1 systemd System and Service Manager
3 CHANGES WITH 256 in spe:
5 Announcements of Future Feature Removals and Incompatible Changes:
7 * Support for flushing of the nscd user/group database caches will be
8 dropped in a future release.
10 * Support for cgroup v1 ('legacy' and 'hybrid' hierarchies) is now
11 considered obsolete and systemd by default will refuse to boot under it.
12 To forcibly reenable cgroup v1 support, SYSTEMD_CGROUP_ENABLE_LEGACY_FORCE=1
13 must be set on kernel command line. The meson option 'default-hierarchy='
14 is also deprecated, i.e. only cgroup v2 ('unified' hierarchy) can be
15 selected as build-time default.
17 * Previously, systemd-networkd did not explicitly remove any bridge VLAN
18 IDs assigned on bridge master and ports. Since v256, if a .network
19 file for an interface has at least one valid settings in [BridgeVLAN]
20 section, then all assigned VLAN IDs on the interface that are not
21 configured in the .network file are removed.
23 * systemd-gpt-auto-generator will stop generating units for ESP or
24 XBOOTLDR partitions if it finds mount entries in the /boot/ or /efi/
25 hierarchies in fstab. This is to prevent the generator from
26 interfering with systems where ESP is explicitly configured to be
27 mounted at some path, for example /boot/efi/ (this type of setup is
28 obsolete but is still commonly found).
30 * The behavior of systemd-sleep and systemd-homed has been updated to
31 freeze user sessions when entering the various sleep modes or when
32 locking a homed-managed home area. This is known to cause issues with
33 the proprietary NVIDIA drivers. Packagers of the NVIDIA proprietary
34 drivers may want to add drop-in configuration files that set
35 SYSTEMD_SLEEP_FREEZE_USER_SESSION=false for systemd-suspend.service
36 and related services, and SYSTEMD_HOME_LOCK_FREEZE_SESSION=false for
37 systemd-homed.service.
41 * systemd-networkd's proxy support gained a new option to configure
42 a private VLAN variant of the proxy ARP supported by the kernel
43 under the name IPv4ProxyARPPrivateVLAN=.
45 * TPM 1.2 PCR measurement support has been removed from
46 systemd-stub. TPM 1.2 is obsolete and — due to the (by today's
47 standards) weak cryptographic algorithms it only supports — does not
48 actually provide the security benefits it's supposed to
49 provide. Given that the rest of systemd's codebase never supported
50 TPM 1.2 the support has now been removed from systemd-stub as well.
54 Announcements of Future Feature Removals and Incompatible Changes:
56 * Support for split-usr (/usr/ mounted separately during late boot,
57 instead of being mounted by the initrd before switching to the rootfs)
58 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
59 /usr/lib/, …) has been removed. For more details, see:
60 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
62 * We intend to remove cgroup v1 support from a systemd release after
63 the end of 2023. If you run services that make explicit use of
64 cgroup v1 features (i.e. the "legacy hierarchy" with separate
65 hierarchies for each controller), please implement compatibility with
66 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
67 Most of Linux userspace has been ported over already.
69 * Support for System V service scripts is now deprecated and will be
70 removed in a future release. Please make sure to update your software
71 *now* to include a native systemd unit file instead of a legacy
72 System V script to retain compatibility with future systemd releases.
74 * Support for the SystemdOptions EFI variable is deprecated.
75 'bootctl systemd-efi-options' will emit a warning when used. It seems
76 that this feature is little-used and it is better to use alternative
77 approaches like credentials and confexts. The plan is to drop support
78 altogether at a later point, but this might be revisited based on
81 * systemd-run's switch --expand-environment= which currently is disabled
82 by default when combined with --scope, will be changed in a future
83 release to be enabled by default.
85 * "systemctl switch-root" is now restricted to initrd transitions only.
87 Transitions between real systems should be done with
88 "systemctl soft-reboot" instead.
90 * The "ip=off" and "ip=none" kernel command line options interpreted by
91 systemd-network-generator will now result in IPv6RA + link-local
92 addressing being disabled, too. Previously DHCP was turned off, but
93 IPv6RA and IPv6 link-local addressing was left enabled.
95 * The NAMING_BRIDGE_MULTIFUNCTION_SLOT naming scheme has been deprecated
98 * SuspendMode=, HibernateState= and HybridSleepState= in the [Sleep]
99 section of systemd-sleep.conf are now deprecated and have no effect.
100 They did not (and could not) take any value other than the respective
101 default. HybridSleepMode= is also deprecated, and will now always use
102 the 'suspend' disk mode.
106 * The way services are spawned has been overhauled. Previously, a
107 process was forked that shared all of the manager's memory (via
108 copy-on-write) while doing all the required setup (e.g.: mount
109 namespaces, CGroup configuration, etc.) before exec'ing the target
110 executable. This was problematic for various reasons: several glibc
111 APIs were called that are not supposed to be used after a fork but
112 before an exec, copy-on-write meant that if either process (the
113 manager or the child) touched a memory page a copy was triggered, and
114 also the memory footprint of the child process was that of the
115 manager, but with the memory limits of the service. From this version
116 onward, the new process is spawned using CLONE_VM and CLONE_VFORK
117 semantics via posix_spawn(3), and it immediately execs a new internal
118 binary, systemd-executor, that receives the configuration to apply
119 via memfd, and sets up the process before exec'ing the target
120 executable. The systemd-executor binary is pinned by file descriptor
121 by each manager instance (system and users), and the reference is
122 updated on daemon-reexec - it is thus important to reexec all running
123 manager instances when the systemd-executor and/or libsystemd*
124 libraries are updated on the filesystem.
126 * Most of the internal process tracking is being changed to use PIDFDs
127 instead of PIDs when the kernel supports it, to improve robustness
130 * A new option SurviveFinalKillSignal= can be used to configure the
131 unit to be skipped in the final SIGTERM/SIGKILL spree on shutdown.
132 This is part of the required configuration to let a unit's processes
133 survive a soft-reboot operation.
135 * System extension images (sysext) can now set
136 EXTENSION_RELOAD_MANAGER=1 in their extension-release files to
137 automatically reload the service manager (PID 1) when
138 merging/refreshing/unmerging on boot. Generally, while this can be
139 used to ship services in system extension images it's recommended to
140 do that via portable services instead.
142 * The ExtensionImages= and ExtensionDirectories= options now support
143 confexts images/directories.
145 * A new option NFTSet= provides a method for integrating dynamic cgroup
146 IDs into firewall rules with NFT sets. The benefit of using this
147 setting is to be able to use control group as a selector in firewall
148 rules easily and this in turn allows more fine grained filtering.
149 Also, NFT rules for cgroup matching use numeric cgroup IDs, which
150 change every time a service is restarted, making them hard to use in
153 * A new option CoredumpReceive= can be set for service and scope units,
154 together with Delegate=yes, to make systemd-coredump on the host
155 forward core files from processes crashing inside the delegated
156 CGroup subtree to systemd-coredump running in the container. This new
157 option is by default used by systemd-nspawn containers that use the
160 * A new ConditionSecurity=measured-uki option is now available, to ensure
161 a unit can only run when the system has been booted from a measured UKI.
163 * MemoryAvailable= now considers physical memory if there are no CGroup
164 memory limits set anywhere in the tree.
166 * The $USER environment variable is now always set for services, while
167 previously it was only set if User= was specified. A new option
168 SetLoginEnvironment= is now supported to determine whether to also set
169 $HOME, $LOGNAME, and $SHELL.
171 * Socket units now support a new pair of
172 PollLimitBurst=/PollLimitInterval= options to configure a limit on
173 how often polling events on the file descriptors backing this unit
174 will be considered within a time window.
176 * Scope units can now be created using PIDFDs instead of PIDs to select
177 the processes they should include.
179 * Sending SIGRTMIN+18 with 0x500 as sigqueue() value will now cause the
180 manager to dump the list of currently pending jobs.
182 * If the kernel supports MOVE_MOUNT_BENEATH, the systemctl and
183 machinectl bind and mount-image verbs will now cause the new mount to
184 replace the old mount (if any), instead of overmounting it.
186 * Units now have MemoryPeak, MemorySwapPeak, MemorySwapCurrent and
187 MemoryZSwapCurrent properties, which respectively contain the values
188 of the cgroup v2's memory.peak, memory.swap.peak, memory.swap.current
189 and memory.zswap.current properties. This information is also shown in
190 "systemctl status" output, if available.
192 TPM2 Support + Disk Encryption & Authentication:
194 * systemd-cryptenroll now allows specifying a PCR bank and explicit hash
195 value in the --tpm2-pcrs= option.
197 * systemd-cryptenroll now allows specifying a TPM2 key handle (nv
198 index) to be used instead of the default SRK via the new
199 --tpm2-seal-key-handle= option.
201 * systemd-cryptenroll now allows TPM2 enrollment using only a TPM2
202 public key (in TPM2B_PUBLIC format) – without access to the TPM2
203 device itself – which enables offline sealing of LUKS images for a
204 specific TPM2 chip, as long as the SRK public key is known. Pass the
205 public to the tool via the new --tpm2-device-key= switch.
207 * systemd-cryptsetup is now installed in /usr/bin/ and is no longer an
208 internal-only executable.
210 * The TPM2 Storage Root Key will now be set up, if not already present,
211 by a new systemd-tpm2-setup.service early boot service. The SRK will
212 be stored in PEM format and TPM2_PUBLIC format (the latter is useful
213 for systemd-cryptenroll --tpm2-device-key=, as mentioned above) for
214 easier access. A new "srk" verb has been added to systemd-analyze to
215 allow extracting it on demand if it is already set up.
217 * The internal systemd-pcrphase executable has been renamed to
220 * The systemd-pcrextend tool gained a new --pcr= switch to override
221 which PCR to measure into.
223 * systemd-pcrextend now exposes a Varlink interface at
224 io.systemd.PCRExtend that can be used to do measurements and event
227 * TPM measurements are now also written to an event log at
228 /run/log/systemd/tpm2-measure.log, using a derivative of the TCG
229 Canonical Event Log format. Previously we'd only log them to the
230 journal, where they however were subject to rotation and similar.
232 * A new component "systemd-pcrlock" has been added that allows managing
233 local TPM2 PCR policies for PCRs 0-7 and similar, which are hard to
234 predict by the OS vendor because of the inherently local nature of
235 what measurements they contain, such as firmware versions of the
236 system and extension cards and suchlike. pcrlock can predict PCR
237 measurements ahead of time based on various inputs, such as the local
238 TPM2 event log, GPT partition tables, PE binaries, UKI kernels, and
239 various other things. It can then pre-calculate a TPM2 policy from
240 this, which it stores in an TPM2 NV index. TPM2 objects (such as disk
241 encryption keys) can be locked against this NV index, so that they
242 are locked against a specific combination of system firmware and
243 state. Alternatives for each component are supported to allowlist
244 multiple kernel versions or boot loader version simultaneously
245 without losing access to the disk encryption keys. The tool can also
246 be used to analyze and validate the local TPM2 event log.
247 systemd-cryptsetup, systemd-cryptenroll, systemd-repart have all been
248 updated to support such policies. There's currently no support for
249 locking the system's root disk against a pcrlock policy, this will be
250 added soon. Moreover, it is currently not possible to combine a
251 pcrlock policy with a signed PCR policy. This component is
252 experimental and its public interface is subject to change.
254 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
256 * bootctl will now show whether the system was booted from a UKI in its
259 * systemd-boot and systemd-stub now use different project keys in their
260 respective SBAT sections, so that they can be revoked individually if
263 * systemd-boot will no longer load unverified Devicetree blobs when UEFI
264 SecureBoot is enabled. For more details see:
265 https://github.com/systemd/systemd/security/advisories/GHSA-6m6p-rjcq-334c
267 * systemd-boot gained new hotkeys to reboot and power off the system
268 from the boot menu ("B" and "O"). If the "auto-poweroff" and
269 "auto-reboot" options in loader.conf are set these entries are also
270 shown as menu items (which is useful on devices lacking a regular
273 * systemd-boot gained a new configuration value "menu-disabled" for the
274 set-timeout option, to allow completely disabling the boot menu,
275 including the hotkey.
277 * systemd-boot will now measure the content of loader.conf in TPM2
280 * systemd-stub will now concatenate the content of all kernel
281 command-line addons before measuring them in TPM2 PCR 12, in a single
282 measurement, instead of measuring them individually.
284 * systemd-stub will now measure and load Devicetree Blob addons, which
285 are searched and loaded following the same model as the existing
286 kernel command-line addons.
288 * systemd-stub will now ignore unauthenticated kernel command line options
289 passed from systemd-boot when running inside Confidential VMs with UEFI
292 * systemd-stub will now load a Devicetree blob even if the firmware did
293 not load any beforehand (e.g.: for ACPI systems).
295 * ukify is no longer considered experimental, and now ships in /usr/bin/.
297 * ukify gained a new verb inspect to describe the sections of a UKI and
298 print the contents of the well-known sections.
300 * ukify gained a new verb genkey to generate a set of key pairs for
301 signing UKIs and their PCR data.
303 * The 90-loaderentry kernel-install hook now supports installing device
306 * kernel-install now supports the --json=, --root=, --image=, and
307 --image-policy= options for the inspect verb.
309 * kernel-install now supports new list and add-all verbs. The former
310 lists all installed kernel images (if those are available in
311 /usr/lib/modules/). The latter will install all the kernels it can
316 * A new option --copy-from= has been added that synthesizes partition
317 definitions from the given image, which are then applied by the
318 systemd-repart algorithm.
320 * A new option --copy-source= has been added, which can be used to specify
321 a directory to which CopyFiles= is considered relative to.
323 * New --make-ddi=confext, --make-ddi=sysext, and --make-ddi=portable
324 options have been added to make it easier to generate these types of
325 DDIs, without having to provide repart.d definitions for them.
327 * The dm-verity salt and UUID will now be derived from the specified
330 * New VerityDataBlockSizeBytes= and VerityHashBlockSizeBytes= can now be
331 configured in repart.d/ configuration files.
333 * A new Subvolumes= setting is now supported in repart.d/ configuration
334 files, to indicate which directories in the target partition should be
337 * A new --tpm2-device-key= option can be used to lock a disk against a
338 specific TPM2 public key. This matches the same switch the
339 systemd-cryptenroll tool now supports (see above).
343 * The journalctl --lines= parameter now accepts +N to show the oldest N
344 entries instead of the newest.
346 * journald now ensures that sealing happens once per epoch, and sets a
347 new compatibility flag to distinguish old journal files that were
348 created before this change, for backward compatibility.
352 * udev will now create symlinks to loopback block devices in the
353 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
354 string field selected during allocation. The systemd-dissect tool and
355 the util-linux losetup command now supports a complementing new switch
356 --loop-ref= for selecting the string. This means a loopback block
357 device may now be allocated under a caller-chosen reference and can
358 subsequently be referenced without first having to look up the block
359 device name the caller ended up with.
361 * udev also creates symlinks to loopback block devices in the
362 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
363 of the inode attached to the loopback block device. This means that
364 attaching a file to a loopback device will implicitly make a handle
365 available to be found via that file's inode information.
367 * udevadm info gained support for JSON output via a new --json= flag, and
368 for filtering output using the same mechanism that udevadm trigger
371 * The predictable network interface naming logic is extended to include
372 the SR-IOV-R "representor" information in network interface names.
373 This feature was intended for v254, but even though the code was
374 merged, the part that actually enabled the feature was forgotten.
375 It is now enabled by default and is part of the new "v255" naming
378 * A new hwdb/rules file has been added that sets the
379 ID_NET_AUTO_LINK_LOCAL_ONLY=1 udev property on all network interfaces
380 that should usually only be configured with link-local addressing
381 (IPv4LL + IPv6LL), i.e. for PC-to-PC cables ("laplink") or
382 Thunderbolt networking. systemd-networkd and NetworkManager (soon)
383 will make use of this information to apply an appropriate network
384 configuration by default.
386 * The ID_NET_DRIVER property on network interfaces is now set
387 relatively early in the udev rule set so that other rules may rely on
388 its use. This is implemented in a new "net-driver" udev built-in.
392 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
393 is now dropped, as it never worked, hence it should not be used by
396 * The 'prefixstable' ipv6 address generation mode now considers the SSID
397 when generating stable addresses, so that a different stable address
398 is used when roaming between wireless networks. If you already use
399 'prefixstable' addresses with wireless networks, the stable address
400 will be changed by the update.
402 * The DHCPv4 client gained a RapidCommit option, true by default, which
403 enables RFC4039 Rapid Commit behavior to obtain a lease in a
404 simplified 2-message exchange instead of the typical 4-message
405 exchange, if also supported by the DHCP server.
407 * The DHCPv4 client gained new InitialCongestionWindow= and
408 InitialAdvertisedReceiveWindow= options for route configurations.
410 * The DHCPv4 client gained a new RequestAddress= option that allows
411 to send a preferred IP address in the initial DHCPDISCOVER message.
413 * The DHCPv4 server and client gained support for IPv6-only mode
416 * The SendHostname= and Hostname= options are now available for the
417 DHCPv6 client, independently of the DHCPv4= option, so that these
418 configuration values can be set independently for each client.
420 * The DHCPv4 and DHCPv6 client state can now be queried via D-Bus,
421 including lease information.
423 * The DHCPv6 client can now be configured to use a custom DUID type.
425 * .network files gained a new IPv4ReversePathFilter= setting in the
426 [Network] section, to control sysctl's rp_filter setting.
428 * .network files gaiend a new HopLimit= setting in the [Route] section,
429 to configure a per-route hop limit.
431 * .network files gained a new TCPRetransmissionTimeoutSec= setting in
432 the [Route] section, to configure a per-route TCP retransmission
435 * A new directive NFTSet= provides a method for integrating network
436 configuration into firewall rules with NFT sets. The benefit of using
437 this setting is that static network configuration or dynamically
438 obtained network addresses can be used in firewall rules with the
439 indirection of NFT set types.
441 * The [IPv6AcceptRA] section supports the following new options:
442 UsePREF64=, UseHopLimit=, UseICMP6RateLimit=, and NFTSet=.
444 * The [IPv6SendRA] section supports the following new options:
445 RetransmitSec=, HopLimit=, HomeAgent=, HomeAgentLifetimeSec=, and
446 HomeAgentPreference=.
448 * A new [IPv6PREF64Prefix] set of options, containing Prefix= and
449 LifetimeSec=, has been introduced to append pref64 options in router
450 advertisements (RFC8781).
452 * The network generator now configures the interfaces with only
453 link-local addressing if "ip=link-local" is specified on the kernel
456 * The prefix of the configuration files generated by the network
457 generator from the kernel command line is now prefixed with '70-',
458 to make them have higher precedence over the default configuration
461 * Added a new -Ddefault-network=BOOL meson option, that causes more
462 .network files to be installed as enabled by default. These configuration
463 files will which match generic setups, e.g. 89-ethernet.network matches
464 all Ethernet interfaces and enables both DHCPv4 and DHCPv6 clients.
466 * If a ID_NET_MANAGED_BY= udev property is set on a network device and
467 it is any other string than "io.systemd.Network" then networkd will
468 not manage this device. This may be used to allow multiple network
469 management services to run in parallel and assign ownership of
470 specific devices explicitly. NetworkManager will soon implement a
475 * systemctl is-failed now checks the system state if no unit is
478 * systemctl will now automatically soft-reboot if a new root file system
479 is found under /run/nextroot/ when a reboot operation is invoked.
483 * Wall messages now work even when utmp support is disabled, using
484 systemd-logind to query the necessary information.
