15399 for all units. More specifically, you can now use a command
15400 like "systemctl set-cgroup-attr foobar.service cpu.shares
15401 2000" to alter the CPU shares a specific service gets. These
15402 settings are stored persistently on disk, and thus allow the
15403 administrator to easily adjust the resource usage of
15404 services with a few simple commands. This dynamic resource
15405 management logic is also available to other programs via the
15406 bus. Almost any kernel cgroup attribute and controller is
15409 * systemd-vconsole-setup will now copy all font settings to
15410 all allocated VTs, where it previously applied them only to
15413 * libsystemd-login gained the new sd_session_get_tty() API
15416 * This release drops support for a few legacy or
15417 distribution-specific LSB facility names when parsing init
15418 scripts: $x-display-manager, $mail-transfer-agent,
15419 $mail-transport-agent, $mail-transfer-agent, $smtp,
15420 $null. Also, the mail-transfer-agent.target unit backing
15421 this has been removed. Distributions which want to retain
15422 compatibility with this should carry the burden for
15423 supporting this themselves and patch support for these back
15424 in, if they really need to. Also, the facilities $syslog and
15425 $local_fs are now ignored, since systemd does not support
15426 early-boot LSB init scripts anymore, and these facilities
15427 are implied anyway for normal services. syslog.target has
15430 * There are new bus calls on PID1's Manager object for
15431 cancelling jobs, and removing snapshot units. Previously,
15432 both calls were only available on the Job and Snapshot
15433 objects themselves.
15435 * systemd-journal-gatewayd gained SSL support.
15437 * The various "environment" files, such as /etc/locale.conf
15438 now support continuation lines with a backslash ("\") as
15439 last character in the line, similarly in style (but different)
15440 to how this is supported in shells.
15442 * For normal user processes the _SYSTEMD_USER_UNIT= field is
15443 now implicitly appended to every log entry logged. systemctl
15444 has been updated to filter by this field when operating on a
15445 user systemd instance.
15447 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
15448 CAP_AUDIT_CONTROL capabilities to the capabilities set for
15449 the container. This makes it easier to boot unmodified
15450 Fedora systems in a container, which however still requires
15451 audit=0 to be passed on the kernel command line. Auditing in
15452 kernel and userspace is unfortunately still too broken in
15453 context of containers, hence we recommend compiling it out
15454 of the kernel or using audit=0. Hopefully this will be fixed
15455 one day for good in the kernel.
15457 * nspawn gained the new --bind= and --bind-ro= parameters to
15458 bind mount specific directories from the host into the
15461 * nspawn will now mount its own devpts file system instance
15462 into the container, in order not to leak pty devices from
15463 the host into the container.
15465 * systemd will now read the firmware boot time performance
15466 information from the EFI variables, if the used boot loader
15467 supports this, and takes it into account for boot performance
15468 analysis via "systemd-analyze". This is currently supported
15469 only in conjunction with Gummiboot, but could be supported
15470 by other boot loaders too. For details see:
15472 https://systemd.io/BOOT_LOADER_INTERFACE
15474 * A new generator has been added that automatically mounts the
15475 EFI System Partition (ESP) to /boot, if that directory
15476 exists, is empty, and no other file system has been
15477 configured to be mounted there.
15479 * logind will now send out PrepareForSleep(false) out
15480 unconditionally, after coming back from suspend. This may be
15481 used by applications as asynchronous notification for
15482 system resume events.
15484 * "systemctl unlock-sessions" has been added, that allows
15485 unlocking the screens of all user sessions at once, similar
15486 to how "systemctl lock-sessions" already locked all users
15487 sessions. This is backed by a new D-Bus call UnlockSessions().
15489 * "loginctl seat-status" will now show the master device of a
15490 seat. (i.e. the device of a seat that needs to be around for
15491 the seat to be considered available, usually the graphics
15494 * tmpfiles gained a new "X" line type, that allows
15495 configuration of files and directories (with wildcards) that
15496 shall be excluded from automatic cleanup ("aging").
15498 * udev default rules set the device node permissions now only
15499 at "add" events, and do not change them any longer with a
15500 later "change" event.
15502 * The log messages for lid events and power/sleep keypresses
15503 now carry a message ID.
15505 * We now have a substantially larger unit test suite, but this
15506 continues to be work in progress.
15508 * udevadm hwdb gained a new --root= parameter to change the
15509 root directory to operate relative to.
15511 * logind will now issue a background sync() request to the kernel
15512 early at shutdown, so that dirty buffers are flushed to disk early
15513 instead of at the last moment, in order to optimize shutdown
15516 * A new bootctl tool has been added that is an interface for
15517 certain boot loader operations. This is currently a preview
15518 and is likely to be extended into a small mechanism daemon
15519 like timedated, localed, hostnamed, and can be used by
15520 graphical UIs to enumerate available boot options, and
15521 request boot into firmware operations.
15523 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
15524 the rest of the package. It also has been updated to work
15525 correctly in initrds.
15527 * polkit previously has been runtime optional, and is now also
15528 compile time optional via a configure switch.
15530 * systemd-analyze has been reimplemented in C. Also "systemctl
15531 dot" has moved into systemd-analyze.
15533 * "systemctl status" with no further parameters will now print
15534 the status of all active or failed units.
15536 * Operations such as "systemctl start" can now be executed
15537 with a new mode "--irreversible" which may be used to queue
15538 operations that cannot accidentally be reversed by a later
15539 job queuing. This is by default used to make shutdown
15540 requests more robust.
15542 * The Python API of systemd now gained a new module for
15543 reading journal files.
15545 * A new tool kernel-install has been added that can install
15546 kernel images according to the Boot Loader Specification:
15548 https://systemd.io/BOOT_LOADER_SPECIFICATION
15550 * Boot time console output has been improved to provide
15551 animated boot time output for hanging jobs.
15553 * A new tool systemd-activate has been added which can be used
15554 to test socket activation with, directly from the command
15555 line. This should make it much easier to test and debug
15556 socket activation in daemons.
15558 * journalctl gained a new "--reverse" (or -r) option to show
15559 journal output in reverse order (i.e. newest line first).
15561 * journalctl gained a new "--pager-end" (or -e) option to jump
15562 to immediately jump to the end of the journal in the
15563 pager. This is only supported in conjunction with "less".
15565 * journalctl gained a new "--user-unit=" option, that works
15566 similarly to "--unit=" but filters for user units rather than
15569 * A number of unit files to ease adoption of systemd in
15570 initrds has been added. This moves some minimal logic from
15571 the various initrd implementations into systemd proper.
15573 * The journal files are now owned by a new group
15574 "systemd-journal", which exists specifically to allow access
15575 to the journal, and nothing else. Previously, we used the
15576 "adm" group for that, which however possibly covers more
15577 than just journal/log file access. This new group is now
15578 already used by systemd-journal-gatewayd to ensure this
15579 daemon gets access to the journal files and as little else
15580 as possible. Note that "make install" will also set FS ACLs
15581 up for /var/log/journal to give "adm" and "wheel" read
15582 access to it, in addition to "systemd-journal" which owns
15583 the journal files. We recommend that packaging scripts also
15584 add read access to "adm" + "wheel" to /var/log/journal, and
15585 all existing/future journal files. To normal users and
15586 administrators little changes, however packagers need to
15587 ensure to create the "systemd-journal" system group at
15588 package installation time.
15590 * The systemd-journal-gatewayd now runs as unprivileged user
15591 systemd-journal-gateway:systemd-journal-gateway. Packaging
15592 scripts need to create these system user/group at
15595 * timedated now exposes a new boolean property CanNTP that
15596 indicates whether a local NTP service is available or not.
15598 * systemd-detect-virt will now also detect xen PVs
15600 * The pstore file system is now mounted by default, if it is
15603 * In addition to the SELinux and IMA policies we will now also
15604 load SMACK policies at early boot.
15606 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
15607 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
15608 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
15609 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
15610 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
15611 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
15612 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
15613 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
15614 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
15615 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
15616 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
15617 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
15618 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
15619 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
15623 * Timer units now support calendar time events in addition to
15624 monotonic time events. That means you can now trigger a unit
15625 based on a calendar time specification such as "Thu,Fri
15626 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
15627 or fifth day of any month of the year 2013, given that it is
15628 a Thursday or a Friday. This brings timer event support
15629 considerably closer to cron's capabilities. For details on
15630 the supported calendar time specification language see
15633 * udev now supports a number of different naming policies for
15634 network interfaces for predictable names, and a combination
15635 of these policies is now the default. Please see this wiki
15636 document for details:
15638 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
15640 * Auke Kok's bootchart implementation has been added to the
15641 systemd tree. It is an optional component that can graph the
15642 boot in quite some detail. It is one of the best bootchart
15643 implementations around and minimal in its code and
15646 * nss-myhostname has been integrated into the systemd source
15647 tree. nss-myhostname guarantees that the local hostname
15648 always stays resolvable via NSS. It has been a weak
15649 requirement of systemd-hostnamed since a long time, and
15650 since its code is actually trivial we decided to just
15651 include it in systemd's source tree. It can be turned off
15652 with a configure switch.
15654 * The read-ahead logic is now capable of properly detecting
15655 whether a btrfs file system is on SSD or rotating media, in
15656 order to optimize the read-ahead scheme. Previously, it was
15657 only capable of detecting this on traditional file systems
15660 * In udev, additional device properties are now read from the
15661 IAB in addition to the OUI database. Also, Bluetooth company
15662 identities are attached to the devices as well.
15664 * In service files %U may be used as specifier that is
15665 replaced by the configured user name of the service.
15667 * nspawn may now be invoked without a controlling TTY. This
15668 makes it suitable for invocation as its own service. This
15669 may be used to set up a simple containerized server system
15670 using only core OS tools.
15672 * systemd and nspawn can now accept socket file descriptors
15673 when they are started for socket activation. This enables
15674 implementation of socket activated nspawn
15675 containers. i.e. think about autospawning an entire OS image
15676 when the first SSH or HTTP connection is received. We expect
15677 that similar functionality will also be added to libvirt-lxc
15680 * journalctl will now suppress ANSI color codes when
15681 presenting log data.
15683 * systemctl will no longer show control group information for
15684 a unit if the control group is empty anyway.
15686 * logind can now automatically suspend/hibernate/shutdown the
15689 * /etc/machine-info and hostnamed now also expose the chassis
15690 type of the system. This can be used to determine whether
15691 the local system is a laptop, desktop, handset or
15692 tablet. This information may either be configured by the
15693 user/vendor or is automatically determined from ACPI and DMI
15694 information if possible.
15696 * A number of polkit actions are now bound together with "imply"
15697 rules. This should simplify creating UIs because many actions
15698 will now authenticate similar ones as well.
15700 * Unit files learnt a new condition ConditionACPower= which
15701 may be used to conditionalize a unit depending on whether an
15702 AC power source is connected or not, of whether the system
15703 is running on battery power.
15705 * systemctl gained a new "is-failed" verb that may be used in
15706 shell scripts and suchlike to check whether a specific unit
15707 is in the "failed" state.
15709 * The EnvironmentFile= setting in unit files now supports file
15710 globbing, and can hence be used to easily read a number of
15711 environment files at once.
15713 * systemd will no longer detect and recognize specific
15714 distributions. All distribution-specific #ifdeffery has been
15715 removed, systemd is now fully generic and
15716 distribution-agnostic. Effectively, not too much is lost as
15717 a lot of the code is still accessible via explicit configure
15718 switches. However, support for some distribution specific
15719 legacy configuration file formats has been dropped. We
15720 recommend distributions to simply adopt the configuration
15721 files everybody else uses now and convert the old
15722 configuration from packaging scripts. Most distributions
15723 already did that. If that's not possible or desirable,
15724 distributions are welcome to forward port the specific
15725 pieces of code locally from the git history.
15727 * When logging a message about a unit systemd will now always
15728 log the unit name in the message meta data.
15730 * localectl will now also discover system locale data that is
15731 not stored in locale archives, but directly unpacked.
15733 * logind will no longer unconditionally use framebuffer
15734 devices as seat masters, i.e. as devices that are required
15735 to be existing before a seat is considered preset. Instead,
15736 it will now look for all devices that are tagged as
15737 "seat-master" in udev. By default, framebuffer devices will
15738 be marked as such, but depending on local systems, other
15739 devices might be marked as well. This may be used to
15740 integrate graphics cards using closed source drivers (such
15741 as NVidia ones) more nicely into logind. Note however, that
15742 we recommend using the open source NVidia drivers instead,
15743 and no udev rules for the closed-source drivers will be
15744 shipped from us upstream.