486 * systemd-logind now sends a new PrepareForShutdownWithMetadata D-Bus
487 signal before shutdown/reboot/soft-reboot that includes additional
488 information compared to the PrepareForShutdown signal. Currently the
489 additional information is the type of operation that is about to be
492 Hibernation & Suspend:
494 * The kernel and OS versions will no longer be checked on resume from
497 * Hibernation into swap files backed by btrfs are now
498 supported. (Previously this was supported only for other file
503 * A new systemd-vmspawn tool has been added, that aims to provide for VMs
504 the same interfaces and functionality that systemd-nspawn provides for
505 containers. For now it supports QEMU as a backend, and exposes some of
506 its options to the user. This component is experimental and its public
507 interface is subject to change.
509 * "systemd-analyze plot" has gained tooltips on each unit name with
510 related-unit information in its svg output, such as Before=,
511 Requires=, and similar properties.
513 * A new varlinkctl tool has been added to allow interfacing with
514 Varlink services, and introspection has been added to all such
515 services. This component is experimental and its public interface is
518 * systemd-sysext and systemd-confext now expose a Varlink service
519 at io.systemd.sysext.
521 * portable services now accept confexts as extensions.
523 * systemd-sysupdate now accepts directories in the MatchPattern= option.
525 * systemd-run will now output the invocation ID of the launched
526 transient unit and its peak memory usage.
528 * systemd-analyze, systemd-tmpfiles, systemd-sysusers, systemd-sysctl,
529 and systemd-binfmt gained a new --tldr option that can be used instead
530 of --cat-config to suppress uninteresting configuration lines, such as
531 comments and whitespace.
533 * resolvectl gained a new "show-server-state" command that shows
534 current statistics of the resolver. This is backed by a new
535 DumpStatistics() Varlink method provided by systemd-resolved.
537 * systemd-timesyncd will now emit a D-Bus signal when the LinkNTPServers
540 * vconsole now supports KEYMAP=@kernel for preserving the kernel keymap
543 * seccomp now supports the LoongArch64 architecture.
545 * seccomp may now be enabled for services running as a non-root User=
546 without NoNewPrivileges=yes.
548 * systemd-id128 now supports a new -P option to show only values. The
549 combination of -P and --app options is also supported.
551 * A new pam_systemd_loadkey.so PAM module is now available, which will
552 automatically fetch the passphrase used by cryptsetup to unlock the
553 root file system and set it as the PAM authtok. This enables, among
554 other things, configuring auto-unlock of the GNOME Keyring / KDE
555 Wallet when autologin is configured.
557 * Many meson options now use the 'feature' type, which means they
558 take enabled/disabled/auto as values.
560 * A new meson option -Dconfigfiledir= can be used to change where
561 configuration files with default values are installed to.
563 * Options and verbs in man pages are now tagged with the version they
564 were first introduced in.
566 * A new component "systemd-storagetm" has been added, which exposes all
567 local block devices as NVMe-TCP devices, fully automatically. It's
568 hooked into a new target unit storage-target-mode.target that is
569 suppsoed to be booted into via
570 rd.systemd.unit=storage-target-mode.target on the kernel command
571 line. This is intended to be used for installers and debugging to
572 quickly get access to the local disk. It's inspired by MacOS "target
573 disk mode". This component is experimental and its public interface is
576 * A new component "systemd-bsod" has been added, which can show logged
577 error messages full screen, if they have a log level of LOG_EMERG log
578 level. This component is experimental and its public interface is
581 * The systemd-dissect tool's --with command will now set the
582 $SYSTEMD_DISSECT_DEVICE environment variable to the block device it
583 operates on for the invoked process.
585 * The systemd-mount tool gained a new --tmpfs switch for mounting a new
586 'tmpfs' instance. This is useful since it does so via .mount units
587 and thus can be executed remotely or in containers.
589 * The various tools in systemd that take "verbs" (such as systemctl,
590 loginctl, machinectl, …) now will suggest a close verb name in case
591 the user specified an unrecognized one.
593 * libsystemd now exports a new function sd_id128_get_app_specific()
594 that generates "app-specific" 128bit IDs from any ID. It's similar to
595 sd_id128_get_machine_app_specific() and
596 sd_id128_get_boot_app_specific() but takes the ID to base calculation
597 on as input. This new functionality is also exposed in the
598 "systemd-id128" tool where you can now combine --app= with `show`.
600 * All tools that parse timestamps now can also parse RFC3339 style
601 timestamps that include the "T" and Z" characters.
603 * New documentation has been added:
605 https://systemd.io/FILE_DESCRIPTOR_STORE
606 https://systemd.io/TPM2_PCR_MEASUREMENTS
607 https://systemd.io/MOUNT_REQUIREMENTS
609 * The codebase now recognizes the suffix .confext.raw and .sysext.raw
610 as alternative to the .raw suffix generally accepted for DDIs. It is
611 recommended to name configuration extensions and system extensions
612 with such suffixes, to indicate their purpose in the name.
614 * The sd-device API gained a new function
615 sd_device_enumerator_add_match_property_required() which allows
616 configuring matches on properties that are strictly required. This is
617 different from the existing sd_device_enumerator_add_match_property()
618 matches of which one one needs to apply.
620 * The MAC address the veth side of an nspawn container shall get
621 assigned may now be controlled via the $SYSTEMD_NSPAWN_NETWORK_MAC
622 environment variable.
624 * The libiptc dependency is now implemented via dlopen(), so that tools
625 such as networkd and nspawn no longer have a hard dependency on the
626 shared library when compiled with support for libiptc.
628 * New rpm macros have been added: %systemd_user_daemon_reexec does
629 daemon-reexec for all user managers, and %systemd_postun_with_reload
630 and %systemd_user_postun_with_reload do a reload for system and user
633 * coredumpctl now propagates SIGTERM to the debugger process.
635 Contributions from: 김인수, Abderrahim Kitouni, Adam Goldman,
636 Adam Williamson, Alexandre Peixoto Ferreira, Alex Hudspith,
637 Alvin Alvarado, André Paiusco, Antonio Alvarez Feijoo,
638 Anton Lundin, Arian van Putten, Arseny Maslennikov, Arthur Shau,
639 Balázs Úr, beh_10257, Benjamin Peterson, Bertrand Jacquin,
640 Brian Norris, Charles Lee, Cheng-Chia Tseng, Chris Patterson,
641 Christian Hergert, Christian Hesse, Christian Kirbach,
642 Clayton Craft, commondservice, cunshunxia, Curtis Klein, cvlc12,
643 Daan De Meyer, Daniele Medri, Daniel P. Berrangé, Daniel Rusek,
644 Daniel Thompson, Dan Nicholson, Dan Streetman, David Rheinsberg,
645 David Santamaría Rogado, David Tardon, dependabot[bot],
646 Diego Viola, Dmitry V. Levin, Emanuele Giuseppe Esposito,
647 Emil Renner Berthing, Emil Velikov, Etienne Dechamps, Fabian Vogt,
648 felixdoerre, Felix Dörre, Florian Schmaus, Franck Bui,
649 Frantisek Sumsal, G2-Games, Gioele Barabucci, Hugo Carvalho,
650 huyubiao, Iago López Galeiras, IllusionMan1212, Jade Lovelace,
651 janana, Jan Janssen, Jan Kuparinen, Jan Macku, Jeremy Fleischman,
652 Jin Liu, jjimbo137, Joerg Behrmann, Johannes Segitz, Jordan Rome,
653 Jordan Williams, Julien Malka, Juno Computers, Khem Raj, khm,
654 Kingbom Dou, Kiran Vemula, Krzesimir Nowak, Laszlo Gombos,
655 Lennart Poettering, linuxlion, Luca Boccassi, Lucas Adriano Salles,
656 Lukas, Lukáš Nykrýn, Maanya Goenka, Maarten, Malte Poll,
657 Marc Pervaz Boocha, Martin Beneš, Martin Joerg, Martin Wilck,
658 Mathieu Tortuyaux, Matthias Schiffer, Maxim Mikityanskiy,
659 Max Kellermann, Michael A Cassaniti, Michael Biebl, Michael Kuhn,
660 Michael Vasseur, Michal Koutný, Michal Sekletár, Mike Yuan,
661 Milton D. Miller II, mordner, msizanoen, NAHO, Nandakumar Raghavan,
662 Neil Wilson, Nick Rosbrook, Nils K, NRK, Oğuz Ersen,
663 Omojola Joshua, onenowy, Paul Meyer, Paymon MARANDI, pelaufer,
664 Peter Hutterer, PhylLu, Pierre GRASSER, Piotr Drąg, Priit Laes,
665 Rahil Bhimjiani, Raito Bezarius, Raul Cheleguini, Reto Schneider,
666 Richard Maw, Robby Red, RoepLuke, Roland Hieber, Roland Singer,
667 Ronan Pigott, Sam James, Sam Leonard, Sergey A, Susant Sahani,
668 Sven Joachim, Tad Fisher, Takashi Sakamoto, Thorsten Kukuk, Tj,
669 Tomasz Świątek, Topi Miettinen, Valentin David,
670 Valentin Lefebvre, Victor Westerhuis, Vincent Haupert,
671 Vishal Chillara Srinivas, Vito Caputo, Warren, Weblate,
672 Xiaotian Wu, xinpeng wang, Yaron Shahrabani, Yo-Jung Lin,
673 Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zeroskyx,
674 Дамјан Георгиевски, наб
676 — Edinburgh, 2023-12-06
680 Announcements of Future Feature Removals and Incompatible Changes:
682 * The next release (v255) will remove support for split-usr (/usr/
683 mounted separately during late boot, instead of being mounted by the
684 initrd before switching to the rootfs) and unmerged-usr (parallel
685 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
687 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
689 * We intend to remove cgroup v1 support from a systemd release after
690 the end of 2023. If you run services that make explicit use of
691 cgroup v1 features (i.e. the "legacy hierarchy" with separate
692 hierarchies for each controller), please implement compatibility with
693 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
694 Most of Linux userspace has been ported over already.
696 * Support for System V service scripts is now deprecated and will be
697 removed in a future release. Please make sure to update your software
698 *now* to include a native systemd unit file instead of a legacy
699 System V script to retain compatibility with future systemd releases.
701 * Support for the SystemdOptions EFI variable is deprecated.
702 'bootctl systemd-efi-options' will emit a warning when used. It seems
703 that this feature is little-used and it is better to use alternative
704 approaches like credentials and confexts. The plan is to drop support
705 altogether at a later point, but this might be revisited based on
708 * EnvironmentFile= now treats the line following a comment line
709 trailing with escape as a non comment line. For details, see:
710 https://github.com/systemd/systemd/issues/27975
712 * PrivateNetwork=yes and NetworkNamespacePath= now imply
713 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
715 * Behaviour of sandboxing options for the per-user service manager
716 units has changed. They now imply PrivateUsers=yes, which means user
717 namespaces will be implicitly enabled when a sandboxing option is
718 enabled in a user unit. Enabling user namespaces has the drawback
719 that system users will no longer be visible (and processes/files will
720 appear as owned by 'nobody') in the user unit.
722 By definition a sandboxed user unit should run with reduced
723 privileges, so impact should be small. This will remove a great
724 source of confusion that has been reported by users over the years,
725 due to how these options require an extra setting to be manually
726 enabled when used in the per-user service manager, which is not
727 needed in the system service manager. For more details, see:
728 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
730 * systemd-run's switch --expand-environment= which currently is disabled
731 by default when combined with --scope, will be changed in a future
732 release to be enabled by default.
734 Security Relevant Changes:
736 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
737 process capability to invoked session processes of regular users on
738 local seats (as well as to systemd --user), unless configured
739 otherwise via data from JSON user records, or via the PAM module's
740 parameter list. This is useful in order allow desktop tools such as
741 GNOME's Alarm Clock application to set a timer for
742 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
743 per-user service unit file may thus use AmbientCapability= to pass
744 the capability to invoked processes. Note that this capability is
745 relatively narrow in focus (in particular compared to other process
746 capabilities such as CAP_SYS_ADMIN) and we already — by default —
747 permit more impactful operations such as system suspend to local
752 * Memory limits that apply while the unit is activating are now
753 supported. Previously IO and CPU settings were already supported via
754 StartupCPUWeight= and similar. The same logic has been added for the
755 various manager and unit memory settings (DefaultStartupMemoryLow=,
756 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
757 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
759 * The service manager gained support for enqueuing POSIX signals to
760 services that carry an additional integer value, exposing the
761 sigqueue() system call. This is accessible via new D-Bus calls
762 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
763 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
764 via the new --kill-value= option.
766 * systemctl gained a new "list-paths" verb, which shows all currently
767 active .path units, similarly to how "systemctl list-timers" shows
768 active timers, and "systemctl list-sockets" shows active sockets.
770 * systemctl gained a new --when= switch which is honoured by the various
771 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
772 scheduling these operations by time, similar in fashion to how this
773 has been supported by SysV shutdown.
775 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
776 supports the new PR_SET_MDWE prctl() call, it is used instead of the
777 seccomp()-based system call filter to achieve the same effect.
779 * A new set of kernel command line options is now understood:
780 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
781 systemd.tty.columns.<name>= allow configuring the TTY type and
782 dimensions for the tty specified via <name>. When systemd invokes a
783 service on a tty (via TTYName=) it will look for these and configure
784 the TTY accordingly. This is particularly useful in VM environments
785 to propagate host terminal settings into the appropriate TTYs of the
788 * A new RootEphemeral= setting is now understood in service units. It
789 takes a boolean argument. If enabled for services that use RootImage=
790 or RootDirectory= an ephemeral copy of the disk image or directory
791 tree is made when the service is started. It is removed automatically
792 when the service is stopped. That ephemeral copy is made using
793 btrfs/xfs reflinks or btrfs snapshots, if available.
795 * The service activation logic gained new settings RestartSteps= and
796 RestartMaxDelaySec= which allow exponentially-growing restart
797 intervals for Restart=.
799 * The service activation logic gained a new setting RestartMode= which
800 can be set to 'direct' to skip the inactive/failed states when
801 restarting, so that dependent units are not notified until the service
802 converges to a final (successful or failed) state. For example, this
803 means that OnSuccess=/OnFailure= units will not be triggered until the
804 service state has converged.
806 * PID 1 will now automatically load the virtio_console kernel module
807 during early initialization if running in a suitable VM. This is done
808 so that early-boot logging can be written to the console if available.
810 * Similarly, virtio-vsock support is loaded early in suitable VM
811 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
812 to the VMM if configured, thus loading this early is beneficial.
814 * A new verb "fdstore" has been added to systemd-analyze to show the
815 current contents of the file descriptor store of a unit. This is
816 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
819 * The service manager will now set a new $FDSTORE environment variable
820 when invoking processes for services that have the file descriptor
823 * A new service option FileDescriptorStorePreserve= has been added that
824 allows tuning the lifecycle of the per-service file descriptor store.
825 If set to "yes", the entries in the fd store are retained even after
826 the service has been fully stopped.
828 * The "systemctl clean" command may now be used to clear the fdstore of
831 * Unit *.preset files gained a new directive "ignore", in addition to
832 the existing "enable" and "disable". As the name suggests, matching
833 units are left unchanged, i.e. neither enabled nor disabled.
835 * Service units gained a new setting DelegateSubgroup=. It takes the
836 name of a sub-cgroup to place any processes the service manager forks
837 off in. Previously, the service manager would place all service
838 processes directly in the top-level cgroup it created for the
839 service. This usually meant that main process in a service with
840 delegation enabled would first have to create a subgroup and move
841 itself down into it, in order to not conflict with the "no processes
842 in inner cgroups" rule of cgroup v2. With this option, this step is
843 now handled by PID 1.
845 * The service manager will now look for .upholds/ directories,
846 similarly to the existing support for .wants/ and .requires/
847 directories. Symlinks in this directory result in Upholds=
850 The [Install] section of unit files gained support for a new
851 UpheldBy= directive to generate .upholds/ symlinks automatically when
854 * The service manager now supports a new kernel command line option
855 systemd.default_device_timeout_sec=, which may be used to override
856 the default timeout for .device units.
858 * A new "soft-reboot" mechanism has been added to the service manager.
859 A "soft reboot" is similar to a regular reboot, except that it
860 affects userspace only: the service manager shuts down any running
861 services and other units, then optionally switches into a new root
862 file system (mounted to /run/nextroot/), and then passes control to a
863 systemd instance in the new file system which then starts the system
864 up again. The kernel is not rebooted and neither is the hardware,
865 firmware or boot loader. This provides a fast, lightweight mechanism
866 to quickly reset or update userspace, without the latency that a full
867 system reset involves. Moreover, open file descriptors may be passed
868 across the soft reboot into the new system where they will be passed
869 back to the originating services. This allows pinning resources
870 across the reboot, thus minimizing grey-out time further. This new
871 reboot mechanism is accessible via the new "systemctl soft-reboot"
874 * Services using RootDirectory= or RootImage= will now have read-only
875 access to a copy of the host's os-release file under
876 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
877 This was already the case for Portable Services, and the feature has
878 now been extended to all services that do not run off the host's
881 * A new service setting MemoryKSM= has been added to enable kernel
882 same-page merging individually for services.
884 * A new service setting ImportCredentials= has been added that augments
885 LoadCredential= and LoadCredentialEncrypted= and searches for
886 credentials to import from the system, and supports globbing.
888 * A new job mode "restart-dependencies" has been added to the service
889 manager (exposed via systemctl --job-mode=). It is only valid when
890 used with "start" jobs, and has the effect that the "start" job will
891 be propagated as "restart" jobs to currently running units that have
892 a BindsTo= or Requires= dependency on the started unit.
894 * A new verb "whoami" has been added to "systemctl" which determines as
895 part of which unit the command is being invoked. It writes the unit
896 name to standard output. If one or more PIDs are specified reports
897 the unit names the processes referenced by the PIDs belong to.
899 * The system and service credential logic has been improved: there's
900 now a clearly defined place where system provisioning tools running
901 in the initrd can place credentials that will be imported into the
902 system's set of credentials during the initrd → host transition: the
903 /run/credentials/@initrd/ directory. Once the credentials placed
904 there are imported into the system credential set they are deleted
905 from this directory, and the directory itself is deleted afterwards
908 * A new kernel command line option systemd.set_credential_binary= has
909 been added, that is similar to the pre-existing
910 systemd.set_credential= but accepts arbitrary binary credential data,
911 encoded in Base64. Note that the kernel command line is not a
912 recommend way to transfer credentials into a system, since it is
913 world-readable from userspace.
915 * The default machine ID to use may now be configured via the
916 system.machine_id system credential. It will only be used if no
917 machine ID was set yet on the host.
919 * On Linux kernel 6.4 and newer system and service credentials will now
920 be placed in a tmpfs instance that has the "noswap" mount option
921 set. Previously, a "ramfs" instance was used. By switching to tmpfs
922 ACL support and overall size limits can now be enforced, without
923 compromising on security, as the memory is never paged out either
926 * The service manager now can detect when it is running in a
927 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
928 accepted by ConditionSecurity= for units that want to conditionalize
929 themselves on this. systemd-detect-virt gained new 'cvm' and
930 '--list-cvm' switches to respectively perform the detection or list
931 all known flavours of confidential VM, depending on the vendor. The
932 manager will publish a 'ConfidentialVirtualization' D-Bus property,
933 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
934 variable for unit generators. Finally, udev rules can match on a new
935 'cvm' key that will be set when in a confidential VM.
936 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
937 strings and QEMU's fw_cfg protocol will not be used to import
938 credentials and kernel command line parameters by the system manager,
939 systemd-boot and systemd-stub, because the hypervisor is considered
940 untrusted in this particular setting.
944 * The sd-journal API gained a new call sd_journal_get_seqnum() to
945 retrieve the current log record's sequence number and sequence number
946 ID, which allows applications to order records the same way as
947 journal does internally. The sequence number is now also exported in
948 the JSON and "export" output of the journal.
950 * journalctl gained a new switch --truncate-newline. If specified
951 multi-line log records will be truncated at the first newline,
952 i.e. only the first line of each log message will be shown.
954 * systemd-journal-upload gained support for --namespace=, similar to
955 the switch of the same name of journalctl.
959 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
960 may be used to exclude certain files from the effect of CopyFiles=.