15746 Contributions from: Adam Williamson, Alessandro Crismani, Auke
15747 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
15748 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
15749 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
15750 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
15751 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
15752 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
15753 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
15754 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
15755 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
15756 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
15757 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
15762 * udev gained support for loading additional device properties
15763 from an indexed database that is keyed by vendor/product IDs
15764 and similar device identifiers. For the beginning this
15765 "hwdb" is populated with data from the well-known PCI and
15766 USB database, but also includes PNP, ACPI and OID data. In
15767 the longer run this indexed database shall grow into
15768 becoming the one central database for non-essential
15769 userspace device metadata. Previously, data from the PCI/USB
15770 database was only attached to select devices, since the
15771 lookup was a relatively expensive operation due to O(n) time
15772 complexity (with n being the number of entries in the
15773 database). Since this is now O(1), we decided to add in this
15774 data for all devices where this is available, by
15775 default. Note that the indexed database needs to be rebuilt
15776 when new data files are installed. To achieve this you need
15777 to update your packaging scripts to invoke "udevadm hwdb
15778 --update" after installation of hwdb data files. For
15779 RPM-based distributions we introduced the new
15780 %udev_hwdb_update macro for this purpose.
15782 * The Journal gained support for the "Message Catalog", an
15783 indexed database to link up additional information with
15784 journal entries. For further details please check:
15786 https://www.freedesktop.org/wiki/Software/systemd/catalog
15788 The indexed message catalog database also needs to be
15789 rebuilt after installation of message catalog files. Use
15790 "journalctl --update-catalog" for this. For RPM-based
15791 distributions we introduced the %journal_catalog_update
15792 macro for this purpose.
15794 * The Python Journal bindings gained support for the standard
15795 Python logging framework.
15797 * The Journal API gained new functions for checking whether
15798 the underlying file system of a journal file is capable of
15799 properly reporting file change notifications, or whether
15800 applications that want to reflect journal changes "live"
15801 need to recheck journal files continuously in appropriate
15804 * It is now possible to set the "age" field for tmpfiles
15805 entries to 0, indicating that files matching this entry
15806 shall always be removed when the directories are cleaned up.
15808 * coredumpctl gained a new "gdb" verb which invokes gdb
15809 right-away on the selected coredump.
15811 * There's now support for "hybrid sleep" on kernels that
15812 support this, in addition to "suspend" and "hibernate". Use
15813 "systemctl hybrid-sleep" to make use of this.
15815 * logind's HandleSuspendKey= setting (and related settings)
15816 now gained support for a new "lock" setting to simply
15817 request the screen lock on all local sessions, instead of
15818 actually executing a suspend or hibernation.
15820 * systemd will now mount the EFI variables file system by
15823 * Socket units now gained support for configuration of the
15824 SMACK security label.
15826 * timedatectl will now output the time of the last and next
15827 daylight saving change.
15829 * We dropped support for various legacy and distro-specific
15830 concepts, such as insserv, early-boot SysV services
15831 (i.e. those for non-standard runlevels such as 'b' or 'S')
15832 or ArchLinux /etc/rc.conf support. We recommend the
15833 distributions who still need support this to either continue
15834 to maintain the necessary patches downstream, or find a
15835 different solution. (Talk to us if you have questions!)
15837 * Various systemd components will now bypass polkit checks for
15838 root and otherwise handle properly if polkit is not found to
15839 be around. This should fix most issues for polkit-less
15840 systems. Quite frankly this should have been this way since
15841 day one. It is absolutely our intention to make systemd work
15842 fine on polkit-less systems, and we consider it a bug if
15843 something does not work as it should if polkit is not around.
15845 * For embedded systems it is now possible to build udev and
15846 systemd without blkid and/or kmod support.
15848 * "systemctl switch-root" is now capable of switching root
15849 more than once. I.e. in addition to transitions from the
15850 initrd to the host OS it is now possible to transition to
15851 further OS images from the host. This is useful to implement
15852 offline updating tools.
15854 * Various other additions have been made to the RPM macros
15855 shipped with systemd. Use %udev_rules_update() after
15856 installing new udev rules files. %_udevhwdbdir,
15857 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
15858 %_sysctldir are now available which resolve to the right
15859 directories for packages to place various data files in.
15861 * journalctl gained the new --full switch (in addition to
15862 --all, to disable ellipsation for long messages.
15864 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
15865 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
15866 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
15867 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
15868 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
15869 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
15870 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
15871 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
15872 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
15876 * journalctl gained new --since= and --until= switches to
15877 filter by time. It also now supports nice filtering for
15878 units via --unit=/-u.
15880 * Type=oneshot services may use ExecReload= and do the
15883 * The journal daemon now supports time-based rotation and
15884 vacuuming, in addition to the usual disk-space based
15887 * The journal will now index the available field values for
15888 each field name. This enables clients to show pretty drop
15889 downs of available match values when filtering. The bash
15890 completion of journalctl has been updated
15891 accordingly. journalctl gained a new switch -F to list all
15892 values a certain field takes in the journal database.
15894 * More service events are now written as structured messages
15895 to the journal, and made recognizable via message IDs.
15897 * The timedated, localed and hostnamed mini-services which
15898 previously only provided support for changing time, locale
15899 and hostname settings from graphical DEs such as GNOME now
15900 also have a minimal (but very useful) text-based client
15901 utility each. This is probably the nicest way to changing
15902 these settings from the command line now, especially since
15903 it lists available options and is fully integrated with bash
15906 * There's now a new tool "systemd-coredumpctl" to list and
15907 extract coredumps from the journal.
15909 * We now install a README each in /var/log/ and
15910 /etc/rc.d/init.d explaining where the system logs and init
15911 scripts went. This hopefully should help folks who go to
15912 that dirs and look into the otherwise now empty void and
15913 scratch their heads.
15915 * When user-services are invoked (by systemd --user) the
15916 $MANAGERPID env var is set to the PID of systemd.
15918 * SIGRTMIN+24 when sent to a --user instance will now result
15919 in immediate termination of systemd.
15921 * gatewayd received numerous feature additions such as a
15922 "follow" mode, for live syncing and filtering.
15924 * browse.html now allows filtering and showing detailed
15925 information on specific entries. Keyboard navigation and
15926 mouse screen support has been added.
15928 * gatewayd/journalctl now supports HTML5/JSON
15929 Server-Sent-Events as output.
15931 * The SysV init script compatibility logic will now
15932 heuristically determine whether a script supports the
15933 "reload" verb, and only then make this available as
15934 "systemctl reload".
15936 * "systemctl status --follow" has been removed, use "journalctl
15939 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
15940 have been removed since they are hardly useful to be
15943 * And I'd like to take the opportunity to specifically mention
15944 Zbigniew for his great contributions. Zbigniew, you rock!
15946 Contributions from: Andrew Eikum, Christian Hesse, Colin
15947 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
15948 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
15949 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
15950 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
15951 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
15952 Jędrzejewski-Szmek, Сковорода Никита Андреевич
15956 * If /etc/vconsole.conf is non-existent or empty we will no
15957 longer load any console font or key map at boot by
15958 default. Instead the kernel defaults will be left
15959 intact. This is definitely the right thing to do, as no
15960 configuration should mean no configuration, and hard-coding
15961 font names that are different on all archs is probably a bad
15962 idea. Also, the kernel default key map and font should be
15963 good enough for most cases anyway, and mostly identical to
15964 the userspace fonts/key maps we previously overloaded them
15965 with. If distributions want to continue to default to a
15966 non-kernel font or key map they should ship a default
15967 /etc/vconsole.conf with the appropriate contents.