962 * systemd-repart's Verity support now implements the Minimize= setting
963 to minimize the size of the resulting partition.
965 * systemd-repart gained a new --offline= switch, which may be used to
966 control whether images shall be built "online" or "offline",
967 i.e. whether to make use of kernel facilities such as loopback block
968 devices and device mapper or not.
970 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
971 partition with some files, it will now default to VFAT rather than
974 * systemd-repart gained a new --architecture= switch. If specified, the
975 per-architecture GPT partition types (i.e. the root and /usr/
976 partitions) configured in the partition drop-in files are
977 automatically adjusted to match the specified CPU architecture, in
978 order to simplify cross-architecture DDI building.
980 * systemd-repart will now default to a minimum size of 300MB for XFS
981 filesystems if no size parameter is specified. This matches what the
982 XFS tools (xfsprogs) can support.
984 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
986 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
987 Instead, pyelftools is now needed, and it will be used to perform the
988 ELF -> PE relocations at build time.
990 * bootctl gained a new switch --print-root-device/-R that prints the
991 block device the root file system is backed by. If specified twice,
992 it returns the whole disk block device (as opposed to partition block
993 device) the root file system is on. It's useful for invocations such
994 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
997 * systemd-stub will now look for the SMBIOS Type 1 field
998 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
999 kernel command line it invokes. This is useful for VMMs such as qemu
1000 to pass additional kernel command lines into the system even when
1001 booting via full UEFI. The contents of the field are measured into
1004 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
1005 value "auto". With this value, a kernel will be automatically
1006 analyzed, and if it qualifies as UKI, it will be installed as if the
1007 setting was to set to "uki", otherwise as "bls".
1009 * systemd-stub can now optionally load UEFI PE "add-on" images that may
1010 contain additional kernel command line information. These "add-ons"
1011 superficially look like a regular UEFI executable, and are expected
1012 to be signed via SecureBoot/shim. However, they do not actually
1013 contain code, but instead a subset of the PE sections that UKIs
1014 support. They are supposed to provide a way to extend UKIs with
1015 additional resources in a secure and authenticated way. Currently,
1016 only the .cmdline PE section may be used in add-ons, in which case
1017 any specified string is appended to the command line embedded into
1018 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
1019 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
1020 the future we expect other sections to be made extensible like this as
1023 * ukify has been updated to allow building these UEFI PE "add-on"
1024 images, using the new 'addon<EFI-ARCH>.efi.stub'.
1026 * ukify now accepts SBAT information to place in the .sbat PE section
1027 of UKIs and addons. If a UKI is built the SBAT information from the
1028 inner kernel is merged with any SBAT information associated with
1029 systemd-stub and the SBAT data specified on the ukify command line.
1031 * The kernel-install script has been rewritten in C, and reuses much of
1032 the infrastructure of existing tools such as bootctl. It also gained
1033 --esp-path= and --boot-path= options to override the path to the ESP,
1034 and the $BOOT partition. Options --make-entry-directory= and
1035 --entry-token= have been added as well, similar to bootctl's options
1038 * A new kernel-install plugin 60-ukify has been added which will
1039 combine kernel/initrd locally into a UKI and optionally sign them
1040 with a local key. This may be used to switch to UKI mode even on
1041 systems where a local kernel or initrd is used. (Typically UKIs are
1042 built and signed by the vendor.)
1044 * The ukify tool now supports "pesign" in addition to the pre-existing
1045 "sbsign" for signing UKIs.
1047 * systemd-measure and systemd-stub now look for the .uname PE section
1048 that should contain the kernel's "uname -r" string.
1050 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
1051 "offline", i.e. without access to a TPM (physical or
1054 Memory Pressure & Control:
1056 * The sd-event API gained new calls sd_event_add_memory_pressure(),
1057 sd_event_source_set_memory_pressure_type(),
1058 sd_event_source_set_memory_pressure_period() to create and configure
1059 an event source that is called whenever the OS signals memory
1060 pressure. Another call sd_event_trim_memory() is provided that
1061 compacts the process' memory use by releasing allocated but unused
1062 malloc() memory back to the kernel. Services can also provide their
1063 own custom callback to do memory trimming. This should improve system
1064 behaviour under memory pressure, as on Linux traditionally provided
1065 no mechanism to return process memory back to the kernel if the
1066 kernel was under memory pressure. This makes use of the kernel's PSI
1067 interface. Most long-running services in systemd have been hooked up
1068 with this, and in particular systems with low memory should benefit
1071 * Service units gained new settings MemoryPressureWatch= and
1072 MemoryPressureThresholdSec= to configure the PSI memory pressure
1073 logic individually. If these options are used, the
1074 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
1075 variables will be set for the invoked processes to inform them about
1076 the requested memory pressure behaviour. (This is used by the
1077 aforementioned sd-events API additions, if set.)
1079 * systemd-analyze gained a new "malloc" verb that shows the output
1080 generated by glibc's malloc_info() on services that support it. Right
1081 now, only the service manager has been updated accordingly. This
1082 call requires privileges.
1084 User & Session Management:
1086 * The sd-login API gained a new call sd_session_get_username() to
1087 return the user name of the owner of a login session. It also gained
1088 a new call sd_session_get_start_time() to retrieve the time the login
1089 session started. A new call sd_session_get_leader() has been added to
1090 return the PID of the "leader" process of a session. A new call
1091 sd_uid_get_login_time() returns the time since the specified user has
1092 most recently been continuously logged in with at least one session.
1094 * JSON user records gained a new set of fields capabilityAmbientSet and
1095 capabilityBoundingSet which contain a list of POSIX capabilities to
1096 set for the logged in users in the ambient and bounding sets,
1097 respectively. homectl gained the ability to configure these two sets
1098 for users via --capability-bounding-set=/--capability-ambient-set=.
1100 * pam_systemd learnt two new module options
1101 default-capability-bounding-set= and default-capability-ambient-set=,
1102 which configure the default bounding sets for users as they are
1103 logging in, if the JSON user record doesn't specify this explicitly
1104 (see above). The built-in default for the ambient set now contains
1105 the CAP_WAKE_ALARM, thus allowing regular users who may log in
1106 locally to resume from a system suspend via a timer.
1108 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
1109 call to update the TTY of a session after it has been allocated. This
1110 is useful for SSH sessions which are typically allocated first, and
1111 for which a TTY is added later.
1113 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
1114 combines the various other sd_pid_notify() flavours into one: takes a
1115 format string, an overriding PID, and a set of file descriptors to
1116 send. It also gained a new call sd_pid_notify_barrier() call which is
1117 equivalent to sd_notify_barrier() but allows the originating PID to
1120 * "loginctl list-users" and "loginctl list-sessions" will now show the
1121 state of each logged in user/session in their tabular output. It will
1122 also show the current idle state of sessions.
1126 * systemd-dissect will now show the intended CPU architecture of an
1129 * systemd-dissect will now install itself as mount helper for the "ddi"
1130 pseudo-file system type. This means you may now mount DDIs directly
1131 via /bin/mount or /etc/fstab, making full use of embedded Verity
1132 information and all other DDI features.
1134 Example: mount -t ddi myimage.raw /some/where
1136 * The systemd-dissect tool gained the new switches --attach/--detach to
1137 attach/detach a DDI to a loopback block device without mounting it.
1138 It will automatically derive the right sector size from the image
1139 and set up Verity and similar, but not mount the file systems in it.
1141 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
1142 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
1143 implied. Given that these file systems are typically untrusted, this
1144 should make mounting them automatically have less of a security
1147 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
1148 systemd-tmpfiles, …) now understand a new switch --image-policy= which
1149 takes a string encoding image dissection policy. With this mechanism
1150 automatic discovery and use of specific partition types and the
1151 cryptographic requirements on the partitions (Verity, LUKS, …) can be
1152 restricted, permitting better control of the exposed attack surfaces
1153 when mounting disk images. systemd-gpt-auto-generator will honour such
1154 an image policy too, configurable via the systemd.image_policy= kernel
1155 command line option. Unit files gained the RootImagePolicy=,
1156 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
1157 disk images a service runs off.
1159 * systemd-analyze gained a new verb "image-policy" to validate and
1160 parse image policy strings.
1162 * systemd-dissect gained support for a new --validate switch to
1163 superficially validate DDI structure, and check whether a specific
1164 image policy allows the DDI.
1166 * systemd-dissect gained support for a new --mtree-hash switch to
1167 optionally disable calculating mtree hashes, which can be slow on
1170 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
1171 are now able to operate on directories too, other than images.
1175 * networkd's GENEVE support as gained a new .network option
1176 InheritInnerProtocol=.
1178 * The [Tunnel] section in .netdev files has gained a new setting
1179 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
1181 * A new global IPv6PrivacyExtensions= setting has been added that
1182 selects the default value of the per-network setting of the same
1185 * The predictable network interface naming logic was extended to
1186 include SR-IOV-R "representor" information in network interface
1187 names. Unfortunately, this feature was not enabled by default and can
1188 only be enabled at compilation time by setting
1189 -Ddefault-net-naming-scheme=v254.
1191 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
1192 the RFC8910 captive portal option.
1196 * udevadm gained the new "verify" verb for validating udev rules files
1199 * udev gained a new tool "iocost" that can be used to configure QoS IO
1200 cost data based on hwdb information onto suitable block devices. Also
1201 see https://github.com/iocost-benchmark/iocost-benchmarks.
1203 TPM2 Support + Disk Encryption & Authentication:
1205 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
1206 ("Storage Root Key") as first step in the TPM2, and then use that
1207 for binding FDE to, if TPM2 support is used. This matches
1208 recommendations of TCG (see
1209 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
1211 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
1212 understand textual identifiers for these PCRs.
1214 * systemd-veritysetup + /etc/veritytab gained support for a series of
1215 new options: hash-offset=, superblock=, format=, data-block-size=,
1216 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
1217 fec-offset=, fec-roots= to configure various aspects of a Verity
1220 * systemd-cryptsetup + /etc/crypttab gained support for a new
1221 veracrypt-pim= option for setting the Personal Iteration Multiplier
1222 of veracrypt volumes.
1224 * systemd-integritysetup + /etc/integritytab gained support for a new
1225 mode= setting for controlling the dm-integrity mode (journal, bitmap,
1226 direct) for the volume.
1228 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
1229 registers, their symbolic names and current values.
1233 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
1234 access right is specified this is equivalent to "x" but only if the
1235 inode in question already has the executable bit set for at least
1236 some user/group. Otherwise the "x" bit will be turned off.
1238 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
1239 with C+ will result in a "merge" copy, i.e. all files of the source
1240 tree are copied into the target tree, even if that tree already
1241 exists, resulting in a combined tree of files already present in the
1242 target tree and those copied in.
1244 * systemd-tmpfiles gained a new --graceful switch. If specified lines
1245 with unknown users/groups will silently be skipped.
1249 * systemd-notify gained two new options --fd= and --fdname= for sending
1250 arbitrary file descriptors to the service manager (while specifying an
1251 explicit name for it).
1253 * systemd-notify gained a new --exec switch, which makes it execute the
1254 specified command line after sending the requested messages. This is
1255 useful for sending out READY=1 first, and then continuing invocation
1256 without changing process ID, so that the tool can be nicely used
1257 within an ExecStart= line of a unit file that uses Type=notify.
1259 sd-event + sd-bus APIs:
1261 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
1262 which may be used to explicitly end a rate-limit state an event
1263 source might be in, resetting all rate limiting counters.
1265 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
1266 sockets, it will now encode the "description" set via
1267 sd_bus_set_description() into the source socket address. It will also
1268 look for this information when accepting a connection. This is useful
1269 to track individual D-Bus connections on a D-Bus broker for debug
1274 * systemd-resolved gained a new resolved.conf setting
1275 StateRetentionSec= which may be used to retain cached DNS records
1276 even after their nominal TTL, and use them in case upstream DNS
1277 servers cannot be reached. This can be used to make name resolution
1278 more resilient in case of network problems.
1280 * resolvectl gained a new verb "show-cache" to show the current cache
1281 contents of systemd-resolved. This verb communicates with the
1282 systemd-resolved daemon and requires privileges.
1286 * Meson >= 0.60.0 is now required to build systemd.
1288 * The default keymap to apply may now be chosen at build-time via the
1289 new -Ddefault-keymap= meson option.
1291 * Most of systemd's long-running services now have a generic handler of
1292 the SIGRTMIN+18 signal handler which executes various operations
1293 depending on the sigqueue() parameter sent along. For example, values
1294 0x100…0x107 allow changing the maximum log level of such
1295 services. 0x200…0x203 allow changing the log target of such
1296 services. 0x300 make the services trim their memory similarly to the
1297 automatic PSI-triggered action, see above. 0x301 make the services
1298 output their malloc_info() data to the logs.
1300 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
1301 files, inspired by systemctl's verbs of the same name which edit unit
1302 files. Similarly, networkctl gained the same verbs for editing
1303 .network, .netdev, .link files.
1305 * A new syscall filter group "@sandbox" has been added that contains
1306 syscalls for sandboxing system calls such as those for seccomp and
1309 * New documentation has been added:
1311 https://systemd.io/COREDUMP
1312 https://systemd.io/MEMORY_PRESSURE
1315 * systemd-firstboot gained a new --reset option. If specified, the
1316 settings in /etc/ it knows how to initialize are reset.
1318 * systemd-sysext is now a multi-call binary and is also installed under
1319 the systemd-confext alias name (via a symlink). When invoked that way
1320 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
1321 powerful, atomic, secure configuration management of sorts, that
1322 locally can merge configuration from multiple confext configuration
1323 images into a single immutable tree.
1325 * The --network-macvlan=, --network-ipvlan=, --network-interface=
1326 switches of systemd-nspawn may now optionally take the intended
1327 network interface inside the container.
1329 * All our programs will now send an sd_notify() message with their exit
1330 status in the EXIT_STATUS= field when exiting, using the usual
1331 protocol, including PID 1. This is useful for VMMs and container
1332 managers to collect an exit status from a system as it shuts down, as
1333 set via "systemctl exit …". This is particularly useful in test cases
1334 and similar, as invocations via a VM can now nicely propagate an exit
1335 status to the host, similar to local processes.
1337 * systemd-run gained a new switch --expand-environment=no to disable
1338 server-side environment variable expansion in specified command
1339 lines. Expansion defaults to enabled for all execution types except
1340 --scope, where it defaults to off (and prints a warning) for backward
1341 compatibility reasons. --scope will be flipped to enabled by default
1342 too in a future release. If you are using --scope and passing a '$'
1343 character in the payload you should start explicitly using
1344 --expand-environment=yes/no according to the use case.
1346 * The systemd-system-update-generator has been updated to also look for
1347 the special flag file /etc/system-update in addition to the existing
1348 support for /system-update to decide whether to enter system update
1351 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
1352 mount options by default.
1354 * systemd-fstab-generator now understands two new kernel command line
1355 options systemd.mount-extra= and systemd.swap-extra=, which configure
1356 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
1357 will be ran on these block devices, like it already happens for
1358 'root='. It also now supports the new fstab.extra and
1359 fstab.extra.initrd credentials that may contain additional /etc/fstab
1360 lines to apply at boot.
1362 * systemd-getty-generator now understands two new credentials
1363 getty.ttys.container and getty.ttys.serial. These credentials may
1364 contain a list of TTY devices – one per line – to instantiate
1365 container-getty@.service and serial-getty@.service on.
1367 * The getty/serial-getty/container-getty units now import the 'agetty.*'
1368 and 'login.*' credentials, which are consumed by the 'login' and
1369 'agetty' programs starting from util-linux v2.40.
1371 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
1372 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
1373 which case the Path= setting is taken relative to the ESP or XBOOTLDR
1374 partitions, rather than the system's root directory /. The relevant
1375 directories are automatically discovered.
1377 * The systemd-ac-power tool gained a new switch --low, which reports
1378 whether the battery charge is considered "low", similar to how the
1379 s2h suspend logic checks this state to decide whether to enter system
1380 suspend or hibernation.
1382 * The /etc/os-release file can now have two new optional fields
1383 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
1386 * When the system hibernates, information about the device and offset
1387 used is now written to a non-volatile EFI variable. On next boot the
1388 system will attempt to resume from the location indicated in this EFI
1389 variable. This should make hibernation a lot more robust, while
1390 requiring no manual configuration of the resume location.
1392 * The $XDG_STATE_HOME environment variable (added in more recent
1393 versions of the XDG basedir specification) is now honoured to
1394 implement the StateDirectory= setting in user services.
1396 * A new component "systemd-battery-check" has been added. It may run
1397 during early boot (usually in the initrd), and checks the battery
1398 charge level of the system. In case the charge level is very low the
1399 user is notified (graphically via Plymouth – if available – as well
1400 as in text form on the console), and the system is turned off after a
1401 10s delay. The feature can be disabled by passing
1402 systemd.battery_check=0 through the kernel command line.
1404 * The 'passwdqc' library is now supported as an alternative to the
1405 'pwquality' library and can be selected at build time.
1407 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
1408 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
1409 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
1410 Arian van Putten, Arthur Shau, A S Alam,
1411 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
1412 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
1413 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
1414 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
1415 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
1416 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
1417 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
1418 David Schroeder, David Tardon, dependabot[bot],
1419 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
1420 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
1421 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
1422 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
1423 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
1424 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
1425 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
1426 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
1427 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
1428 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
1429 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
1430 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
1431 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
1432 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
1433 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
1434 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
1435 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
1436 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
1437 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
1438 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
1439 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
1440 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
1441 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
1442 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
1443 saikat0511, Samanta Navarro, Sam James, Sam Morris,
1444 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
1445 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
1446 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
1447 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
1448 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
1449 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
1450 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
1451 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
1452 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
1453 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
1454 Дамјан Георгиевски, наб
1456 — Edinburgh, 2023-07-28
1460 Announcements of Future Feature Removals and Incompatible Changes:
1462 * We intend to remove cgroup v1 support from systemd release after the
1463 end of 2023. If you run services that make explicit use of cgroup v1
1464 features (i.e. the "legacy hierarchy" with separate hierarchies for
1465 each controller), please implement compatibility with cgroup v2 (i.e.
1466 the "unified hierarchy") sooner rather than later. Most of Linux
1467 userspace has been ported over already.
1469 * We intend to remove support for split-usr (/usr mounted separately
1470 during boot) and unmerged-usr (parallel directories /bin and
1471 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
1472 half of 2023, in the first release that falls into that time window.
1473 For more details, see:
1474 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1476 * We intend to change behaviour w.r.t. units of the per-user service
1477 manager and sandboxing options, so that they work without having to
1478 manually enable PrivateUsers= as well, which is not required for
1479 system units. To make this work, we will implicitly enable user
1480 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
1481 user unit. The drawback is that system users will no longer be visible
1482 (and appear as 'nobody') to the user unit when a sandboxing option is
1483 enabled. By definition a sandboxed user unit should run with reduced
1484 privileges, so impact should be small. This will remove a great source
1485 of confusion that has been reported by users over the years, due to
1486 how these options require an extra setting to be manually enabled when
1487 used in the per-user service manager, as opposed as to the system
1488 service manager. We plan to enable this change in the next release
1489 later this year. For more details, see:
1490 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
1492 Deprecations and incompatible changes:
1494 * systemctl will now warn when invoked without /proc/ mounted
1495 (e.g. when invoked after chroot() into an directory tree without the
1496 API mount points like /proc/ being set up.) Operation in such an
1497 environment is not fully supported.
1499 * The return value of 'systemctl is-active|is-enabled|is-failed' for
1500 unknown units is changed: previously 1 or 3 were returned, but now 4
1501 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
1503 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
1504 systemd-hwdb (added in 2014) should be used instead.
1506 * 'bootctl --json' now outputs a single JSON array, instead of a stream
1507 of newline-separated JSON objects.