15969 Contributions from: Colin Walters, Daniel J Walsh, Dave
15970 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
15971 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15975 * journalctl gained a new --cursor= switch to show entries
15976 starting from the specified location in the journal.
15978 * We now enforce a size limit on journal entry fields exported
15979 with "-o json" in journalctl. Fields larger than 4K will be
15980 assigned null. This can be turned off with --all.
15982 * An (optional) journal gateway daemon is now available as
15983 "systemd-journal-gatewayd.service". This service provides
15984 access to the journal via HTTP and JSON. This functionality
15985 will be used to implement live log synchronization in both
15986 pull and push modes, but has various other users too, such
15987 as easy log access for debugging of embedded devices. Right
15988 now it is already useful to retrieve the journal via HTTP:
15990 # systemctl start systemd-journal-gatewayd.service
15991 # wget http://localhost:19531/entries
15993 This will download the journal contents in a
15994 /var/log/messages compatible format. The same as JSON:
15996 # curl -H"Accept: application/json" http://localhost:19531/entries
15998 This service is also accessible via a web browser where a
15999 single static HTML5 app is served that uses the JSON logic
16000 to enable the user to do some basic browsing of the
16001 journal. This will be extended later on. Here's an example
16002 screenshot of this app in its current state:
16004 https://0pointer.de/public/journal-gatewayd
16006 Contributions from: Kay Sievers, Lennart Poettering, Robert
16007 Milasan, Tom Gundersen
16011 * The bash completion logic is now available for journalctl
16014 * We do not mount the "cpuset" controller anymore together with
16015 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
16016 started if no parameters are assigned to it. "cpuset" hence
16017 broke code that assumed it could create "cpu" groups and
16020 * journalctl -f will now subscribe to terminal size changes,
16021 and line break accordingly.
16023 Contributions from: Dave Reisner, Kay Sievers, Lennart
16024 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
16028 * nspawn will now create a symlink /etc/localtime in the
16029 container environment, copying the host's timezone
16030 setting. Previously this has been done via a bind mount, but
16031 since symlinks cannot be bind mounted this has now been
16032 changed to create/update the appropriate symlink.
16034 * journalctl -n's line number argument is now optional, and
16035 will default to 10 if omitted.
16037 * journald will now log the maximum size the journal files may
16038 take up on disk. This is particularly useful if the default
16039 built-in logic of determining this parameter from the file
16040 system size is used. Use "systemctl status
16041 systemd-journald.service" to see this information.
16043 * The multi-seat X wrapper tool has been stripped down. As X
16044 is now capable of enumerating graphics devices via udev in a
16045 seat-aware way the wrapper is not strictly necessary
16046 anymore. A stripped down temporary stop-gap is still shipped
16047 until the upstream display managers have been updated to
16048 fully support the new X logic. Expect this wrapper to be
16049 removed entirely in one of the next releases.
16051 * HandleSleepKey= in logind.conf has been split up into
16052 HandleSuspendKey= and HandleHibernateKey=. The old setting
16053 is not available anymore. X11 and the kernel are
16054 distinguishing between these keys and we should too. This
16055 also means the inhibition lock for these keys has been split
16058 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
16059 Poettering, Lukas Nykryn, Václav Pavlín
16063 * Whenever a unit changes state we will now log this to the
16064 journal and show along the unit's own log output in
16065 "systemctl status".
16067 * ConditionPathIsMountPoint= can now properly detect bind
16068 mount points too. (Previously, a bind mount of one file
16069 system to another place in the same file system could not be
16070 detected as mount, since they shared struct stat's st_dev
16073 * We will now mount the cgroup controllers cpu, cpuacct,
16074 cpuset and the controllers net_cls, net_prio together by
16077 * nspawn containers will now have a virtualized boot
16078 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
16079 over with a randomized ID at container initialization). This
16080 has the effect of making "journalctl -b" do the right thing
16083 * The JSON output journal serialization has been updated not
16084 to generate "endless" list objects anymore, but rather one
16085 JSON object per line. This is more in line how most JSON
16086 parsers expect JSON objects. The new output mode
16087 "json-pretty" has been added to provide similar output, but
16088 neatly aligned for readability by humans.
16090 * We dropped all explicit sync() invocations in the shutdown
16091 code. The kernel does this implicitly anyway in the kernel
16092 reboot() syscall. halt(8)'s -n option is now a compatibility
16095 * We now support virtualized reboot() in containers, as
16096 supported by newer kernels. We will fall back to exit() if
16097 CAP_SYS_REBOOT is not available to the container. Also,
16098 nspawn makes use of this now and will actually reboot the
16099 container if the containerized OS asks for that.
16101 * journalctl will only show local log output by default
16102 now. Use --merge (-m) to show remote log output, too.
16104 * libsystemd-journal gained the new sd_journal_get_usage()
16105 call to determine the current disk usage of all journal
16106 files. This is exposed in the new "journalctl --disk-usage"
16109 * journald gained a new configuration setting SplitMode= in
16110 journald.conf which may be used to control how user journals
16111 are split off. See journald.conf(5) for details.
16113 * A new condition type ConditionFileNotEmpty= has been added.
16115 * tmpfiles' "w" lines now support file globbing, to write
16116 multiple files at once.
16118 * We added Python bindings for the journal submission
16119 APIs. More Python APIs for a number of selected APIs will
16120 likely follow. Note that we intend to add native bindings
16121 only for the Python language, as we consider it common
16122 enough to deserve bindings shipped within systemd. There are
16123 various projects outside of systemd that provide bindings
16124 for languages such as PHP or Lua.
16126 * Many conditions will now resolve specifiers such as %i. In
16127 addition, PathChanged= and related directives of .path units
16128 now support specifiers as well.
16130 * There's now a new RPM macro definition for the system preset
16133 * journald will now warn if it ca not forward a message to the
16134 syslog daemon because its socket is full.
16136 * timedated will no longer write or process /etc/timezone,
16137 except on Debian. As we do not support late mounted /usr
16138 anymore /etc/localtime always being a symlink is now safe,
16139 and hence the information in /etc/timezone is not necessary
16142 * logind will now always reserve one VT for a text getty (VT6
16143 by default). Previously if more than 6 X sessions where
16144 started they took up all the VTs with auto-spawned gettys,
16145 so that no text gettys were available anymore.
16147 * udev will now automatically inform the btrfs kernel logic
16148 about btrfs RAID components showing up. This should make
16149 simple hotplug based btrfs RAID assembly work.
16151 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
16152 (but not for its children which will stay at the kernel
16153 default). This should allow setups with a lot more listening
16156 * systemd will now always pass the configured timezone to the
16157 kernel at boot. timedated will do the same when the timezone
16160 * logind's inhibition logic has been updated. By default,
16161 logind will now handle the lid switch, the power and sleep
16162 keys all the time, even in graphical sessions. If DEs want
16163 to handle these events on their own they should take the new
16164 handle-power-key, handle-sleep-key and handle-lid-switch
16165 inhibitors during their runtime. A simple way to achieve
16166 that is to invoke the DE wrapped in an invocation of:
16168 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
16170 * Access to unit operations is now checked via SELinux taking
16171 the unit file label and client process label into account.
16173 * systemd will now notify the administrator in the journal
16174 when he over-mounts a non-empty directory.
16176 * There are new specifiers that are resolved in unit files,
16177 for the hostname (%H), the machine ID (%m) and the boot ID
16180 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
16181 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
16182 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
16183 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
16184 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
16185 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
16186 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
16190 * Support for reading structured kernel messages from
16191 /dev/kmsg has now been added and is enabled by default.
16193 * Support for reading kernel messages from /proc/kmsg has now
16194 been removed. If you want kernel messages in the journal
16195 make sure to run a recent kernel (>= 3.5) that supports
16196 reading structured messages from /dev/kmsg (see
16197 above). /proc/kmsg is now exclusive property of classic
16198 syslog daemons again.
16200 * The libudev API gained the new
16201 udev_device_new_from_device_id() call.
16203 * The logic for file system namespace (ReadOnlyDirectory=,
16204 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
16205 require pivot_root() anymore. This means fewer temporary
16206 directories are created below /tmp for this feature.
16208 * nspawn containers will now see and receive all submounts
16209 made on the host OS below the root file system of the
16212 * Forward Secure Sealing is now supported for Journal files,
16213 which provide cryptographical sealing of journal files so
16214 that attackers cannot alter log history anymore without this
16215 being detectable. Lennart will soon post a blog story about
16216 this explaining it in more detail.
16218 * There are two new service settings RestartPreventExitStatus=
16219 and SuccessExitStatus= which allow configuration of exit
16220 status (exit code or signal) which will be excepted from the
16221 restart logic, resp. consider successful.
16223 * journalctl gained the new --verify switch that can be used
16224 to check the integrity of the structure of journal files and
16225 (if Forward Secure Sealing is enabled) the contents of
16228 * nspawn containers will now be run with /dev/stdin, /dev/fd/
16229 and similar symlinks pre-created. This makes running shells
16230 as container init process a lot more fun.
16232 * The fstab support can now handle PARTUUID= and PARTLABEL=
16235 * A new ConditionHost= condition has been added to match
16236 against the hostname (with globs) and machine ID. This is
16237 useful for clusters where a single OS image is used to
16238 provision a large number of hosts which shall run slightly
16239 different sets of services.
16241 * Services which hit the restart limit will now be placed in a
16244 Contributions from: Bertram Poettering, Dave Reisner, Huang
16245 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
16246 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
16250 * When running in --user mode systemd will now become a
16251 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
16252 tree a lot more organized.
16254 * A new PartOf= unit dependency type has been introduced that
16255 may be used to group services in a natural way.
16257 * "systemctl enable" may now be used to enable instances of
16260 * journalctl now prints error log levels in red, and
16261 warning/notice log levels in bright white. It also supports
16262 filtering by log level now.
16264 * cgtop gained a new -n switch (similar to top), to configure
16265 the maximum number of iterations to run for. It also gained
16266 -b, to run in batch mode (accepting no input).
16268 * The suffix ".service" may now be omitted on most systemctl
16269 command lines involving service unit names.
16271 * There's a new bus call in logind to lock all sessions, as
16272 well as a loginctl verb for it "lock-sessions".
16274 * libsystemd-logind.so gained a new call sd_journal_perror()
16275 that works similar to libc perror() but logs to the journal
16276 and encodes structured information about the error number.
16278 * /etc/crypttab entries now understand the new keyfile-size=
16281 * shutdown(8) now can send a (configurable) wall message when
16282 a shutdown is cancelled.
16284 * The mount propagation mode for the root file system will now
16285 default to "shared", which is useful to make containers work
16286 nicely out-of-the-box so that they receive new mounts from
16287 the host. This can be undone locally by running "mount
16288 --make-rprivate /" if needed.
16290 * The prefdm.service file has been removed. Distributions
16291 should maintain this unit downstream if they intend to keep
16292 it around. However, we recommend writing normal unit files
16293 for display managers instead.
16295 * Since systemd is a crucial part of the OS we will now
16296 default to a number of compiler switches that improve
16297 security (hardening) such as read-only relocations, stack
16298 protection, and suchlike.
16300 * The TimeoutSec= setting for services is now split into
16301 TimeoutStartSec= and TimeoutStopSec= to allow configuration
16302 of individual time outs for the start and the stop phase of
16305 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
16306 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
16307 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
16308 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
16309 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
16310 Gundersen, Zbigniew Jędrzejewski-Szmek
16314 * The journal and id128 C APIs are now fully documented as man
16317 * Extra safety checks have been added when transitioning from
16318 the initial RAM disk to the main system to avoid accidental
16321 * /etc/crypttab entries now understand the new keyfile-offset=
16324 * systemctl -t can now be used to filter by unit load state.
16326 * The journal C API gained the new sd_journal_wait() call to
16327 make writing synchronous journal clients easier.
16329 * journalctl gained the new -D switch to show journals from a
16330 specific directory.
16332 * journalctl now displays a special marker between log
16333 messages of two different boots.
16335 * The journal is now explicitly flushed to /var via a service
16336 systemd-journal-flush.service, rather than implicitly simply
16337 by seeing /var/log/journal to be writable.
16339 * journalctl (and the journal C APIs) can now match for much
16340 more complex expressions, with alternatives and
16343 * When transitioning from the initial RAM disk to the main
16344 system we will now kill all processes in a killing spree to
16345 ensure no processes stay around by accident.
16347 * Three new specifiers may be used in unit files: %u, %h, %s
16348 resolve to the user name, user home directory resp. user
16349 shell. This is useful for running systemd user instances.
16351 * We now automatically rotate journal files if their data
16352 object hash table gets a fill level > 75%. We also size the
16353 hash table based on the configured maximum file size. This
16354 together should lower hash collisions drastically and thus
16355 speed things up a bit.