1509 * Udev rules in 60-evdev.rules have been changed to load hwdb
1510 properties for all modalias patterns. Previously only the first
1511 matching pattern was used. This could change what properties are
1512 assigned if the user has more and less specific patterns that could
1513 match the same device, but it is expected that the change will have
1514 no effect for most users.
1516 * systemd-networkd-wait-online exits successfully when all interfaces
1517 are ready or unmanaged. Previously, if neither '--any' nor
1518 '--interface=' options were used, at least one interface had to be in
1519 configured state. This change allows the case where systemd-networkd
1520 is enabled, but no interfaces are configured, to be handled
1521 gracefully. It may occur in particular when a different network
1522 manager is also enabled and used.
1524 * Some compatibility helpers were dropped: EmergencyAction= in the user
1525 manager, as well as measuring kernel command line into PCR 8 in
1526 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
1529 * The '-Dupdate-helper-user-timeout=' build-time option has been
1530 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
1531 integer as parameter instead of a string.
1533 * The DDI image dissection logic (which backs RootImage= in service
1534 unit files, the --image= switch in various tools such as
1535 systemd-nspawn, as well as systemd-dissect) will now only mount file
1536 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
1537 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
1538 variable. These file systems are fairly well supported and maintained
1539 in current kernels, while others are usually more niche, exotic or
1540 legacy and thus typically do not receive the same level of security
1543 * The default per-link multicast DNS mode is changed to "yes"
1544 (that was previously "no"). As the default global multicast DNS mode
1545 has been "yes" (but can be changed by the build option), now the
1546 multicast DNS is enabled on all links by default. You can disable the
1547 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
1548 or on an interface by calling "resolvectl mdns INTERFACE no".
1552 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
1553 (UKIs) has been added. This replaces functionality provided by
1554 'dracut --uefi' and extends it with automatic calculation of PE file
1555 offsets, insertion of signed PCR policies generated by
1556 systemd-measure, support for initrd concatenation, signing of the
1557 embedded Linux image and the combined image with sbsign, and
1558 heuristics to autodetect the kernel uname and verify the splash
1561 Changes in systemd and units:
1563 * A new service type Type=notify-reload is defined. When such a unit is
1564 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
1565 service process. The manager will then wait until it receives a
1566 "RELOADING=1" followed by a "READY=1" notification from the unit as
1567 response (via sd_notify()). Otherwise, this type is the same as
1568 Type=notify. A new setting ReloadSignal= may be used to change the
1569 signal to send from the default of SIGHUP.
1571 user@.service, systemd-networkd.service, systemd-udevd.service, and
1572 systemd-logind have been updated to this type.
1574 * Initrd environments which are not on a pure memory file system (e.g.
1575 overlayfs combination as opposed to tmpfs) are now supported. With
1576 this change, during the initrd → host transition ("switch root")
1577 systemd will erase all files of the initrd only when the initrd is
1578 backed by a memory file system such as tmpfs.
1580 * New per-unit MemoryZSwapMax= option has been added to configure
1581 memory.zswap.max cgroup properties (the maximum amount of zswap
1584 * A new LogFilterPatterns= option has been added for units. It may be
1585 used to specify accept/deny regular expressions for log messages
1586 generated by the unit, that shall be enforced by systemd-journald.
1587 Rejected messages are neither stored in the journal nor forwarded.
1588 This option may be used to suppress noisy or uninteresting messages
1591 * The manager has a new
1592 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
1593 query process ownership via a PIDFD, which is more resilient against
1594 PID recycling issues.
1596 * Scope units now support OOMPolicy=. Login session scopes default to
1597 OOMPolicy=continue, allowing login scopes to survive the OOM killer
1598 terminating some processes in the scope.
1600 * systemd-fstab-generator now supports x-systemd.makefs option for
1601 /sysroot/ (in the initrd).
1603 * The maximum rate at which daemon reloads are executed can now be
1604 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
1605 options. (Or the equivalent on the kernel command line:
1606 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
1607 addition, systemd now logs the originating unit and PID when a reload
1608 request is received over D-Bus.
1610 * When enabling a swap device systemd will now reinitialize the device
1611 when the page size of the swap space does not match the page size of
1612 the running kernel. Note that this requires the 'swapon' utility to
1613 provide the '--fixpgsz' option, as implemented by util-linux, and it
1614 is not supported by busybox at the time of writing.
1616 * systemd now executes generator programs in a mount namespace
1617 "sandbox" with most of the file system read-only and write access
1618 restricted to the output directories, and with a temporary /tmp/
1619 mount provided. This provides a safeguard against programming errors
1620 in the generators, but also fixes here-docs in shells, which
1621 previously didn't work in early boot when /tmp/ wasn't available
1622 yet. (This feature has no security implications, because the code is
1623 still privileged and can trivially exit the sandbox.)
1625 * The system manager will now parse a new "vmm.notify_socket"
1626 system credential, which may be supplied to a VM via SMBIOS. If
1627 found, the manager will send a "READY=1" notification on the
1628 specified socket after boot is complete. This allows readiness
1629 notification to be sent from a VM guest to the VM host over a VSOCK
1632 * The sample PAM configuration file for systemd-user@.service now
1633 includes a call to pam_namespace. This puts children of user@.service
1634 in the expected namespace. (Many distributions replace their file
1635 with something custom, so this change has limited effect.)
1637 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
1638 can be used to override the mount units burst late limit for
1639 parsing '/proc/self/mountinfo', which was introduced in v249.
1642 * Drop-ins for init.scope changing control group resource limits are
1643 now applied, while they were previously ignored.
1645 * New build-time configuration options '-Ddefault-timeout-sec=' and
1646 '-Ddefault-user-timeout-sec=' have been added, to let distributions
1647 choose the default timeout for starting/stopping/aborting system and
1648 user units respectively.
1650 * Service units gained a new setting OpenFile= which may be used to
1651 open arbitrary files in the file system (or connect to arbitrary
1652 AF_UNIX sockets in the file system), and pass the open file
1653 descriptor to the invoked process via the usual file descriptor
1654 passing protocol. This is useful to give unprivileged services access
1655 to select files which have restrictive access modes that would
1656 normally not allow this. It's also useful in case RootDirectory= or
1657 RootImage= is used to allow access to files from the host environment
1658 (which is after all not visible from the service if these two options
1663 * The new net naming scheme "v253" has been introduced. In the new
1664 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
1665 a PCI bus. This extends the coverage of predictable interface names
1666 in some embedded systems.
1668 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
1669 a more informative path on some embedded systems.
1671 * Partition block devices will now also get symlinks in
1672 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
1673 block device nodes via the kernel's "diskseq" value. Previously those
1674 symlinks were only created for the main block device.
1676 * A new operator '-=' is supported for SYMLINK variables. This allows
1677 symlinks to be unconfigured even if an earlier rule added them.
1679 * 'udevadm --trigger --settle' now also works for network devices
1680 that are being renamed.
1682 Changes in sd-boot, bootctl, and the Boot Loader Specification:
1684 * systemd-boot now passes its random seed directly to the kernel's RNG
1685 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
1686 means the RNG gets seeded very early in boot before userspace has
1689 * systemd-boot will pass a disk-backed random seed – even when secure
1690 boot is enabled – if it can additionally get a random seed from EFI
1691 itself (via EFI's RNG protocol), or a prior seed in
1692 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
1694 * systemd-boot-system-token.service was renamed to
1695 systemd-boot-random-seed.service and extended to always save a random
1696 seed to ESP on every boot when a compatible boot loader is used. This
1697 allows a refreshed random seed to be used in the boot loader.
1699 * systemd-boot handles various seed inputs using a domain- and
1700 field-separated hashing scheme.
1702 * systemd-boot's 'random-seed-mode' option has been removed. A system
1703 token is now always required to be present for random seeds to be
1706 * systemd-boot now supports being loaded from other locations than the
1707 ESP, for example for direct kernel boot under QEMU or when embedded
1710 * systemd-boot now parses SMBIOS information to detect
1711 virtualization. This information is used to skip some warnings which
1712 are not useful in a VM and to conditionalize other aspects of
1715 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
1716 Secure Boot automated certificate enrollment from the ESP only if it
1717 is considered 'safe' to do so. At the moment 'safe' means running in
1720 * systemd-stub now processes random seeds in the same way as
1721 systemd-boot already does, in case a unified kernel image is being
1722 used from a different bootloader than systemd-boot, or without any
1725 * bootctl will now generate a system token on all EFI systems, even
1726 virtualized ones, and is activated in the case that the system token
1727 is missing from either sd-boot and sd-stub booted systems.
1729 * bootctl now implements two new verbs: 'kernel-identify' prints the
1730 type of a kernel image file, and 'kernel-inspect' provides
1731 information about the embedded command line and kernel version of
1734 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
1735 as for kernel-install.
1737 * The JSON output of "bootctl list" will now contain two more fields:
1738 isDefault and isSelected are boolean fields set to true on the
1739 default and currently booted boot menu entries.
1741 * bootctl gained a new verb "unlink" for removing a boot loader entry
1742 type #1 file from disk in a safe and robust way.
1744 * bootctl also gained a new verb "cleanup" that automatically removes
1745 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
1746 is not referenced anymore by any installed Type #1 boot loader
1747 specification entry. This is particularly useful in environments where
1748 a large number of entries reference the same or partly the same
1749 resources (for example, for snapshot-based setups).
1751 Changes in kernel-install:
1753 * A new "installation layout" can be configured as layout=uki. With
1754 this setting, a Boot Loader Specification Type#1 entry will not be
1755 created. Instead, a new kernel-install plugin 90-uki-copy.install
1756 will copy any .efi files from the staging area into the boot
1757 partition. A plugin to generate the UKI .efi file must be provided
1760 Changes in systemctl:
1762 * 'systemctl reboot' has dropped support for accepting a positional
1763 argument as the argument to the reboot(2) syscall. Please use the
1764 --reboot-argument= option instead.
1766 * 'systemctl disable' will now warn when called on units without
1767 install information. A new --no-warn option has been added that
1768 silences this warning.
1770 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
1771 of the drop-in to edit. (Previously, 'override.conf' was always
1774 * 'systemctl list-dependencies' now respects --type= and --state=.
1776 * 'systemctl kexec' now supports XEN VMM environments.
1778 * 'systemctl edit' will now tell the invoked editor to jump into the
1779 first line with actual unit file data, skipping over synthesized
1782 Changes in systemd-networkd and related tools:
1784 * The [DHCPv4] section in .network file gained new SocketPriority=
1785 setting that assigns the Linux socket priority used by the DHCPv4 raw
1786 socket. This may be used in conjunction with the
1787 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
1788 desired ethernet 802.1Q frame priority for DHCPv4 initial
1789 packets. This cannot be achieved with netfilter mangle tables because
1790 of the raw socket bypass.
1792 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
1793 new QuickAck= boolean setting that enables the TCP quick ACK mode for
1794 the routes configured by the acquired DHCPv4 lease or received router
1795 advertisements (RAs).
1797 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
1798 routes) now accepts three values, for high, medium, and low preference
1799 of the router (which can be set with the RouterPreference=) setting.
1801 * systemd-networkd-wait-online now supports matching via alternative
1804 * The [DHCPv6] section in .network file gained new SendRelease=
1805 setting which enables the DHCPv6 client to send release when
1806 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
1807 It is enabled by default.
1809 * If the Address= setting in [Network] or [Address] sections in .network
1810 specified without its prefix length, then now systemd-networkd assumes
1811 /32 for IPv4 or /128 for IPv6 addresses.
1813 * networkctl shows network and link file dropins in status output.
1815 Changes in systemd-dissect:
1817 * systemd-dissect gained a new option --list, to print the paths of
1818 all files and directories in a DDI.
1820 * systemd-dissect gained a new option --mtree, to generate a file
1821 manifest compatible with BSD mtree(5) of a DDI
1823 * systemd-dissect gained a new option --with, to execute a command with
1824 the specified DDI temporarily mounted and used as working
1825 directory. This is for example useful to convert a DDI to "tar"
1826 simply by running it within a "systemd-dissect --with" invocation.
1828 * systemd-dissect gained a new option --discover, to search for
1829 Discoverable Disk Images (DDIs) in well-known directories of the
1830 system. This will list machine, portable service and system extension
1833 * systemd-dissect now understands 2nd stage initrd images stored as a
1834 Discoverable Disk Image (DDI).
1836 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
1837 disk UUID stored in the GPT header) among the other data it can show.
1839 * systemd-dissect gained a new --in-memory switch to operate on an
1840 in-memory copy of the specified DDI file. This is useful to access a
1841 DDI with write access without persisting any changes. It's also
1842 useful for accessing a DDI without keeping the originating file
1845 * The DDI dissection logic will now automatically detect the intended
1846 sector size of disk images stored in files, based on the GPT
1847 partition table arrangement. Loopback block devices for such DDIs
1848 will then be configured automatically for the right sector size. This
1849 is useful to make dealing with modern 4K sector size DDIs fully
1850 automatic. The systemd-dissect tool will now show the detected sector
1851 size among the other DDI information in its output.
1853 Changes in systemd-repart:
1855 * systemd-repart gained new options --include-partitions= and
1856 --exclude-partitions= to filter operation on partitions by type UUID.
1857 This allows systemd-repart to be used to build images in which the
1858 type of one partition is set based on the contents of another
1859 partition (for example when the boot partition shall include a verity
1860 hash of the root partition).
1862 * systemd-repart also gained a --defer-partitions= option that is
1863 similar to --exclude-partitions=, but the size of the partition is
1864 still taken into account when sizing partitions, but without
1867 * systemd-repart gained a new --sector-size= option to specify what
1868 sector size should be used when an image is created.
1870 * systemd-repart now supports generating erofs file systems via
1871 CopyFiles= (a read-only file system similar to squashfs).
1873 * The Minimize= option was extended to accept "best" (which means the
1874 most minimal image possible, but may require multiple attempts) and
1875 "guess" (which means a reasonably small image).
1877 * The systemd-growfs binary now comes with a regular unit file template
1878 systemd-growfs@.service which can be instantiated directly for any
1879 desired file system. (Previously, the unit was generated dynamically
1880 by various generators, but no regular unit file template was
1883 Changes in journal tools:
1885 * Various systemd tools will append extra fields to log messages when
1886 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
1887 this includes information about D-Bus messages when sd-bus is used,
1888 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
1889 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
1890 Details of what is logged and when are subject to change.
1892 * The systemd-journald-audit.socket can now be disabled via the usual
1893 "systemctl disable" mechanism to stop collection of audit
1894 messages. Please note that it is not enabled statically anymore and
1895 must be handled by the preset/enablement logic in package
1896 installation scripts.
1898 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
1899 be used to curtail disk use by systemd-journal-remote. This is
1900 similar to the options supported by systemd-journald.
1902 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
1905 * When enrolling new keys systemd-cryptenroll now supports unlocking
1906 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
1907 password was strictly required to be specified.
1909 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
1910 (except for tokens with user verification, UV) to identify tokens
1911 before authentication. Multiple FIDO2 tokens can now be enrolled at
1912 the same time, and systemd-cryptsetup will automatically select one
1913 that corresponds to one of the available LUKS key slots.
1915 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
1916 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
1917 bank and number into which the volume key should be measured. This is
1918 automatically enabled for the encrypted root volume discovered and
1919 activated by systemd-gpt-auto-generator.
1921 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
1922 "noexec,nosuid,nodev".
1924 * systemd-gpt-auto-generator will now honour the rootfstype= and
1925 rootflags= kernel command line switches for root file systems it
1926 discovers, to match behaviour in case an explicit root fs is
1927 specified via root=.
1929 * systemd-pcrphase gained new options --machine-id and --file-system=
1930 to measure the machine-id and mount point information into PCR 15.
1931 New service unit files systemd-pcrmachine.service and
1932 systemd-pcrfs@.service have been added that invoke the tool with
1933 these switches during early boot.
1935 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
1936 with a success exit code even if no TPM device is detected.
1938 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
1939 making it harder to brute-force.
1941 Changes in other tools:
1943 * systemd-homed gained support for luksPbkdfForceIterations (the
1944 intended number of iterations for the PBKDF operation on LUKS).
1946 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
1947 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
1948 may now be used to specify additional arguments for mkfs when
1949 systemd-homed formats a file system.
1951 * systemd-hostnamed now exports the contents of
1952 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
1953 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
1954 unprivileged code to access those values.
1956 systemd-hostnamed also exports the SUPPORT_END= field from
1957 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
1958 this to show the status of the installed system.
1960 * systemd-measure gained an --append= option to sign multiple phase
1961 paths with different signing keys. This allows secrets to be
1962 accessible only in certain parts of the boot sequence. Note that
1963 'ukify' provides similar functionality in a more accessible form.
1965 * systemd-timesyncd will now write a structured log message with
1966 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
1967 on a on-disk timestamp, similarly to what it did when reaching
1968 synchronization via NTP.
1970 * systemd-timesyncd will now update the on-disk timestamp file on each
1971 boot at least once, making it more likely that the system time
1972 increases in subsequent boots.
1974 * systemd-vconsole-setup gained support for system/service credentials:
1975 vconsole.keymap/vconsole.keymap_toggle and
1976 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
1977 the similarly-named options in vconsole.conf.
1979 * systemd-localed will now save the XKB keyboard configuration to
1980 /etc/vconsole.conf, and also read it from there with a higher
1981 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
1982 file. Previously, this information was stored in the former file in
1983 converted form, and only in latter file in the original form. Tools
1984 which want to access keyboard configuration can now do so from a
1987 * systemd-resolved gained support for configuring the nameservers and
1988 search domains via kernel command line (nameserver=, domain=) and
1989 credentials (network.dns, network.search_domains).
1991 * systemd-resolved will now synthesize host names for the DNS stub
1992 addresses it supports. Specifically when "_localdnsstub" is resolved,
1993 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
1994 127.0.0.54 is returned.
1996 * systemd-notify will now send a "RELOADING=1" notification when called
1997 with --reloading, and "STOPPING=1" when called with --stopping. This
1998 can be used to implement notifications from units where it's easier
1999 to call a program than to use the sd-daemon library.
2001 * systemd-analyze's 'plot' command can now output its information in
2002 JSON, controlled via the --json= switch. Also, new --table, and
2003 --no-legend options have been added.
2005 * 'machinectl enable' will now automatically enable machines.target
2006 unit in addition to adding the machine unit to the target.
2008 Similarly, 'machinectl start|stop' gained a --now option to enable or
2009 disable the machine unit when starting or stopping it.
2011 * systemd-sysusers will now create /etc/ if it is missing.
2013 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
2014 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
2015 added to provide the new initial value for the new automated battery
2016 estimation functionality. If 'HibernateDelaySec=' is set to any value,
2017 the automated estimate (and thus the automated hibernation on low
2018 battery to avoid data loss) functionality will be disabled.
2020 * Default tmpfiles.d/ configuration will now automatically create
2021 credentials storage directory '/etc/credstore/' with the appropriate,
2022 secure permissions. If '/run/credstore/' exists, its permissions will
2023 be fixed too in case they are not correct.
2025 Changes in libsystemd and shared code:
2027 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
2028 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
2030 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
2031 128-bit ID in files such as /etc/machine-id has an invalid
2032 format. They also accept NULL as output parameter in more places,
2033 which is useful when the caller only wants to validate the inputs and
2034 does not need the output value.
2036 * sd-login gained new functions sd_pidfd_get_session(),
2037 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
2038 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
2039 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
2040 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
2041 but accept a PIDFD instead of a PID.
2043 * sd-path (and systemd-path) now export four new paths:
2044 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
2045 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
2046 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
2047 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
2049 * sd_notify() now supports AF_VSOCK as transport for notification
2050 messages (in addition to the existing AF_UNIX support). This is
2051 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
2053 * Detection of chroot() environments now works if /proc/ is not
2054 mounted. This affects systemd-detect-virt --chroot, but also means
2055 that systemd tools will silently skip various operations in such an
2058 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
2059 virtualization is now detected.
2061 Changes in the build system:
2063 * Standalone variants of systemd-repart and systemd-shutdown may now be
2064 built (if -Dstandalone=true).
2066 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
2067 example, allow scripts to conditionalize execution on AC power
2070 * The libp11kit library is now loaded through dlopen(3).
2072 Changes in the documentation:
2074 * Specifications that are not closely tied to systemd have moved to
2075 https://uapi-group.org/specifications/: the Boot Loader Specification
2076 and the Discoverable Partitions Specification.
2078 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
2079 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
2080 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
2081 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
2082 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
2083 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
2084 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
2085 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
2086 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
2087 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
2088 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
2089 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
2090 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
2091 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
2092 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
2093 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
2094 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
2095 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
2096 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
2097 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
2098 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
2099 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
2100 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
2101 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
2102 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
2103 reuben olinsky, Richard E. van der Luit, Richard Phibel,
2104 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
2105 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
2106 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
2107 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
2108 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
2109 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
2110 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
2111 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
2114 — Warsaw, 2023-02-15
2118 Announcements of Future Feature Removals:
2120 * We intend to remove cgroup v1 support from systemd release after the
2121 end of 2023. If you run services that make explicit use of cgroup v1
2122 features (i.e. the "legacy hierarchy" with separate hierarchies for
2123 each controller), please implement compatibility with cgroup v2 (i.e.
2124 the "unified hierarchy") sooner rather than later. Most of Linux
2125 userspace has been ported over already.
2127 * We intend to remove support for split-usr (/usr mounted separately
2128 during boot) and unmerged-usr (parallel directories /bin and
2129 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
2130 half of 2023, in the first release that falls into that time window.
2131 For more details, see:
2132 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
2134 Compatibility Breaks:
2136 * ConditionKernelVersion= checks that use the '=' or '!=' operators
2137 will now do simple string comparisons (instead of version comparisons
2138 à la stverscmp()). Version comparisons are still done for the
2139 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
2140 specified, a shell-style glob match is now done. This creates a minor
2141 incompatibility compared to older systemd versions when the '*', '?',
2142 '[', ']' characters are used, as these will now match as shell globs
2143 instead of literally. Given that kernel version strings typically do
2144 not include these characters we expect little breakage through this
2147 * The service manager will now read the SELinux label used for SELinux
2148 access checks from the unit file at the time it loads the file.
2149 Previously, the label would be read at the moment of the access
2150 check, which was problematic since at that time the unit file might
2151 already have been updated or removed.
2155 * systemd-measure is a new tool for calculating and signing expected
2156 TPM2 PCR values for a given unified kernel image (UKI) booted via
2157 sd-stub. The public key used for the signature and the signed
2158 expected PCR information can be embedded inside the UKI. This
2159 information can be extracted from the UKI by external tools and code
2160 in the image itself and is made available to userspace in the booted
2163 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
2164 updated to make use of this information if available in the booted
2165 kernel: when locking an encrypted volume/credential to the TPM
2166 systemd-cryptenroll/systemd-creds will use the public key to bind the
2167 volume/credential to any kernel that carries PCR information signed
2168 by the same key pair. When unlocking such volumes/credentials
2169 systemd-cryptsetup/systemd-creds will use the signature embedded in
2170 the booted UKI to gain access.
2172 Binding TPM-based disk encryption to public keys/signatures of PCR
2173 values — instead of literal PCR values — addresses the inherent
2174 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
2175 disks remain accessible even if the UKI is updated, without any TPM
2176 specific preparation during the OS update — as long as each UKI
2177 carries the necessary PCR signature information.
2179 Net effect: if you boot a properly prepared kernel, TPM-bound disk
2180 encryption now defaults to be locked to kernels which carry PCR
2181 signatures from the same key pair. Example: if a hypothetical distro
2182 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
2183 by default – bound to only FooOS kernels, and encrypted volumes bound
2184 to the TPM cannot be unlocked on kernels from other sources. (But do
2185 note this behaviour requires preparation/enabling in the UKI, and of
2186 course users can always enroll non-TPM ways to unlock the volume.)
2188 * systemd-pcrphase is a new tool that is invoked at six places during
2189 system runtime, and measures additional words into TPM2 PCR 11, to
2190 mark milestones of the boot process. This allows binding access to
2191 specific TPM2-encrypted secrets to specific phases of the boot
2192 process. (Example: LUKS2 disk encryption key only accessible in the
2193 initrd, but not later.)
2195 Changes in systemd itself, i.e. the manager and units
2197 * The cpu controller is delegated to user manager units by default, and
2198 CPUWeight= settings are applied to the top-level user slice units
2199 (app.slice, background.slice, session.slice). This provides a degree
2200 of resource isolation between different user services competing for
2203 * Systemd can optionally do a full preset in the "first boot" condition
2204 (instead of just enable-only). This behaviour is controlled by the
2205 compile-time option -Dfirst-boot-full-preset. Right now it defaults
2206 to 'false', but the plan is to switch it to 'true' for the subsequent
2209 * Drop-ins are now allowed for transient units too.
2211 * Systemd will set the taint flag 'support-ended' if it detects that
2212 the OS image is past its end-of-support date. This date is declared
2213 in a new /etc/os-release field SUPPORT_END= described below.
2215 * Two new settings ConditionCredential= and AssertCredential= can be
2216 used to skip or fail units if a certain system credential is not
2219 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
2221 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
2222 specify the SMACK security label to use when not specified in a unit
2225 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
2226 specify the default timeout when waiting for device units to
2229 * C.UTF-8 is used as the default locale if nothing else has been
2232 * [Condition|Assert]Firmware= have been extended to support certain
2233 SMBIOS fields. For example
2235 ConditionFirmware=smbios-field(board_name = "Custom Board")
2237 conditionalizes the unit to run only when
2238 /sys/class/dmi/id/board_name contains "Custom Board" (without the
2241 * ConditionFirstBoot= now correctly evaluates as true only during the
2242 boot phase of the first boot. A unit executed later, after booting
2243 has completed, will no longer evaluate this condition as true.
2245 * Socket units will now create sockets in the SELinuxContext= of the
2246 associated service unit, if any.
2248 * Boot phase transitions (start initrd → exit initrd → boot complete →
2249 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
2250 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
2251 unsealed only in the initrd.
2253 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
2254 also be provided to ExecStartPre= processes.
2256 * Various units are now correctly ordered against
2257 initrd-switch-root.target where previously a conflict without
2258 ordering was configured. A stop job for those units would be queued,
2259 but without the ordering it could be executed only after
2260 initrd-switch-root.service, leading to units not being restarted in
2261 the host system as expected.
2263 * In order to fully support the IPMI watchdog driver, which has not yet
2264 been ported to the new common watchdog device interface,
2265 /dev/watchdog0 will be tried first and systemd will silently fallback
2266 to /dev/watchdog if it is not found.
2268 * New watchdog-related D-Bus properties are now published by systemd:
2269 WatchdogDevice, WatchdogLastPingTimestamp,
2270 WatchdogLastPingTimestampMonotonic.
2272 * At shutdown, API virtual files systems (proc, sys, etc.) will be
2275 * At shutdown, systemd will now log about processes blocking unmounting
2278 * A new meson build option 'clock-valid-range-usec-max' was added to
2279 allow disabling system time correction if RTC returns a timestamp far
2282 * Propagated restart jobs will no longer be discarded while a unit is
2285 * PID 1 will now import system credentials from SMBIOS Type 11 fields
2286 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
2287 simple, fast and generic path for supplying credentials to a VM,
2288 without involving external tools such as cloud-init/ignition.
2290 * The CPUWeight= setting of unit files now accepts a new special value
2291 "idle", which configures "idle" level scheduling for the unit.
2293 * Service processes that are activated due to a .timer or .path unit
2294 triggering will now receive information about this via environment
2295 variables. Note that this is information is lossy, as activation
2296 might be coalesced and only one of the activating triggers will be
2297 reported. This is hence more suited for debugging or tracing rather
2298 than for behaviour decisions.
2300 * The riscv_flush_icache(2) system call has been added to the list of
2301 system calls allowed by default when SystemCallFilter= is used.
2303 * The selinux context derived from the target executable, instead of
2304 'init_t' used for the manager itself, is now used when creating
2305 listening sockets for units that specify SELinuxContextFromNet=yes.
2307 Changes in sd-boot, bootctl, and the Boot Loader Specification:
2309 * The Boot Loader Specification has been cleaned up and clarified.
2310 Various corner cases in version string comparisons have been fixed
2311 (e.g. comparisons for empty strings). Boot counting is now part of
2312 the main specification.
2314 * New PCRs measurements are performed during boot: PCR 11 for the
2315 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
2316 took place this is now reported to userspace via the new
2317 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
2319 * As before, systemd-stub will measure kernel parameters and system
2320 credentials into PCR 12. It will now report this fact via the
2321 StubPcrKernelParameters EFI variable to userspace.
2323 * The UEFI monotonic boot counter is now included in the updated random
2324 seed file maintained by sd-boot, providing some additional entropy.
2326 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
2327 of arranging the image manually and jumping to the kernel entry
2328 point. sd-stub also installs a temporary UEFI SecurityOverride to
2329 allow the (unsigned) nested image to be booted. This is safe because
2330 the outer (signed) stub+kernel binary must have been verified before
2331 the stub was executed.
2333 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
2334 is now supported by sd-boot.
2336 * bootctl gained a bunch of new options: --all-architectures to install
2337 binaries for all supported EFI architectures, --root= and --image=
2338 options to operate on a directory or disk image, and
2339 --install-source= to specify the source for binaries to install,
2340 --efi-boot-option-description= to control the name of the boot entry.
2342 * The sd-boot stub exports a StubFeatures flag, which is used by
2343 bootctl to show features supported by the stub that was used to boot.
2345 * The PE section offsets that are used by tools that assemble unified
2346 kernel images have historically been hard-coded. This may lead to
2347 overlapping PE sections which may break on boot. The UKI will now try
2348 to detect and warn about this.
2350 Any tools that assemble UKIs must update to calculate these offsets
2351 dynamically. Future sd-stub versions may use offsets that will not
2352 work with the currently used set of hard-coded offsets!
2354 * sd-stub now accepts (and passes to the initrd and then to the full
2355 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
2356 signatures of expected PCR values, to allow sealing secrets via the
2357 TPM2 against pre-calculated PCR measurements.
2359 Changes in the hardware database:
2361 * 'systemd-hwdb query' now supports the --root= option.
2363 Changes in systemctl:
2365 * systemctl now supports --state= and --type= options for the 'show'
2368 * systemctl gained a new verb 'list-automounts' to list automount
2371 * systemctl gained support for a new --image= switch to be able to
2372 operate on the specified disk image (similar to the existing --root=
2373 which operates relative to some directory).
2375 Changes in systemd-networkd:
2377 * networkd can set Linux NetLabel labels for integration with the
2378 network control in security modules via a new NetLabel= option.
2380 * The RapidCommit= is (re-)introduced to enable faster configuration
2381 via DHCPv6 (RFC 3315).
2383 * networkd gained a new option TCPCongestionControlAlgorithm= that
2384 allows setting a per-route TCP algorithm.
2386 * networkd gained a new option KeepFileDescriptor= to allow keeping a
2387 reference (file descriptor) open on TUN/TAP interfaces, which is
2388 useful to avoid link flaps while the underlying service providing the
2389 interface is being serviced.
2391 * RouteTable= now also accepts route table names.
2393 Changes in systemd-nspawn:
2395 * The --bind= and --overlay= options now support relative paths.
2397 * The --bind= option now supports a 'rootidmap' value, which will
2398 use id-mapped mounts to map the root user inside the container to the
2399 owner of the mounted directory on the host.
2401 Changes in systemd-resolved:
2403 * systemd-resolved now persists DNSOverTLS in its state file too. This
2404 fixes a problem when used in combination with NetworkManager, which
2405 sends the setting only once, causing it to be lost if resolved was
2406 restarted at any point.
2408 * systemd-resolved now exposes a Varlink socket at
2409 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
2410 root. Processed DNS requests in a JSON format will be published to
2411 any clients connected to this socket.
2413 resolvectl gained a 'monitor' verb to make use of this.
2415 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
2416 instead of returning SERVFAIL, as per RFC:
2417 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
2419 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
2420 is still supported.)
2422 Changes in libsystemd and other libraries:
2424 * libsystemd now exports sd_bus_error_setfv() (a convenience function
2425 for setting bus errors), sd_id128_string_equal (a convenience
2426 function for 128-bit ID string comparisons), and
2427 sd_bus_message_read_strv_extend() (a function to incrementally read
2430 * libsystemd now exports sd_device_get_child_first()/_next() as a
2431 high-level interface for enumerating child devices. It also supports
2432 sd_device_new_child() for opening a child device given a device
2435 * libsystemd now exports sd_device_monitor_set()/get_description()
2436 which allow setting a custom description that will be used in log
2437 messages by sd_device_monitor*.
2439 * Private shared libraries (libsystemd-shared-nnn.so,
2440 libsystemd-core-nnn.so) are now installed into arch-specific
2441 directories to allow multi-arch installs.
2443 * A new sd-gpt.h header is now published, listing GUIDs from the
2444 Discoverable Partitions specification. For more details see:
2445 https://systemd.io/DISCOVERABLE_PARTITIONS/
2447 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
2448 database given an explicit path to the file.
2450 * The signal number argument to sd_event_add_signal() now can now be
2451 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
2452 be automatically invoked to block the specified signal. This is
2453 useful to simplify invocations as the caller doesn't have to do this
2456 * A new convenience call sd_event_set_signal_exit() has been added to
2457 sd-event to set up signal handling so that the event loop
2458 automatically terminates cleanly on SIGTERM/SIGINT.
2460 Changes in other components:
2462 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
2463 can now be provided via the credential mechanism.
2465 * systemd-analyze gained a new verb 'compare-versions' that implements
2466 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
2467 'dpkg --compare-versions').
2469 * 'systemd-analyze dump' is extended to accept glob patterns for unit
2470 names to limit the output to matching units.
2472 * tmpfiles.d/ lines can read file contents to write from a credential.
2473 The new modifier char '^' is used to specify that the argument is a
2474 credential name. This mechanism is used to automatically populate
2475 /etc/motd, /etc/issue, and /etc/hosts from credentials.
2477 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
2478 an inode if the specification is prefixed with ':' and the inode
2481 * Default tmpfiles.d/ configuration now carries a line to automatically
2482 use an 'ssh.authorized_keys.root' credential if provided to set up
2483 the SSH authorized_keys file for the root user.
2485 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
2488 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
2489 in base64. This is useful to write arbitrary binary data into files.
2491 * The pkgconfig and rpm macros files now export the directory for user
2492 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
2494 * Detection of Apple Virtualization and detection of Parallels and
2495 KubeVirt virtualization on non-x86 archs have been added.
2497 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
2498 user when their system will become unsupported.
2500 * When performing suspend-then-hibernate, the system will estimate the
2501 discharge rate and use that to set the delay until hibernation and
2502 hibernate immediately instead of suspending when running from a
2503 battery and the capacity is below 5%.
2505 * systemd-sysctl gained a --strict option to fail when a sysctl
2506 setting is unknown to the kernel.
2508 * machinectl supports --force for the 'copy-to' and 'copy-from'
2511 * coredumpctl gained the --root and --image options to look for journal
2512 files under the specified root directory, image, or block device.
2514 * 'journalctl -o' and similar commands now implement a new output mode
2515 "short-delta". It is similar to "short-monotonic", but also shows the
2516 time delta between subsequent messages.
2518 * journalctl now respects the --quiet flag when verifying consistency
2521 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
2522 will indicate whether a message was logged in the 'initrd' phase or
2523 in the 'system' phase of the boot process.
2525 * Journal files gained a new compatibility flag
2526 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
2527 to the storage format that allow reducing size on disk. As with other
2528 compatibility flags, older journalctl versions will not be able to
2529 read journal files using this new format. The environment variable
2530 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
2531 disable this functionality. It is enabled by default.
2533 * systemd-run's --working-directory= switch now works when used in
2534 combination with --scope.
2536 * portablectl gained a --force flag to skip certain sanity checks. This
2537 is implemented using new flags accepted by systemd-portabled for the
2538 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
2539 flag now means that the attach/detach checks whether the units are
2540 already present and running will be skipped. Similarly,
2541 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
2542 image name matches the name declared inside of the image will be
2543 skipped. Callers must be sure to do those checks themselves if
2546 * systemd-portabled will now use the original filename to check
2547 extension-release.NAME for correctness, in case it is passed a
2550 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
2553 * sysext's extension-release files now support '_any' as a special
2554 value for the ID= field, to allow distribution-independent extensions
2555 (e.g.: fully statically compiled binaries, scripts). It also gained
2556 support for a new ARCHITECTURE= field that may be used to explicitly
2557 restrict an image to hosts of a specific architecture.
2559 * systemd-repart now supports creating squashfs partitions. This
2560 requires mksquashfs from squashfs-tools.
2562 * systemd-repart gained a --split flag to also generate split
2563 artifacts, i.e. a separate file for each partition. This is useful in
2564 conjunction with systemd-sysupdate or other tools, or to generate
2565 split dm-verity artifacts.
2567 * systemd-repart is now able to generate dm-verity partitions, including
2570 * systemd-repart can now set a partition UUID to zero, allowing it to
2571 be filled in later, such as when using verity partitions.
2573 * systemd-repart now supports drop-ins for its configuration files.
2575 * Package metadata logged by systemd-coredump in the system journal is
2578 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
2580 * systemd-oomd now automatically links against libatomic, if available.
2582 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
2585 * scope units now also provide oom-kill status.
2587 * systemd-pstore will now try to load only the efi_pstore kernel module
2588 before running, ensuring that pstore can be used.
2590 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
2591 session after a preconfigure timeout.
2593 * systemd-homed will now wait up to 30 seconds for workers to terminate,
2594 rather than indefinitely.
2596 * homectl gained a new '--luks-sector-size=' flag that allows users to
2597 select the preferred LUKS sector size. Must be a power of 2 between 512
2598 and 4096. systemd-userdbd records gained a corresponding field.
2600 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
2601 variable when generating the 'sp_lstchg' field, to ensure an image
2602 build can be reproducible.
2604 * 'udevadm wait' will now listen to kernel uevents too when called with
2607 * When naming network devices udev will now consult the Devicetree
2608 "alias" fields for the device.
2610 * systemd-udev will now create infiniband/by-path and
2611 infiniband/by-ibdev links for Infiniband verbs devices.
2613 * systemd-udev-trigger.service will now also prioritize input devices.
2615 * ConditionACPower= and systemd-ac-power will now assume the system is
2616 running on AC power if no battery can be found.
2618 * All features and tools using the TPM2 will now communicate with it
2619 using a bind key. Beforehand, the tpm2 support used encrypted sessions
2620 by creating a primary key that was used to encrypt traffic. This
2621 creates a problem as the key created for encrypting the traffic could
2622 be faked by an active interposer on the bus. In cases when a pin is
2623 used, a bind key will be used. The pin is used as the auth value for
2624 the seal key, aka the disk encryption key, and that auth value will be
2625 used in the session establishment. An attacker would need the pin
2626 value to create the secure session and thus an active interposer
2627 without the pin cannot interpose on TPM2 traffic.
2629 * systemd-growfs no longer requires udev to run.
2631 * systemd-backlight now will better support systems with multiple
2634 * systemd-cryptsetup's keyfile-timeout= option now also works when a
2635 device is used as a keyfile.
2637 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
2638 unlocking key from a key file (instead of prompting the user). Note
2639 that this is the key for unlocking the volume in order to be able to
2640 enroll a new key, but it is not the key that is enrolled.
2642 * systemd-dissect gained a new --umount switch that will safely and
2643 synchronously unmount all partitions of an image previously mounted
2644 with 'systemd-dissect --mount'.
2646 * When using gcrypt, all systemd tools and services will now configure
2647 it to prefer the OS random number generator if present.