16357 * journalctl gained the new "--header" switch to introspect
16358 header data of journal files.
16360 * A new setting SystemCallFilters= has been added to services which may
16361 be used to apply deny lists or allow lists to system calls. This is
16362 based on SECCOMP Mode 2 of Linux 3.5.
16364 * nspawn gained a new --link-journal= switch (and quicker: -j)
16365 to link the container journal with the host. This makes it
16366 very easy to centralize log viewing on the host for all
16367 guests while still keeping the journal files separated.
16369 * Many bugfixes and optimizations
16371 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
16372 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
16373 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
16378 * Several tools now understand kernel command line arguments,
16379 which are only read when run in an initial RAM disk. They
16380 usually follow closely their normal counterparts, but are
16383 * There's a new tool to analyze the readahead files that are
16384 automatically generated at boot. Use:
16386 /usr/lib/systemd/systemd-readahead analyze /.readahead
16388 * We now provide an early debug shell on tty9 if this enabled. Use:
16390 systemctl enable debug-shell.service
16392 * All plymouth related units have been moved into the Plymouth
16393 package. Please make sure to upgrade your Plymouth version
16396 * systemd-tmpfiles now supports getting passed the basename of
16397 a configuration file only, in which case it will look for it
16398 in all appropriate directories automatically.
16400 * udevadm info now takes a /dev or /sys path as argument, and
16401 does the right thing. Example:
16403 udevadm info /dev/sda
16404 udevadm info /sys/class/block/sda
16406 * systemctl now prints a warning if a unit is stopped but a
16407 unit that might trigger it continues to run. Example: a
16408 service is stopped but the socket that activates it is left
16411 * "systemctl status" will now mention if the log output was
16412 shortened due to rotation since a service has been started.
16414 * The journal API now exposes functions to determine the
16415 "cutoff" times due to rotation.
16417 * journald now understands SIGUSR1 and SIGUSR2 for triggering
16418 immediately flushing of runtime logs to /var if possible,
16419 resp. for triggering immediate rotation of the journal
16422 * It is now considered an error if a service is attempted to
16423 be stopped that is not loaded.
16425 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
16427 * systemd-analyze now supports Python 3
16429 * tmpfiles now supports cleaning up directories via aging
16430 where the first level dirs are always kept around but
16431 directories beneath it automatically aged. This is enabled
16432 by prefixing the age field with '~'.
16434 * Seat objects now expose CanGraphical, CanTTY properties
16435 which is required to deal with very fast bootups where the
16436 display manager might be running before the graphics drivers
16437 completed initialization.
16439 * Seat objects now expose a State property.
16441 * We now include RPM macros for service enabling/disabling
16442 based on the preset logic. We recommend RPM based
16443 distributions to make use of these macros if possible. This
16444 makes it simpler to reuse RPM spec files across
16447 * We now make sure that the collected systemd unit name is
16448 always valid when services log to the journal via
16451 * There's a new man page kernel-command-line(7) detailing all
16452 command line options we understand.
16454 * The fstab generator may now be disabled at boot by passing
16455 fstab=0 on the kernel command line.
16457 * A new kernel command line option modules-load= is now understood
16458 to load a specific kernel module statically, early at boot.
16460 * Unit names specified on the systemctl command line are now
16461 automatically escaped as needed. Also, if file system or
16462 device paths are specified they are automatically turned
16463 into the appropriate mount or device unit names. Example:
16465 systemctl status /home
16466 systemctl status /dev/sda
16468 * The SysVConsole= configuration option has been removed from
16469 system.conf parsing.
16471 * The SysV search path is no longer exported on the D-Bus
16474 * The Names= option has been removed from unit file parsing.
16476 * There's a new man page bootup(7) detailing the boot process.
16478 * Every unit and every generator we ship with systemd now
16479 comes with full documentation. The self-explanatory boot is
16482 * A couple of services gained "systemd-" prefixes in their
16483 name if they wrap systemd code, rather than only external
16484 code. Among them fsck@.service which is now
16485 systemd-fsck@.service.
16487 * The HaveWatchdog property has been removed from the D-Bus
16490 * systemd.confirm_spawn= on the kernel command line should now
16493 * There's a new man page crypttab(5) which details all options
16494 we actually understand.
16496 * systemd-nspawn gained a new --capability= switch to pass
16497 additional capabilities to the container.
16499 * timedated will now read known NTP implementation unit names
16500 from /usr/lib/systemd/ntp-units.d/*.list,
16501 systemd-timedated-ntp.target has been removed.
16503 * journalctl gained a new switch "-b" that lists log data of
16504 the current boot only.
16506 * The notify socket is in the abstract namespace again, in
16507 order to support daemons which chroot() at start-up.
16509 * There is a new Storage= configuration option for journald
16510 which allows configuration of where log data should go. This
16511 also provides a way to disable journal logging entirely, so
16512 that data collected is only forwarded to the console, the
16513 kernel log buffer or another syslog implementation.
16515 * Many bugfixes and optimizations
16517 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
16518 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
16519 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
16520 Shawn Landden, Tom Gundersen
16524 * "systemctl help <unit>" now shows the man page if one is
16527 * Several new man pages have been added.
16529 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
16530 MaxLevelConsole= can now be specified in
16531 journald.conf. These options allow reducing the amount of
16532 data stored on disk or forwarded by the log level.
16534 * TimerSlackNSec= can now be specified in system.conf for
16535 PID1. This allows system-wide power savings.
16537 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
16538 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
16543 * logind is now capable of (optionally) handling power and
16544 sleep keys as well as the lid switch.
16546 * journalctl now understands the syntax "journalctl
16547 /usr/bin/avahi-daemon" to get all log output of a specific
16550 * CapabilityBoundingSet= in system.conf now also influences
16551 the capability bound set of usermode helpers of the kernel.
16553 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
16554 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
16555 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
16556 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
16560 * Note that we skipped 139 releases here in order to set the
16561 new version to something that is greater than both udev's
16562 and systemd's most recent version number.
16564 * udev: all udev sources are merged into the systemd source tree now.
16565 All future udev development will happen in the systemd tree. It
16566 is still fully supported to use the udev daemon and tools without
16567 systemd running, like in initramfs or other init systems. Building
16568 udev though, will require the *build* of the systemd tree, but
16569 udev can be properly *run* without systemd.
16571 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
16572 should be used to create dead device nodes as workarounds for broken
16575 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
16576 no longer supported. udev_monitor_new_from_netlink() needs to be
16577 used to subscribe to events.