2649 * All example code shipped with documentation has been relicensed from CC0
2652 * Unit tests will no longer fail when running on a system without
2655 Experimental features:
2657 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
2658 and bpftool >= 7.0).
2660 * sd-boot can automatically enroll SecureBoot keys from files found on
2661 the ESP. This enrollment can be either automatic ('force' mode) or
2662 controlled by the user ('manual' mode). It is sufficient to place the
2663 SecureBoot keys in the right place in the ESP and they will be picked
2664 up by sd-boot and shown in the boot menu.
2666 * The mkosi config in systemd gained support for automatically
2667 compiling a kernel with the configuration appropriate for testing
2668 systemd. This may be useful when developing or testing systemd in
2669 tandem with the kernel.
2671 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
2672 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
2673 Alexander Graf, Alexander Shopov, Alexander Wilson,
2674 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
2675 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
2676 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
2677 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
2678 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
2679 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
2680 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
2681 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
2682 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
2683 David Jaša, David Rheinsberg, David Seifert, David Tardon,
2684 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
2685 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
2686 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
2687 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
2688 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
2689 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
2690 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
2691 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
2692 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
2693 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
2694 JeroenHD, jiangchuangang, João Loureiro,
2695 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
2696 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
2697 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
2698 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
2699 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
2700 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
2701 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
2702 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
2703 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
2704 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
2705 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
2706 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
2707 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
2708 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
2709 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
2710 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
2711 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
2712 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
2713 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
2714 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
2715 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
2716 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
2717 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
2718 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
2720 – The Great Beyond, 2022-10-31 👻
2724 Backwards-incompatible changes:
2726 * The minimum kernel version required has been bumped from 3.13 to 4.15,
2727 and CLOCK_BOOTTIME is now assumed to always exist.
2729 * C11 with GNU extensions (aka "gnu11") is now used to build our
2730 components. Public API headers are still restricted to ISO C89.
2732 * In v250, a systemd-networkd feature that automatically configures
2733 routes to addresses specified in AllowedIPs= was added and enabled by
2734 default. However, this causes network connectivity issues in many
2735 existing setups. Hence, it has been disabled by default since
2736 systemd-stable 250.3. The feature can still be used by explicitly
2737 configuring RouteTable= setting in .netdev files.
2739 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
2740 when a Condition*= check does not succeed, restoring the JobRemoved
2741 signal to the behaviour it had before v250.
2743 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
2744 GetImageMetadataWithExtensions() have been fixed to provide an extra
2745 return parameter, containing the actual extension release metadata.
2746 The current implementation was judged to be broken and unusable, and
2747 thus the usual procedure of adding a new set of methods was skipped,
2748 and backward compatibility broken instead on the assumption that
2749 nobody can be affected given the current state of this interface.
2751 * All kernels supported by systemd mix bytes returned by RDRAND (or
2752 similar) into the entropy pool at early boot. This means that on
2753 those systems, even if /dev/urandom is not yet initialized, it still
2754 returns bytes that are of at least RDRAND quality. For that reason,
2755 we no longer have reason to invoke RDRAND from systemd itself, which
2756 has historically been a source of bugs. Furthermore, kernels ≥5.6
2757 provide the getrandom(GRND_INSECURE) interface for returning random
2758 bytes before the entropy pool is initialized without warning into
2759 kmsg, which is what we attempt to use if available. systemd's direct
2760 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
2761 running an older kernel may experience kmsg warnings that were not
2762 seen with 250. For newer kernels, non-x86 systems, or older x86
2763 systems, there should be no visible changes.
2765 * sd-boot will now measure the kernel command line into TPM PCR 12
2766 rather than PCR 8. This improves usefulness of the measurements on
2767 systems where sd-boot is chainloaded from Grub. Grub measures all
2768 commands its executes into PCR 8, which makes it very hard to use
2769 reasonably, hence separate ourselves from that and use PCR 12
2770 instead, which is what certain Ubuntu editions already do. To retain
2771 compatibility with systems running older systemd systems a new meson
2772 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
2773 If enabled, the measurement is done twice: into the new-style PCR 12
2774 *and* the old-style PCR 8. It's strongly advised to migrate all users
2775 to PCR 12 for this purpose in the long run, as we intend to remove
2776 this compatibility feature in two years' time.
2778 * busctl capture now writes output in the newer pcapng format instead
2781 * A udev rule that imported hwdb matches for USB devices with lowercase
2782 hexadecimal vendor/product ID digits was added in systemd 250. This
2783 has been reverted, since uppercase hexadecimal digits are supposed to
2784 be used, and we already had a rule with the appropriate match.
2786 Users might need to adjust their local hwdb entries.
2788 * arch_prctl(2) has been moved to the @default set in the syscall filters
2789 (as exposed via the SystemCallFilter= setting in service unit files).
2790 It is apparently used by the linker now.
2792 * The tmpfiles entries that create the /run/systemd/netif directory and
2793 its subdirectories were moved from tmpfiles.d/systemd.conf to
2794 tmpfiles.d/systemd-network.conf.
2796 Users might need to adjust their files that override tmpfiles.d/systemd.conf
2797 to account for this change.
2799 * The requirement for Portable Services images to contain a well-formed
2800 os-release file (i.e.: contain at least an ID field) is now enforced.
2801 This applies to base images and extensions, and also to systemd-sysext.
2803 Changes in the Boot Loader Specification, kernel-install and sd-boot:
2805 * kernel-install's and bootctl's Boot Loader Specification Type #1
2806 entry generation logic has been reworked. The user may now pick
2807 explicitly by which "token" string to name the installation's boot
2808 entries, via the new /etc/kernel/entry-token file or the new
2809 --entry-token= switch to bootctl. By default — as before — the
2810 entries are named after the local machine ID. However, in "golden
2811 image" environments, where the machine ID shall be initialized on
2812 first boot (as opposed to at installation time before first boot) the
2813 machine ID will not be available at build time. In this case the
2814 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
2815 file) may be used to override the "token" for the entries, for
2816 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
2817 make the OS images independent of any machine ID, and ensure that the
2818 images will not carry any identifiable information before first boot,
2819 but on the other hand means that multiple parallel installations of
2820 the very same image on the same disk cannot be supported.
2822 Summary: if you are building golden images that shall acquire
2823 identity information exclusively on first boot, make sure to both
2824 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
2825 value of the IMAGE_ID= or ID= field of /etc/os-release or another
2826 suitable identifier before deploying the image.
2828 * The Boot Loader Specification has been extended with
2829 /loader/entries.srel file located in the EFI System Partition (ESP)
2830 that disambiguates the format of the entries in the /loader/entries/
2831 directory (in order to discern them from incompatible uses of this
2832 directory by other projects). For entries that follow the
2833 Specification, the string "type1" is stored in this file.
2835 bootctl will now write this file automatically when installing the
2836 systemd-boot boot loader.
2838 * kernel-install supports a new initrd_generator= setting in
2839 /etc/kernel/install.conf, that is exported as
2840 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
2841 allows choosing different initrd generators.
2843 * kernel-install will now create a "staging area" (an initially-empty
2844 directory to gather files for a Boot Loader Specification Type #1
2845 entry). The path to this directory is exported as
2846 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
2847 drop files there instead of writing them directly to the final
2848 location. kernel-install will move them when all files have been
2849 prepared successfully.
2851 * New option sort-key= has been added to the Boot Loader Specification
2852 to override the sorting order of the entries in the boot menu. It is
2853 read by sd-boot and bootctl, and will be written by kernel-install,
2854 with the default value of IMAGE_ID= or ID= fields from
2855 os-release. Together, this means that on multiboot installations,
2856 entries should be grouped and sorted in a predictable way.
2858 * The sort order of boot entries has been updated: entries which have
2859 the new field sort-key= are sorted by it first, and all entries
2860 without it are ordered later. After that, entries are sorted by
2861 version so that newest entries are towards the beginning of the list.
2863 * The kernel-install tool gained a new 'inspect' verb which shows the
2864 paths and other settings used.
2866 * sd-boot can now optionally beep when the menu is shown and menu
2867 entries are selected, which can be useful on machines without a
2868 working display. (Controllable via a loader.conf setting.)
2870 * The --make-machine-id-directory= switch to bootctl has been replaced
2871 by --make-entry-directory=, given that the entry directory is not
2872 necessarily named after the machine ID, but after some other suitable
2873 ID as selected via --entry-token= described above. The old name of
2874 the option is still understood to maximize compatibility.
2876 * 'bootctl list' gained support for a new --json= switch to output boot
2877 menu entries in JSON format.
2879 * 'bootctl is-installed' now supports the --graceful, and various verbs
2880 omit output with the new option --quiet.
2882 Changes in systemd-homed:
2884 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
2885 of activated home directories it manages (if the kernel and selected
2886 file systems support it). So far it mapped three UID ranges: the
2887 range from 0…60000, the user's own UID, and the range 60514…65534,
2888 leaving everything else unmapped (in other words, the 16-bit UID range
2889 is mapped almost fully, with the exception of the UID subrange used
2890 for systemd-homed users, with one exception: the user's own UID).
2891 Unmapped UIDs may not be used for file ownership in the home
2892 directory — any chown() attempts with them will fail. With this
2893 release a fourth range is added to these mappings:
2894 524288…1879048191. This range is the UID range intended for container
2897 https://systemd.io/UIDS-GIDS
2899 This range may be used for container managers that place container OS
2900 trees in the home directory (which is a questionable approach, for
2901 quota, permission, SUID handling and network file system
2902 compatibility reasons, but nonetheless apparently commonplace). Note
2903 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
2904 UID assignments from the range are not managed or mapped by
2905 `systemd-homed`, and must be managed with other mechanisms, in the
2906 context of the local system.
2908 Typically, a better approach to user namespacing in relevant
2909 container managers would be to leave container OS trees on disk at
2910 UID offset 0, but then map them to a dynamically allocated runtime
2911 UID range via another UID mount map at container invocation
2912 time. That way user namespace UID ranges become strictly a runtime
2913 concept, and do not leak into persistent file systems, persistent
2914 user databases or persistent configuration, thus greatly simplifying
2915 handling, and improving compatibility with home directories intended
2916 to be portable like the ones managed by systemd-homed.
2918 Changes in shared libraries:
2920 * A new libsystemd-core-<version>.so private shared library is
2921 installed under /usr/lib/systemd/system, mirroring the existing
2922 libsystemd-shared-<version>.so library. This allows the total
2923 installation size to be reduced by binary code reuse.
2925 * The <version> tag used in the name of libsystemd-shared.so and
2926 libsystemd-core.so can be configured via the meson option
2927 'shared-lib-tag'. Distributions may build subsequent versions of the
2928 systemd package with unique tags (e.g. the full package version),
2929 thus allowing multiple installations of those shared libraries to be
2930 available at the same time. This is intended to fix an issue where
2931 programs that link to those libraries would fail to execute because
2932 they were installed earlier or later than the appropriate version of
2935 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
2936 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
2937 format instead of as a simple series of hex characters.
2939 * The sd-device API gained two new calls sd_device_new_from_devname()
2940 and sd_device_new_from_path() which permit allocating an sd_device
2941 object from a device node name or file system path.
2943 * sd-device also gained a new call sd_device_open() which will open the
2944 device node associated with a device for which an sd_device object
2945 has been allocated. The call is supposed to address races around
2946 device nodes being removed/recycled due to hotplug events, or media
2947 change events: the call checks internally whether the major/minor of
2948 the device node and the "diskseq" (in case of block devices) match
2949 with the metadata loaded in the sd_device object, thus ensuring that
2950 the device once opened really matches the provided sd_device object.
2952 Changes in PID1, systemctl, and systemd-oomd:
2954 * A new set of service monitor environment variables will be passed to
2955 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
2956 handler unit as OnFailure=/OnSuccess=. The variables are:
2957 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
2958 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
2959 handler needs to watch multiple units, use a templated handler.
2961 * A new ExtensionDirectories= setting in service unit files allows
2962 system extensions to be loaded from a directory. (It is similar to
2963 ExtensionImages=, but takes paths to directories, instead of
2966 'portablectl attach --extension=' now also accepts directory paths.
2968 * The user.delegate and user.invocation_id extended attributes on
2969 cgroups are used in addition to trusted.delegate and
2970 trusted.invocation_id. The latter pair requires privileges to set,
2971 but the former doesn't and can be also set by the unprivileged user
2974 (Only supported on kernels ≥5.6.)
2976 * Units that were killed by systemd-oomd will now have a service result
2977 of 'oom-kill'. The number of times a service was killed is tallied
2978 in the 'user.oomd_ooms' extended attribute.
2980 The OOMPolicy= unit file setting is now also honoured by
2983 * In unit files the new %y/%Y specifiers can be used to refer to
2984 normalized unit file path, which is particularly useful for symlinked
2987 The new %q specifier resolves to the pretty hostname
2988 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
2990 The new %d specifier resolves to the credentials directory of a
2991 service (same as $CREDENTIALS_DIRECTORY).
2993 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
2994 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
2995 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
2996 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
2997 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
2998 MountFlags= service settings now also work in unprivileged user
2999 services, i.e. those run by the user's --user service manager, as long
3000 as user namespaces are enabled on the system.
3002 * Services with Restart=always and a failing ExecCondition= will no
3003 longer be restarted, to bring ExecCondition= behaviour in line with
3004 Condition*= settings.
3006 * LoadCredential= now accepts a directory as the argument; all files
3007 from the directory will be loaded as credentials.
3009 * A new D-Bus property ControlGroupId is now exposed on service units,
3010 that encapsulates the service's numeric cgroup ID that newer kernels
3011 assign to each cgroup.
3013 * PID 1 gained support for configuring the "pre-timeout" of watchdog
3014 devices and the associated governor, via the new
3015 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
3016 options in /etc/systemd/system.conf.
3018 * systemctl's --timestamp= option gained a new choice "unix", to show
3019 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
3021 * A new "taint" flag named "old-kernel" is introduced which is set when
3022 the kernel systemd runs on is older then the current baseline version
3023 (see above). The flag is shown in "systemctl status" output.
3025 * Two additional taint flags "short-uid-range" and "short-gid-range"
3026 have been added as well, which are set when systemd notices it is run
3027 within a userns namespace that does not define the full 0…65535 UID
3030 * A new "unmerged-usr" taint flag has been added that is set whenever
3031 running on systems where /bin/ + /sbin/ are *not* symlinks to their
3032 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
3035 * Generators invoked by PID 1 will now have a couple of useful
3036 environment variables set describing the execution context a
3037 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
3038 system service manager, or from the per-user service
3039 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
3040 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
3041 systemd considers the current boot to be a "first"
3042 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
3043 detected and which type of hypervisor/container
3044 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
3045 kernel is built for.
3047 * PID 1 will now automatically pick up system credentials from qemu's
3048 fw_cfg interface, thus allowing passing arbitrary data into VM
3049 systems similar to how this is already supported for passing them
3050 into `systemd-nspawn` containers. Credentials may now also be passed
3051 in via the new kernel command line option `systemd.set_credential=`
3052 (note that kernel command line options are world-readable during
3053 runtime, and only useful for credentials that require no
3054 confidentiality). The credentials that can be passed to unified
3055 kernels that use the `systemd-stub` UEFI stub are now similarly
3056 picked up automatically. Automatic importing of system credentials
3057 this way can be turned off via the new
3058 `systemd.import_credentials=no` kernel command line option.
3060 * LoadCredential= will now automatically look for credentials in the
3061 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
3062 the argument is not an absolute path. Similarly,
3063 LoadCredentialEncrypted= will check the same directories plus
3064 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
3065 /usr/lib/credstore.encrypted/. The idea is to use those directories
3066 as the system-wide location for credentials that services should pick
3069 * System and service credentials are described in great detail in a new
3072 https://systemd.io/CREDENTIALS
3074 Changes in systemd-journald:
3076 * The journal JSON export format has been added to listed of stable
3077 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
3079 * journalctl --list-boots now supports JSON output and the --reverse option.
3081 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
3082 updated, BUILDING_IMAGES is new:
3084 https://systemd.io/JOURNAL_EXPORT_FORMATS
3085 https://systemd.io/BUILDING_IMAGES
3089 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
3090 calculators, etc.), the other AV production devices (DJ tables,
3091 keypads, etc.) that should accessible to the seat owner user by
3094 * udevadm trigger gained a new --prioritized-subsystem= option to
3095 process certain subsystems (and all their parent devices) earlier.
3097 systemd-udev-trigger.service now uses this new option to trigger
3098 block and TPM devices first, hopefully making the boot a bit faster.
3100 * udevadm trigger now implements --type=all, --initialized-match,
3101 --initialized-nomatch to trigger both subsystems and devices, only
3102 already-initialized devices, and only devices which haven't been
3103 initialized yet, respectively.
3105 * udevadm gained a new "wait" command for safely waiting for a specific
3106 device to show up in the udev device database. This is useful in
3107 scripts that asynchronously allocate a block device (e.g. through
3108 repartitioning, or allocating a loopback device or similar) and need
3109 to synchronize on the creation to complete.
3111 * udevadm gained a new "lock" command for locking one or more block
3112 devices while formatting it or writing a partition table to it. It is
3113 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
3114 usable in scripts dealing with block devices.
3116 * udevadm info will show a couple of additional device fields in its
3117 output, and will not apply a limited set of coloring to line types.
3119 * udevadm info --tree will now show a tree of objects (i.e. devices and
3120 suchlike) in the /sys/ hierarchy.
3122 * Block devices will now get a new set of device symlinks in
3123 /dev/disk/by-diskseq/<nr>, which may be used to reference block
3124 device nodes via the kernel's "diskseq" value. Note that this does
3125 not guarantee that opening a device by a symlink like this will
3126 guarantee that the opened device actually matches the specified
3127 diskseq value. To be safe against races, the actual diskseq value of
3128 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
3129 the one in the symlink path.
3131 * .link files gained support for setting MDI/MID-X on a link.
3133 * .link files gained support for [Match] Firmware= setting to match on
3134 the device firmware description string. By mistake, it was previously
3135 only supported in .network files.
3137 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
3138 and [SR-IOV] section to configure SR-IOV virtual functions.
3140 Changes in systemd-networkd:
3142 * The default scope for unicast routes configured through [Route]
3143 section is changed to "link", to make the behavior consistent with
3144 "ip route" command. The manual configuration of [Route] Scope= is
3147 * A new unit systemd-networkd-wait-online@<interface>.service has been
3148 added that can be used to wait for a specific network interface to be
3151 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
3152 that configures the eponymous kernel attribute on the bridge.
3154 * .netdev files now can be used to create virtual WLAN devices, and
3155 configure various settings on them, via the [WLAN] section.
3157 * .link/.network files gained support for [Match] Kind= setting to match
3158 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
3160 This value is also shown by 'networkctl status'.
3162 * The Local= setting in .netdev files for various virtual network
3163 devices gained support for specifying, in addition to the network
3164 address, the name of a local interface which must have the specified
3167 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
3168 files, to configure tunnels in external mode (a.k.a. collect metadata
3171 * [Network] L2TP= setting was removed. Please use interface specifier in
3172 Local= setting in .netdev files of corresponding L2TP interface.
3174 * New [DHCPServer] BootServerName=, BootServerAddress=, and
3175 BootFilename= settings can be used to configure the server address,
3176 server name, and file name sent in the DHCP packet (e.g. to configure
3179 Changes in systemd-resolved:
3181 * systemd-resolved is started earlier (in sysinit.target), so it
3182 available earlier and will also be started in the initrd if installed
3185 Changes in disk encryption:
3187 * systemd-cryptenroll can now control whether to require the user to
3188 enter a PIN when using TPM-based unlocking of a volume via the new
3189 --tpm2-with-pin= option.
3191 Option tpm2-pin= can be used in /etc/crypttab.
3193 * When unlocking devices via TPM, TPM2 parameter encryption is now
3194 used, to ensure that communication between CPU and discrete TPM chips
3195 cannot be eavesdropped to acquire disk encryption keys.
3197 * A new switch --fido2-credential-algorithm= has been added to
3198 systemd-cryptenroll allowing selection of the credential algorithm to
3199 use when binding encryption to FIDO2 tokens.
3201 Changes in systemd-hostnamed:
3203 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
3204 to override the values gleaned from the hwdb.
3206 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
3207 /sys/class/dmi/id) to override the chassis that is reported by
3210 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
3211 for reading the hardware serial number, as reportd by DMI. It also
3212 exposes a new method D-Bus property FirmwareVersion that encode the
3213 firmware version of the system.
3215 Changes in other components:
3217 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
3218 handling with the values that were configured during systemd build
3219 (if /etc/locale.conf has not been created through some other
3220 mechanism). This means that /etc/locale.conf should always have
3221 reasonable contents and we avoid a potential mismatch in defaults.
3223 * The userdbctl tool will now show UID range information as part of the
3224 list of known users.
3226 * A new build-time configuration setting default-user-shell= can be
3227 used to set the default shell for user records and nspawn shell
3228 invocations (instead of the default /bin/bash).
3230 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
3231 information dynamically at runtime via IPC.
3233 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
3234 whether a functioning TPM2 infrastructure is available, i.e. if
3235 firmware, kernel driver and systemd all have TPM2 support enabled and
3238 * The systemd-creds tool gained support for generating encrypted
3239 credentials that are using an empty encryption key. While this
3240 provides no integrity nor confidentiality it's useful to implement
3241 codeflows that work the same on TPM-ful and TPM2-less systems. The
3242 service manager will only accept credentials "encrypted" that way if
3243 a TPM2 device cannot be detected, to ensure that credentials
3244 "encrypted" like that cannot be used to trick TPM2 systems.
3246 * When deciding whether to colorize output, all systemd programs now
3247 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
3250 * Meson's new install_tag feature is now in use for several components,
3251 allowing to build and install select binaries only: pam, nss, devel
3252 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
3253 $ meson build systemd-boot
3254 $ meson install --tags systemd-boot --no-rebuild
3255 https://mesonbuild.com/Installing.html#installation-tags
3257 * A new build configuration option has been added, to allow selecting the
3258 default compression algorithm used by systemd-journald and systemd-coredump.
3259 This allows to build-in support for decompressing all supported formats,
3260 but choose a specific one for compression. E.g.:
3261 $ meson -Ddefault-compression=xz
3263 Experimental features:
3265 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
3266 loader.conf that implements booting Microsoft Windows from the
3267 sd-boot in a way that first reboots the system, to reset the TPM
3268 PCRs. This improves compatibility with BitLocker's TPM use, as the
3269 PCRs will only record the Windows boot process, and not sd-boot
3270 itself, thus retaining the PCR measurements not involving sd-boot.
3271 Note that this feature is experimental for now, and is likely going
3272 to be generalized and renamed in a future release, without retaining
3273 compatibility with the current implementation.
3275 * A new systemd-sysupdate component has been added that automatically
3276 discovers, downloads, and installs A/B-style updates for the host
3277 installation itself, or container images, portable service images,
3278 and other assets. See the new systemd-sysupdate man page for updates.
3280 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
3281 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
3282 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
3283 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
3284 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
3285 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
3286 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
3287 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
3288 David, David Bond, Davide Cavalca, David Tardon, davijosw,
3289 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
3290 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
3291 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
3292 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
3293 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
3294 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
3295 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
3296 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
3297 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
3298 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
3299 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
3300 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
3301 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
3302 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
3303 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
3304 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
3305 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
3306 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
3307 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
3308 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
3309 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
3310 Simon Ellmann, Sonali Srivastava, Stefan Seering,
3311 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
3312 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
3313 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
3314 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
3315 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
3316 Zbigniew Jędrzejewski-Szmek, наб
3318 — Edinburgh, 2022-05-21
3322 * Support for encrypted and authenticated credentials has been added.
3323 This extends the credential logic introduced with v247 to support
3324 non-interactive symmetric encryption and authentication, based on a
3325 key that is stored on the /var/ file system or in the TPM2 chip (if
3326 available), or the combination of both (by default if a TPM2 chip
3327 exists the combination is used, otherwise the /var/ key only). The
3328 credentials are automatically decrypted at the moment a service is
3329 started, and are made accessible to the service itself in unencrypted
3330 form. A new tool 'systemd-creds' encrypts credentials for this
3331 purpose, and two new service file settings LoadCredentialEncrypted=
3332 and SetCredentialEncrypted= configure such credentials.
3334 This feature is useful to store sensitive material such as SSL
3335 certificates, passwords and similar securely at rest and only decrypt
3336 them when needed, and in a way that is tied to the local OS
3337 installation or hardware.
3339 * systemd-gpt-auto-generator can now automatically set up discoverable
3340 LUKS2 encrypted swap partitions.
3342 * The GPT Discoverable Partitions Specification has been substantially
3343 extended with support for root and /usr/ partitions for the majority
3344 of architectures systemd supports. This includes platforms that do
3345 not natively support UEFI, because even though GPT is specified under
3346 UEFI umbrella, it is useful on other systems too. Specifically,
3347 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
3348 Portable Services use the concept without requiring UEFI.
3350 * The GPT Discoverable Partitions Specifications has been extended with
3351 a new set of partitions that may carry PKCS#7 signatures for Verity
3352 partitions, encoded in a simple JSON format. This implements a simple
3353 mechanism for building disk images that are fully authenticated and
3354 can be tested against a set of cryptographic certificates. This is
3355 now implemented for the various systemd tools that can operate with
3356 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
3357 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
3358 The PKCS#7 signatures are passed to the kernel (where they are
3359 checked against certificates from the kernel keyring), or can be
3360 verified against certificates provided in userspace (via a simple
3361 drop-in file mechanism).
3363 * systemd-dissect's inspection logic will now report for which uses a
3364 disk image is intended. Specifically, it will display whether an
3365 image is suitable for booting on UEFI or in a container (using
3366 systemd-nspawn's --image= switch), whether it can be used as portable
3367 service, or attached as system extension.
3369 * The system-extension.d/ drop-in files now support a new field
3370 SYSEXT_SCOPE= that may encode which purpose a system extension image
3371 is for: one of "initrd", "system" or "portable". This is useful to
3372 make images more self-descriptive, and to ensure system extensions
3373 cannot be attached in the wrong contexts.
3375 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
3376 be used in portable service images to indicate which unit prefixes
3379 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
3380 now is able to decode images for non-native architectures as well.
3381 This allows systemd-nspawn to boot images of non-native architectures
3382 if the corresponding user mode emulator is installed and
3383 systemd-binfmtd is running.
3385 * systemd-logind gained new settings HandlePowerKeyLongPress=,
3386 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
3387 HandleHibernateKeyLongPress= which may be used to configure actions
3388 when the relevant keys are pressed for more than 5s. This is useful
3389 on devices that only have hardware for a subset of these keys. By
3390 default, if the reboot key is pressed long the poweroff operation is
3391 now triggered, and when the suspend key is pressed long the hibernate
3392 operation is triggered. Long pressing the other two keys currently
3393 does not trigger any operation by default.
3395 * When showing unit status updates on the console during boot and
3396 shutdown, and a service is slow to start so that the cylon animation
3397 is shown, the most recent sd_notify() STATUS= text is now shown as
3398 well. Services may use this to make the boot/shutdown output easier
3399 to understand, and to indicate what precisely a service that is slow
3400 to start or stop is waiting for. In particular, the per-user service
3401 manager instance now reports what it is doing and which service it is
3402 waiting for this way to the system service manager.
3404 * The service manager will now re-execute on reception of the
3405 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
3406 only when running as PID 1. There was no signal to request this when
3407 running as per-user service manager, i.e. as any other PID than 1.
3408 SIGRTMIN+25 works for both system and user managers.
3410 * The hardware watchdog logic in PID 1 gained support for operating
3411 with the default timeout configured in the hardware, instead of
3412 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
3413 request this behavior.
3415 * A new kernel command line option systemd.watchdog_sec= is now
3416 understood which may be used to override the hardware watchdog
3417 time-out for the boot.
3419 * A new setting DefaultOOMScoreAdjust= is now supported in
3420 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
3421 to set the default process OOM score adjustment value for processes
3422 started by the service manager. For per-user service managers this
3423 now defaults to 100, but for per-system service managers is left as
3424 is. This means that by default now services forked off the user
3425 service manager are more likely to be killed by the OOM killer than
3426 system services or the managers themselves.
3428 * A new per-service setting RestrictFileSystems= as been added that
3429 restricts the file systems a service has access to by their type.
3430 This is based on the new BPF LSM of the Linux kernel. It provides an
3431 effective way to make certain API file systems unavailable to
3432 services (and thus minimizing attack surface). A new command
3433 "systemd-analyze filesystems" has been added that lists all known
3434 file system types (and how they are grouped together under useful
3437 * Services now support a new setting RestrictNetworkInterfaces= for
3438 restricting access to specific network interfaces.
3440 * Service unit files gained new settings StartupAllowedCPUs= and
3441 StartupAllowedMemoryNodes=. These are similar to their counterparts
3442 without the "Startup" prefix and apply during the boot process
3443 only. This is useful to improve boot-time behavior of the system and
3444 assign resources differently during boot than during regular
3445 runtime. This is similar to the preexisting StartupCPUWeight=
3448 * Related to this: the various StartupXYZ= settings
3449 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
3450 during shutdown. The settings not prefixed with "Startup" hence apply
3451 during regular runtime, and those that are prefixed like that apply
3452 during boot and shutdown.
3454 * A new per-unit set of conditions/asserts
3455 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
3456 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
3457 pressure is above the configured threshold, using the kernel PSI
3458 feature. For more details see systemd.unit(5) and
3459 https://docs.kernel.org/accounting/psi.html
3461 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
3462 ProtectKernelLogs=yes can now be used.
3464 * The default maximum numbers of inodes have been raised from 64k to 1M
3465 for /dev/, and from 400k to 1M for /tmp/.
3467 * The per-user service manager learnt support for communicating with
3468 systemd-oomd to acquire OOM kill information.
3470 * A new service setting ExecSearchPath= has been added that allows
3471 changing the search path for executables for services. It affects
3472 where we look for the binaries specified in ExecStart= and similar,
3473 and the specified directories are also added the $PATH environment
3474 variable passed to invoked processes.
3476 * A new setting RuntimeRandomizedExtraSec= has been added for service
3477 and scope units that allows extending the runtime time-out as
3478 configured by RuntimeMaxSec= with a randomized amount.
3480 * The syntax of the service unit settings RuntimeDirectory=,
3481 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
3482 if the specified value is now suffixed with a colon, followed by
3483 another filename, the latter will be created as symbolic link to the
3484 specified directory. This allows creating these service directories
3485 together with alias symlinks to make them available under multiple
3488 * Service unit files gained two new settings TTYRows=/TTYColumns= for
3489 configuring rows/columns of the TTY device passed to
3490 stdin/stdout/stderr of the service. This is useful to propagate TTY
3491 dimensions to a virtual machine.
3493 * A new service unit file setting ExitType= has been added that
3494 specifies when to assume a service has exited. By default systemd
3495 only watches the main process of a service. By setting
3496 ExitType=cgroup it can be told to wait for the last process in a
3499 * Automount unit files gained a new setting ExtraOptions= that can be
3500 used to configure additional mount options to pass to the kernel when
3501 mounting the autofs instance.
3503 * "Urlification" (generation of ESC sequences that generate clickable
3504 hyperlinks in modern terminals) may now be turned off altogether
3507 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
3508 settings that default to 200 and 2 s respectively. The ratelimit
3509 ensures that a path unit cannot cause PID1 to busy-loop when it is
3510 trying to trigger a service that is skipped because of a Condition*=
3511 not being satisfied. This matches the configuration and behaviour of
3514 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
3515 as a plug-in for cryptsetup. This means the plain cryptsetup command
3516 may now be used to unlock volumes set up this way.
3518 * The TPM2 logic in cryptsetup will now automatically detect systems
3519 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
3520 updates the SHA1 banks. In such a case PCR policies will be
3521 automatically bound to the latter, not the former. This makes the PCR
3522 policies reliable, but of course do not provide the same level of
3523 trust as SHA256 banks.
3525 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
3526 RSA primary keys in addition to ECC, improving compatibility with
3527 TPM2 chips that do not support ECC. RSA keys are much slower to use
3528 than ECC, and hence are only used if ECC is not available.
3530 * /etc/crypttab gained support for a new token-timeout= setting for
3531 encrypted volumes that allows configuration of the maximum time to
3532 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
3533 the logic will query the user for a regular passphrase/recovery key
3536 * Support for activating dm-integrity volumes at boot via a new file
3537 /etc/integritytab and the tool systemd-integritysetup have been
3538 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
3539 with dm-integrity instead of dm-crypt/dm-verity.
3541 * The systemd-veritysetup-generator now understands a new usrhash=
3542 kernel command line option for specifying the Verity root hash for
3543 the partition backing the /usr/ file system. A matching set of
3544 systemd.verity_usr_* kernel command line options has been added as
3545 well. These all work similar to the corresponding options for the
3548 * The sd-device API gained a new API call sd_device_get_diskseq() to
3549 return the DISKSEQ property of a device structure. The "disk
3550 sequence" concept is a new feature recently introduced to the Linux
3551 kernel that allows detecting reuse cycles of block devices, i.e. can
3552 be used to recognize when loopback block devices are reused for a
3553 different purpose or CD-ROM drives get their media changed.
3555 * A new unit systemd-boot-update.service has been added. If enabled
3556 (the default) and the sd-boot loader is detected to be installed, it
3557 is automatically updated to the newest version when out of date. This
3558 is useful to ensure the boot loader remains up-to-date, and updates
3559 automatically propagate from the OS tree in /usr/.
3561 * sd-boot will now build with SBAT by default in order to facilitate
3562 working with recent versions of Shim that require it to be present.
3564 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
3565 This is used to robustly generate boot entry titles for Windows.
3567 * A new generic target unit factory-reset.target has been added. It is
3568 hooked into systemd-logind similar in fashion to
3569 reboot/poweroff/suspend/hibernate, and is supposed to be used to
3570 initiate a factory reset operation. What precisely this operation
3571 entails is up for the implementer to decide, the primary goal of the
3572 new unit is provide a framework where to plug in the implementation
3573 and how to trigger it.
3575 * A new meson build-time option 'clock-valid-range-usec-max' has been
3576 added which takes a time in µs and defaults to 15 years. If the RTC
3577 time is noticed to be more than the specified time ahead of the
3578 built-in epoch of systemd (which by default is the release timestamp
3579 of systemd) it is assumed that the RTC is not working correctly, and
3580 the RTC is reset to the epoch. (It already is reset to the epoch when
3581 noticed to be before it.) This should increase the chance that time
3582 doesn't accidentally jump too far ahead due to faulty hardware or
3585 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
3586 which may be used to automatically save the current system time to
3587 disk in regular intervals. This is useful to maintain a roughly
3588 monotonic clock even without RTC hardware and with some robustness
3589 against abnormal system shutdown.
3591 * systemd-analyze verify gained support for a pair of new --image= +
3592 --root= switches for verifying units below a specific root
3593 directory/image instead of on the host.
3595 * systemd-analyze verify gained support for verifying unit files under
3596 an explicitly specified unit name, independently of what the filename
3599 * systemd-analyze verify gained a new switch --recursive-errors= which
3600 controls whether to only fail on errors found in the specified units
3601 or recursively any dependent units.
3603 * systemd-analyze security now supports a new --offline mode for
3604 analyzing unit files stored on disk instead of loaded units. It may
3605 be combined with --root=/--image to analyze unit files under a root
3606 directory or disk image. It also learnt a new --threshold= parameter
3607 for specifying an exposure level threshold: if the exposure level
3608 exceeds the specified value the call will fail. It also gained a new
3609 --security-policy= switch for configuring security policies to
3610 enforce on the units. A policy is a JSON file that lists which tests
3611 shall be weighted how much to determine the overall exposure
3612 level. Altogether these new features are useful for fully automatic
3613 analysis and enforcement of security policies on unit files.
3615 * systemd-analyze security gain a new --json= switch for JSON output.
3617 * systemd-analyze learnt a new --quiet switch for reducing
3618 non-essential output. It's honored by the "dot", "syscall-filter",
3619 "filesystems" commands.
3621 * systemd-analyze security gained a --profile= option that can be used
3622 to take into account a portable profile when analyzing portable
3623 services, since a lot of the security-related settings are enabled
3626 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
3627 files, binaries and executables and prints metadata information,
3628 including the build-id and other info described on:
3629 https://systemd.io/COREDUMP_PACKAGE_METADATA/
3631 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
3632 section, for automatically propagating DNS settings from other
3635 * The static lease DHCP server logic in systemd-networkd may now serve
3636 IP addresses outside of the configured IP pool range for the server.
3638 * CAN support in systemd-networkd gained four new settings Loopback=,
3639 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
3640 control modes. It gained a number of further settings for tweaking
3643 * The [CAN] section in .network file gained new TimeQuantaNSec=,
3644 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
3645 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
3646 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
3647 DataSyncJumpWidth= settings to control bit-timing processed by the
3650 * DHCPv4 client support in systemd-networkd learnt a new Label= option
3651 for configuring the address label to apply to configure IPv4
3654 * The [IPv6AcceptRA] section of .network files gained support for a new
3655 UseMTU= setting that may be used to control whether to apply the
3656 announced MTU settings to the local interface.
3658 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
3659 setting to control whether the DHCPv4 client request and process the
3662 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
3663 [DHCPPrefixDelegation], as now the prefix delegation is also supported
3664 with DHCPv4 protocol by enabling the Use6RD= setting.
3666 * The [DHCPPrefixDelegation] section in .network file gained a new
3667 setting UplinkInterface= to specify the upstream interface.
3669 * The [DHCPv6] section in .network file gained a new setting
3670 UseDelegatedPrefix= to control whether the delegated prefixes will be
3671 propagated to the downstream interfaces.
3673 * The [IPv6AcceptRA] section of .network files now understands two new
3674 settings UseGateway=/UseRoutePrefix= for explicitly configuring
3675 whether to use the relevant fields from the IPv6 Router Advertisement
3678 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
3679 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
3680 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
3681 [IPv6AcceptRA] section to control when the DHCPv6 client is started
3682 and how the delegated prefixes are handled by the DHCPv6 client.
3684 * The IPv6Token= section in the [Network] section is deprecated, and
3685 the [IPv6AcceptRA] section gained the Token= setting for its
3686 replacement. The [IPv6Prefix] section also gained the Token= setting.
3687 The Token= setting gained 'eui64' mode to explicitly configure an
3688 address with the EUI64 algorithm based on the interface MAC address.
3689 The 'prefixstable' mode can now optionally take a secret key. The
3690 Token= setting in the [DHCPPrefixDelegation] section now supports all
3691 algorithms supported by the same settings in the other sections.
3693 * The [RoutingPolicyRule] section of .network file gained a new
3694 SuppressInterfaceGroup= setting.
3696 * The IgnoreCarrierLoss= setting in the [Network] section of .network
3697 files now allows a duration to be specified, controlling how long to
3698 wait before reacting to carrier loss.
3700 * The [DHCPServer] section of .network file gained a new Router=
3701 setting to specify the router address.
3703 * The [CAKE] section of .network files gained various new settings
3704 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
3705 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
3706 and UseRawPacketSize= for configuring CAKE.
3708 * systemd-networkd now ships with new default .network files:
3709 80-container-vb.network which matches host-side network bridge device
3710 created by systemd-nspawn's --network-bridge or --network-zone
3711 switch, and 80-6rd-tunnel.network which matches automatically created
3712 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
3714 * systemd-networkd's handling of Endpoint= resolution for WireGuard
3715 interfaces has been improved.