16579 * udev: when udevd is started by systemd, processes which are left
16580 behind by forking them off of udev rules, are unconditionally cleaned
16581 up and killed now after the event handling has finished. Services or
16582 daemons must be started as systemd services. Services can be
16583 pulled-in by udev to get started, but they can no longer be directly
16584 forked by udev rules.
16586 * udev: the daemon binary is called systemd-udevd now and installed
16587 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
16588 to adapt to that, create symlink, or rename the binary after building
16591 * libudev no longer provides these symbols:
16592 udev_monitor_from_socket()
16593 udev_queue_get_failed_list_entry()
16594 udev_get_{dev,sys,run}_path()
16595 The versions number was bumped and symbol versioning introduced.
16597 * systemd-loginctl and systemd-journalctl have been renamed
16598 to loginctl and journalctl to match systemctl.
16600 * The config files: /etc/systemd/systemd-logind.conf and
16601 /etc/systemd/systemd-journald.conf have been renamed to
16602 logind.conf and journald.conf. Package updates should rename
16603 the files to the new names on upgrade.
16605 * For almost all files the license is now LGPL2.1+, changed
16606 from the previous GPL2.0+. Exceptions are some minor stuff
16607 of udev (which will be changed to LGPL2.1 eventually, too),
16608 and the MIT licensed sd-daemon.[ch] library that is suitable
16609 to be used as drop-in files.
16611 * systemd and logind now handle system sleep states, in
16612 particular suspending and hibernating.
16614 * logind now implements a sleep/shutdown/idle inhibiting logic
16615 suitable for a variety of uses. Soonishly Lennart will blog
16616 about this in more detail.
16618 * var-run.mount and var-lock.mount are no longer provided
16619 (which previously bind mounted these directories to their new
16620 places). Distributions which have not converted these
16621 directories to symlinks should consider stealing these files
16622 from git history and add them downstream.
16624 * We introduced the Documentation= field for units and added
16625 this to all our shipped units. This is useful to make it
16626 easier to explore the boot and the purpose of the various
16629 * All smaller setup units (such as
16630 systemd-vconsole-setup.service) now detect properly if they
16631 are run in a container and are skipped when
16632 appropriate. This guarantees an entirely noise-free boot in
16633 Linux container environments such as systemd-nspawn.
16635 * A framework for implementing offline system updates is now
16636 integrated, for details see:
16637 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
16639 * A new service type Type=idle is available now which helps us
16640 avoiding ugly interleaving of getty output and boot status
16643 * There's now a system-wide CapabilityBoundingSet= option to
16644 globally reduce the set of capabilities for the
16645 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
16646 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
16647 even CAP_NET_ADMIN system-wide for secure systems.
16649 * There are now system-wide DefaultLimitXXX= options to
16650 globally change the defaults of the various resource limits
16651 for all units started by PID 1.
16653 * Harald Hoyer's systemd test suite has been integrated into
16654 systemd which allows easy testing of systemd builds in qemu
16655 and nspawn. (This is really awesome! Ask us for details!)
16657 * The fstab parser is now implemented as generator, not inside
16660 * systemctl will now warn you if .mount units generated from
16661 /etc/fstab are out of date due to changes in fstab that
16662 have not been read by systemd yet.
16664 * systemd is now suitable for usage in initrds. Dracut has
16665 already been updated to make use of this. With this in place
16666 initrds get a slight bit faster but primarily are much
16667 easier to introspect and debug since "systemctl status" in
16668 the host system can be used to introspect initrd services,
16669 and the journal from the initrd is kept around too.
16671 * systemd-delta has been added, a tool to explore differences
16672 between user/admin configuration and vendor defaults.
16674 * PrivateTmp= now affects both /tmp and /var/tmp.
16676 * Boot time status messages are now much prettier and feature
16677 proper english language. Booting up systemd has never been
16680 * Read-ahead pack files now include the inode number of all
16681 files to pre-cache. When the inode changes the pre-caching
16682 is not attempted. This should be nicer to deal with updated
16683 packages which might result in changes of read-ahead
16686 * We now temporaritly lower the kernel's read_ahead_kb variable
16687 when collecting read-ahead data to ensure the kernel's
16688 built-in read-ahead does not add noise to our measurements
16689 of necessary blocks to pre-cache.
16691 * There's now RequiresMountsFor= to add automatic dependencies
16692 for all mounts necessary for a specific file system path.
16694 * MountAuto= and SwapAuto= have been removed from
16695 system.conf. Mounting file systems at boot has to take place
16698 * nspawn now learned a new switch --uuid= to set the machine
16699 ID on the command line.
16701 * nspawn now learned the -b switch to automatically search
16702 for an init system.
16704 * vt102 is now the default TERM for serial TTYs, upgraded from
16707 * systemd-logind now works on VT-less systems.
16709 * The build tree has been reorganized. The individual
16710 components now have directories of their own.
16712 * A new condition type ConditionPathIsReadWrite= is now available.
16714 * nspawn learned the new -C switch to create cgroups for the
16715 container in other hierarchies.
16717 * We now have support for hardware watchdogs, configurable in
16720 * The scheduled shutdown logic now has a public API.
16722 * We now mount /tmp as tmpfs by default, but this can be
16723 masked and /etc/fstab can override it.
16725 * Since udisks does not make use of /media anymore we are not
16726 mounting a tmpfs on it anymore.
16728 * journalctl gained a new --local switch to only interleave
16729 locally generated journal files.
16731 * We can now load the IMA policy at boot automatically.
16733 * The GTK tools have been split off into a systemd-ui.
16735 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
16736 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
16737 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
16738 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
16739 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
16740 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
16741 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
16742 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
16743 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
16748 * This is mostly a bugfix release
16750 * Support optional initialization of the machine ID from the
16751 KVM or container configured UUID.
16753 * Support immediate reboots with "systemctl reboot -ff"
16755 * Show /etc/os-release data in systemd-analyze output
16757 * Many bugfixes for the journal, including endianness fixes and
16758 ensuring that disk space enforcement works
16760 * sd-login.h is C++ compatible again
16762 * Extend the /etc/os-release format on request of the Debian
16765 * We now refuse non-UTF8 strings used in various configuration
16766 and unit files. This is done to ensure we do not pass invalid
16767 data over D-Bus or expose it elsewhere.
16769 * Register Mimo USB Screens as suitable for automatic seat
16772 * Read SELinux client context from journal clients in a race
16775 * Reorder configuration file lookup order. /etc now always
16776 overrides /run in order to allow the administrator to always
16777 and unconditionally override vendor-supplied or
16778 automatically generated data.
16780 * The various user visible bits of the journal now have man
16781 pages. We still lack man pages for the journal API calls
16784 * We now ship all man pages in HTML format again in the
16787 Contributions from: Dave Reisner, Dirk Eibach, Frederic
16788 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
16789 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
16794 * This is mostly a bugfix release
16796 * systems lacking /etc/os-release are no longer supported.