3717 * systemd-networkd will now automatically configure routes to addresses
3718 specified in AllowedIPs=. This feature can be controlled via
3719 RouteTable= and RouteMetric= settings in [WireGuard] or
3720 [WireGuardPeer] sections.
3722 * systemd-networkd will now once again automatically generate persistent
3723 MAC addresses for batadv and bridge interfaces. Users can disable this
3724 by using MACAddress=none in .netdev files.
3726 * systemd-networkd and systemd-udevd now support IP over InfiniBand
3727 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
3728 systemd.netdev files gained the [IPoIB] section.
3730 * systemd-networkd and systemd-udevd now support net.ifname_policy=
3731 option on the kernel command-line. This is implemented through the
3732 systemd-network-generator service that automatically generates
3733 appropriate .link, .network, and .netdev files.
3735 * The various systemd-udevd "ethtool" buffer settings now understand
3736 the special value "max" to configure the buffers to the maximum the
3739 * systemd-udevd's .link files may now configure a large variety of
3740 NIC coalescing settings, plus more hardware offload settings.
3742 * .link files gained a new WakeOnLanPassword= setting in the [Link]
3743 section that allows to specify a WoL "SecureOn" password on hardware
3746 * systemd-nspawn's --setenv= switch now supports an additional syntax:
3747 if only a variable name is specified (i.e. without being suffixed by
3748 a '=' character and a value) the current value of the environment
3749 variable is propagated to the container. e.g. --setenv=FOO will
3750 lookup the current value of $FOO in the environment, and pass it down
3751 to the container. Similar behavior has been added to homectl's,
3752 machinectl's and systemd-run's --setenv= switch.
3754 * systemd-nspawn gained a new switch --suppress-sync= which may be used
3755 to optionally suppress the effect of the sync()/fsync()/fdatasync()
3756 system calls for the container payload. This is useful for build
3757 system environments where safety against abnormal system shutdown is
3758 not essential as all build artifacts can be regenerated any time, but
3759 the performance win is beneficial.
3761 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
3762 same value that PID 1 uses for most forked off processes.
3764 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
3765 uidmap/nouidmap options as last parameter. If "uidmap" is used the
3766 bind mounts are created with UID mapping taking place that ensures
3767 the host's file ownerships are mapped 1:1 to container file
3768 ownerships, even if user namespacing is used. This way
3769 files/directories bound into containers will no longer show up as
3770 owned by the nobody user as they typically did if no special care was
3771 taken to shift them manually.
3773 * When discovering Windows installations sd-boot will now attempt to
3774 show the Windows version.
3776 * The color scheme to use in sd-boot may now be configured at
3779 * sd-boot gained the ability to change screen resolution during
3780 boot-time, by hitting the "r" key. This will cycle through available
3781 resolutions and save the last selection.
3783 * sd-boot learnt a new hotkey "f". When pressed the system will enter
3784 firmware setup. This is useful in environments where it is difficult
3785 to hit the right keys early enough to enter the firmware, and works
3786 on any firmware regardless which key it natively uses.
3788 * sd-boot gained support for automatically booting into the menu item
3789 selected on the last boot (using the "@saved" identifier for menu
3792 * sd-boot gained support for automatically loading all EFI drivers
3793 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
3794 Partition (ESP). These drivers are loaded before the menu entries are
3795 loaded. This is useful e.g. to load additional file system drivers
3796 for the XBOOTLDR partition.
3798 * systemd-boot will now paint the input cursor on its own instead of
3799 relying on the firmware to do so, increasing compatibility with broken
3800 firmware that doesn't make the cursor reasonably visible.
3802 * sd-boot now embeds a .osrel PE section like we expect from Boot
3803 Loader Specification Type #2 Unified Kernels. This means sd-boot
3804 itself may be used in place of a Type #2 Unified Kernel. This is
3805 useful for debugging purposes as it allows chain-loading one a
3806 (development) sd-boot instance from another.
3808 * sd-boot now supports a new "devicetree" field in Boot Loader
3809 Specification Type #1 entries: if configured the specified device
3810 tree file is installed before the kernel is invoked. This is useful
3811 for installing/applying new devicetree files without updating the
3814 * Similarly, sd-stub now can read devicetree data from a PE section
3815 ".dtb" and apply it before invoking the kernel.
3817 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
3818 gained the ability to pick up credentials and sysext files, wrap them
3819 in a cpio archive, and pass as an additional initrd to the invoked
3820 Linux kernel, in effect placing those files in the /.extra/ directory
3821 of the initrd environment. This is useful to implement trusted initrd
3822 environments which are fully authenticated but still can be extended
3823 (via sysexts) and parameterized (via encrypted/authenticated
3824 credentials, see above).
3826 Credentials can be located next to the kernel image file (credentials
3827 specific to a single boot entry), or in one of the shared directories
3828 (credentials applicable to multiple boot entries).
3830 * sd-stub now comes with a full man page, that explains its feature set
3831 and how to combine a kernel image, an initrd and the stub to build a
3832 complete EFI unified kernel image, implementing Boot Loader
3833 Specification Type #2.
3835 * sd-stub may now provide the initrd to the executed kernel via the
3836 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
3837 non-x86 architectures.
3839 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
3840 may be used to set the boot menu time-out of the boot loader (for all
3841 or just the subsequent boot).
3843 * bootctl and kernel-install will now read variables
3844 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
3845 /etc/kernel/install.conf. When set, it specifies the layout to use
3846 for installation directories on the boot partition, so that tools
3847 don't need to guess it based on the already-existing directories. The
3848 only value that is defined natively is "bls", corresponding to the
3850 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
3851 kernel-install that implement a different layout can declare other
3852 values for this variable.
3854 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
3855 assumption that if the user installed sd-boot to the ESP, they intend
3856 to use the entry layout understood by sd-boot. It'll also write
3857 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
3858 (and it wasn't specified in the config file yet). Similarly,
3859 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
3860 wasn't specified in the config file yet). Effectively, those changes
3861 mean that the machine-id used for boot loader entry installation is
3862 "frozen" upon first use and becomes independent of the actual
3865 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
3866 images created for distribution ("golden images") are built with no
3867 machine-id, so that a unique machine-id can be created on the first
3868 boot. But those images may contain boot loader entries with the
3869 machine-id used during build included in paths. Using a "frozen"
3870 value allows unambiguously identifying entries that match the
3871 specific installation, while still permitting parallel installations
3874 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
3875 kernel-install to guess the installation layout. This fixes the
3876 problem where a (possibly empty) directory in the boot partition is
3877 created from a different layout causing kernel-install plugins to
3878 assume the wrong layout. A particular example of how this may happen
3879 is the grub2 package in Fedora which includes directories under /boot
3880 directly in its file list. Various other packages pull in grub2 as a
3881 dependency, so it may be installed even if unused, breaking
3882 installations that use the bls layout.
3884 * bootctl and systemd-bless-boot can now be linked statically.
3886 * systemd-sysext now optionally doesn't insist on extension-release.d/
3887 files being placed in the image under the image's file name. If the
3888 file system xattr user.extension-release.strict is set on the
3889 extension release file, it is accepted regardless of its name. This
3890 relaxes security restrictions a bit, as system extension may be
3891 attached under a wrong name this way.
3893 * udevadm's test-builtin command learnt a new --action= switch for
3894 testing the built-in with the specified action (in place of the
3897 * udevadm info gained new switches --property=/--value for showing only
3898 specific udev properties/values instead of all.
3900 * A new hwdb database has been added that contains matches for various
3901 types of signal analyzers (protocol analyzers, logic analyzers,
3902 oscilloscopes, multimeters, bench power supplies, etc.) that should
3903 be accessible to regular users.
3905 * A new hwdb database entry has been added that carries information
3906 about types of cameras (regular or infrared), and in which direction
3907 they point (front or back).
3909 * A new rule to allow console users access to rfkill by default has been
3912 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
3913 now also owned by the system group "sgx".
3915 * A new build-time meson option "extra-net-naming-schemes=" has been
3916 added to define additional naming schemes for udev's network
3917 interface naming logic. This is useful for enterprise distributions
3918 and similar which want to pin the schemes of certain distribution
3919 releases under a specific name and previously had to patch the
3920 sources to introduce new named schemes.
3922 * The predictable naming logic for network interfaces has been extended
3923 to generate stable names from Xen netfront device information.
3925 * hostnamed's chassis property can now be sourced from chassis-type
3926 field encoded in devicetree (in addition to the existing DMI
3929 * systemd-cgls now optionally displays cgroup IDs and extended
3930 attributes for each cgroup. (Controllable via the new --xattr= +
3931 --cgroup-id= switches.)
3933 * coredumpctl gained a new --all switch for operating on all
3934 Journal files instead of just the local ones.
3936 * systemd-coredump will now use libdw/libelf via dlopen() rather than
3937 directly linking, allowing users to easily opt-out of backtrace/metadata
3938 analysis of core files, and reduce image sizes when this is not needed.
3940 * systemd-coredump will now analyze core files with libdw/libelf in a
3941 forked, sandboxed process.
3943 * systemd-homed will now try to unmount an activate home area in
3944 regular intervals once the user logged out fully. Previously this was
3945 attempted exactly once but if the home directory was busy for some
3946 reason it was not tried again.
3948 * systemd-homed's LUKS2 home area backend will now create a BSD file
3949 system lock on the image file while the home area is active
3950 (i.e. mounted). If a home area is found to be locked, logins are
3951 politely refused. This should improve behavior when using home areas
3952 images that are accessible via the network from multiple clients, and
3953 reduce the chance of accidental file system corruption in that case.
3955 * Optionally, systemd-homed will now drop the kernel buffer cache once
3956 a user has fully logged out, configurable via the new --drop-caches=
3959 * systemd-homed now makes use of UID mapped mounts for the home areas.
3960 If the kernel and used file system support it, files are now
3961 internally owned by the "nobody" user (i.e. the user typically used
3962 for indicating "this ownership is not mapped"), and dynamically
3963 mapped to the UID used locally on the system via the UID mapping
3964 mount logic of recent kernels. This makes migrating home areas
3965 between different systems cheaper because recursively chown()ing file
3966 system trees is no longer necessary.
3968 * systemd-homed's CIFS backend now optionally supports CIFS service
3969 names with a directory suffix, in order to place home directories in
3970 a subdirectory of a CIFS share, instead of the top-level directory.
3972 * systemd-homed's CIFS backend gained support for specifying additional
3973 mount options in the JSON user record (cifsExtraMountOptions field,
3974 and --cifs-extra-mount-options= homectl switch). This is for example
3975 useful for configuring mount options such as "noserverino" that some
3976 SMB3 services require (use that to run a homed home directory from a
3977 FritzBox SMB3 share this way).
3979 * systemd-homed will now default to btrfs' zstd compression for home
3980 areas. This is inspired by Fedora's recent decision to switch to zstd
3983 * Additional mount options to use when mounting the file system of
3984 LUKS2 volumes in systemd-homed has been added. Via the
3985 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
3986 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
3987 systemd-homed or via the luksExtraMountOptions user record JSON
3988 property. (Exposed via homectl --luks-extra-mount-options)
3990 * homectl's resize command now takes the special size specifications
3991 "min" and "max" to shrink/grow the home area to the minimum/maximum
3992 size possible, taking disk usage/space constraints and file system
3993 limitations into account. Resizing is now generally graceful: the
3994 logic will try to get as close to the specified size as possible, but
3995 not consider it a failure if the request couldn't be fulfilled
3998 * systemd-homed gained the ability to automatically shrink home areas
3999 on logout to their minimal size and grow them again on next
4000 login. This ensures that while inactive, a home area only takes up
4001 the minimal space necessary, but once activated, it provides
4002 sufficient space for the user's needs. This behavior is only
4003 supported if btrfs is used as file system inside the home area
4004 (because only for btrfs online growing/shrinking is implemented in
4005 the kernel). This behavior is now enabled by default, but may be
4006 controlled via the new --auto-resize-mode= setting of homectl.
4008 * systemd-homed gained support for automatically re-balancing free disk
4009 space among active home areas, in case the LUKS2 backends are used,
4010 and no explicit disk size was requested. This way disk space is
4011 automatically managed and home areas resized in regular intervals and
4012 manual resizing when disk space becomes scarce should not be
4013 necessary anymore. This behavior is only supported if btrfs is used
4014 within the home areas (as only then online shrinking and growing is
4015 supported), and may be configured via the new rebalanceWeight JSON
4016 user record field (as exposed via the new --rebalance-weight= homectl
4017 setting). Re-balancing is mostly automatic, but can also be requested
4018 explicitly via "homectl rebalance", which is synchronous, and thus
4019 may be used to wait until the rebalance run is complete.
4021 * userdbctl gained a --json= switch for configured the JSON formatting
4022 to use when outputting user or group records.
4024 * userdbctl gained a new --multiplexer= switch for explicitly
4025 configuring whether to use the systemd-userdbd server side user
4026 record resolution logic.
4028 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
4029 for chaining up another command to execute after completing the
4030 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
4031 configuration of a single command to invoke, this maybe used to
4032 invoke multiple: first userdbctl's own implementation, and then any
4033 other also configured in the command line.
4035 * The sd-event API gained a new function sd_event_add_inotify_fd() that
4036 is similar to sd_event_add_inotify() but accepts a file descriptor
4037 instead of a path in the file system for referencing the inode to
4040 * The sd-event API gained a new function
4041 sd_event_source_set_ratelimit_expire_callback() that may be used to
4042 define a callback function that is called whenever an event source
4043 leaves the rate limiting phase.
4045 * New documentation has been added explaining which steps are necessary
4046 to port systemd to a new architecture:
4048 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
4050 * The x-systemd.makefs option in /etc/fstab now explicitly supports
4051 ext2, ext3, and f2fs file systems.
4053 * Mount units and units generated from /etc/fstab entries with 'noauto'
4054 are now ordered the same as other units. Effectively, they will be
4055 started earlier (if something actually pulled them in) and stopped
4056 later, similarly to normal mount units that are part of
4057 fs-local.target. This change should be invisible to users, but
4058 should prevent those units from being stopped too early during
4061 * The systemd-getty-generator now honors a new kernel command line
4062 argument systemd.getty_auto= and a new environment variable
4063 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
4064 example useful to turn off gettys inside of containers or similar
4067 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
4068 (in addition to 127.0.0.53, as before). If DNS requests are sent to
4069 this address they are propagated in "bypass" mode only, i.e. are
4070 almost not processed locally, but mostly forwarded as-is to the
4071 current upstream DNS servers. This provides a stable DNS server
4072 address that proxies all requests dynamically to the right upstream
4073 DNS servers even if these dynamically change. This stub does not do
4074 mDNS/LLMNR resolution. However, it will translate look-ups to
4075 DNS-over-TLS if necessary. This new stub is particularly useful in
4076 container/VM environments, or for tethering setups: use DNAT to
4077 redirect traffic to any IP address to this stub.
4079 * systemd-importd now honors new environment variables
4080 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
4081 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
4082 generation, btrfs quota setup and disk synchronization.
4084 * systemd-importd and systemd-resolved can now be optionally built with
4085 OpenSSL instead of libgcrypt.
4087 * systemd-repart no longer requires OpenSSL.
4089 * systemd-sysusers will no longer create the redundant 'nobody' group
4090 by default, as the 'nobody' user is already created with an
4091 appropriate primary group.
4093 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
4095 * systemctl show-environment gained support for --output=json.
4097 * pam_systemd will now first try to use the X11 abstract socket, and
4098 fallback to the socket file in /tmp/.X11-unix/ only if that does not
4101 * systemd-journald will no longer go back to volatile storage
4102 regardless of configuration when its unit is restarted.
4104 * Initial support for the LoongArch architecture has been added (system
4105 call lists, GPT partition table UUIDs, etc).
4107 * systemd-journald's own logging messages are now also logged to the
4108 journal itself when systemd-journald logs to /dev/kmsg.
4110 * systemd-journald now re-enables COW for archived journal files on
4111 filesystems that support COW. One benefit of this change is that
4112 archived journal files will now get compressed on btrfs filesystems
4113 that have compression enabled.
4115 * systemd-journald now deduplicates fields in a single log message
4116 before adding it to the journal. In archived journal files, it will
4117 also punch holes for unused parts and truncate the file as
4118 appropriate, leading to reductions in disk usage.
4120 * journalctl --verify was extended with more informative error
4123 * More of sd-journal's functions are now resistant against journal file
4126 * The shutdown command learnt a new option --show, to display the
4129 * A LICENSES/ directory is now included in the git tree. It contains a
4130 README.md file that explains the licenses used by source files in
4131 this repository. It also contains the text of all applicable
4132 licenses as they appear on spdx.org.
4134 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
4135 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
4136 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
4137 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
4138 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
4139 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
4140 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
4141 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
4142 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
4143 Christian Brauner, Christian Göttsche, Christian Wehrli,
4144 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
4145 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
4146 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
4147 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
4148 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
4149 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
4150 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
4151 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
4152 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
4153 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
4154 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
4155 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
4156 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
4157 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
4158 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
4159 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
4160 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
4161 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
4162 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
4163 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
4164 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
4165 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
4166 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
4167 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
4168 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
4169 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
4170 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
4171 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
4172 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
4173 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
4174 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
4175 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
4176 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
4177 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
4178 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
4179 Дамјан Георгиевски, наб
4181 — Warsaw, 2021-12-23
4185 * When operating on disk images via the --image= switch of various
4186 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
4187 no 'root=' parameter on the kernel command line, and multiple
4188 suitable root or /usr/ partitions exist in the image, then a simple
4189 comparison inspired by strverscmp() is done on the GPT partition
4190 label, and the newest partition is picked. This permits a simple and
4191 generic whole-file-system A/B update logic where new operating system
4192 versions are dropped into partitions whose label is then updated with
4193 a matching version identifier.
4195 * systemd-sysusers now supports querying the passwords to set for the
4196 users it creates via the "credentials" logic introduced in v247: the
4197 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
4198 credentials are consulted for the password to use (either in UNIX
4199 hashed form, or literally). By default these credentials are inherited
4200 down from PID1 (which in turn imports it from a container manager if
4201 there is one). This permits easy configuration of user passwords
4202 during first boot. Example:
4204 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
4206 Note that systemd-sysusers operates in purely additive mode: it
4207 executes no operation if the declared users already exist, and hence
4208 doesn't set any passwords as effect of the command line above if the
4209 specified root user exists already in the image. (Note that
4210 --volatile=yes ensures it doesn't, though.)
4212 * systemd-firstboot now also supports querying various system
4213 parameters via the credential subsystems. Thus, as above this may be
4214 used to initialize important system parameters on first boot of
4215 previously unprovisioned images (i.e. images with a mostly empty
4218 * PID 1 may now show both the unit name and the unit description
4219 strings in its status output during boot. This may be configured with
4220 StatusUnitFormat=combined in system.conf or
4221 systemd.status-unit-format=combined on the kernel command line.
4223 * The systemd-machine-id-setup tool now supports a --image= switch for
4224 provisioning a machine ID file into an OS disk image, similar to how
4225 --root= operates on an OS file tree. This matches the existing switch
4226 of the same name for systemd-tmpfiles, systemd-firstboot, and
4227 systemd-sysusers tools.
4229 * Similarly, systemd-repart gained support for the --image= switch too.
4230 In combination with the existing --size= option, this makes the tool
4231 particularly useful for easily growing disk images in a single
4232 invocation, following the declarative rules included in the image
4235 * systemd-repart's partition configuration files gained support for a
4236 new switch MakeDirectories= which may be used to create arbitrary
4237 directories inside file systems that are created, before registering
4238 them in the partition table. This is useful in particular for root
4239 partitions to create mount point directories for other partitions
4240 included in the image. For example, a disk image that contains a
4241 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
4242 create /home/ and /var/ as empty directories in the root file system
4243 on its creation, so that the resulting image can be mounted
4244 immediately, even in read-only mode.
4246 * systemd-repart's CopyBlocks= setting gained