16798 * Various functionality updates to libsystemd-login.so
16800 * Track class of PAM logins to distinguish greeters from
16801 normal user logins.
16803 Contributions from: Kay Sievers, Lennart Poettering, Michael
16808 * This is an important bugfix release for v41.
16810 * Building man pages is now optional which should be useful
16811 for those building systemd from git but unwilling to install
16814 * Watchdog support for supervising services is now usable. In
16815 a future release support for hardware watchdogs
16816 (i.e. /dev/watchdog) will be added building on this.
16818 * Service start rate limiting is now configurable and can be
16819 turned off per service. When a start rate limit is hit a
16820 reboot can automatically be triggered.
16822 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
16824 Contributions from: Benjamin Franzke, Bill Nottingham,
16825 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
16826 Schmidt, Michał Górny, Piotr Drąg
16830 * The systemd binary is installed /usr/lib/systemd/systemd now;
16831 An existing /sbin/init symlink needs to be adapted with the
16834 * The code that loads kernel modules has been ported to invoke
16835 libkmod directly, instead of modprobe. This means we do not
16836 support systems with module-init-tools anymore.
16838 * Watchdog support is now already useful, but still not
16841 * A new kernel command line option systemd.setenv= is
16842 understood to set system wide environment variables
16843 dynamically at boot.
16845 * We now limit the set of capabilities of systemd-journald.
16847 * We now set SIGPIPE to ignore by default, since it only is
16848 useful in shell pipelines, and has little use in general
16849 code. This can be disabled with IgnoreSIPIPE=no in unit
16852 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
16853 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
16858 * This is mostly a bugfix release
16860 * We now expose the reason why a service failed in the
16861 "Result" D-Bus property.
16863 * Rudimentary service watchdog support (will be completed over
16864 the next few releases.)
16866 * When systemd forks off in order execute some service we will
16867 now immediately changes its argv[0] to reflect which process
16868 it will execute. This is useful to minimize the time window
16869 with a generic argv[0], which makes bootcharts more useful
16871 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
16872 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
16873 Mike Kazantsev, Ray Strode
16877 * This is mostly a test release, but incorporates many
16880 * New systemd-cgtop tool to show control groups by their
16883 * Linking against libacl for ACLs is optional again. If
16884 disabled, support tracking device access for active logins
16885 goes becomes unavailable, and so does access to the user
16886 journals by the respective users.
16888 * If a group "adm" exists, journal files are automatically
16889 owned by them, thus allow members of this group full access
16890 to the system journal as well as all user journals.
16892 * The journal now stores the SELinux context of the logging
16893 client for all entries.
16895 * Add C++ inclusion guards to all public headers
16897 * New output mode "cat" in the journal to print only text
16898 messages, without any meta data like date or time.
16900 * Include tiny X server wrapper as a temporary stop-gap to
16901 teach XOrg udev display enumeration. This is used by display
16902 managers such as gdm, and will go away as soon as XOrg
16903 learned native udev hotplugging for display devices.
16905 * Add new systemd-cat tool for executing arbitrary programs
16906 with STDERR/STDOUT connected to the journal. Can also act as
16907 BSD logger replacement, and does so by default.
16909 * Optionally store all locally generated coredumps in the
16910 journal along with meta data.
16912 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
16913 writing short strings to files (for usage for /sys), and for
16914 creating symlinks, character and block device nodes.
16916 * New unit file option ControlGroupPersistent= to make cgroups
16917 persistent, following the mechanisms outlined in
16918 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
16920 * Support multiple local RTCs in a sane way
16922 * No longer monopolize IO when replaying readahead data on
16923 rotating disks, since we might starve non-file-system IO to
16924 death, since fanotify() will not see accesses done by blkid,
16927 * Do not show kernel threads in systemd-cgls anymore, unless
16928 requested with new -k switch.
16930 Contributions from: Dan Horák, Kay Sievers, Lennart
16931 Poettering, Michal Schmidt
16935 * This is mostly a test release, but incorporates many
16938 * The git repository moved to:
16939 git://anongit.freedesktop.org/systemd/systemd
16940 ssh://git.freedesktop.org/git/systemd/systemd
16942 * First release with the journal
16943 https://0pointer.de/blog/projects/the-journal.html
16945 * The journal replaces both systemd-kmsg-syslogd and
16946 systemd-stdout-bridge.
16948 * New sd_pid_get_unit() API call in libsystemd-logind
16950 * Many systemadm clean-ups
16952 * Introduce remote-fs-pre.target which is ordered before all
16953 remote mounts and may be used to start services before all
16956 * Added Mageia support
16958 * Add bash completion for systemd-loginctl
16960 * Actively monitor PID file creation for daemons which exit in
16961 the parent process before having finished writing the PID
16962 file in the daemon process. Daemons which do this need to be
16963 fixed (i.e. PID file creation must have finished before the
16964 parent exits), but we now react a bit more gracefully to them.
16966 * Add colourful boot output, mimicking the well-known output
16967 of existing distributions.
16969 * New option PassCredentials= for socket units, for
16970 compatibility with a recent kernel ABI breakage.
16972 * /etc/rc.local is now hooked in via a generator binary, and
16973 thus will no longer act as synchronization point during
16976 * systemctl list-unit-files now supports --root=.
16978 * systemd-tmpfiles now understands two new commands: z, Z for
16979 relabelling files according to the SELinux database. This is
16980 useful to apply SELinux labels to specific files in /sys,
16981 among other things.
16983 * Output of SysV services is now forwarded to both the console
16984 and the journal by default, not only just the console.
16986 * New man pages for all APIs from libsystemd-login.
16988 * The build tree got reorganized and the build system is a
16989 lot more modular allowing embedded setups to specifically
16990 select the components of systemd they are interested in.
16992 * Support for Linux systems lacking the kernel VT subsystem is
16995 * configure's --with-rootdir= got renamed to
16996 --with-rootprefix= to follow the naming used by udev and
16999 * Unless specified otherwise we will now install to /usr instead
17000 of /usr/local by default.
17002 * Processes with '@' in argv[0][0] are now excluded from the
17003 final shut-down killing spree, following the logic explained
17005 https://systemd.io/ROOT_STORAGE_DAEMONS/
17007 * All processes remaining in a service cgroup when we enter
17008 the START or START_PRE states are now killed with
17009 SIGKILL. That means it is no longer possible to spawn
17010 background processes from ExecStart= lines (which was never
17011 supported anyway, and bad style).
17013 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
17014 reloading of units together.
17016 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
17017 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
17018 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
17019 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
17020 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek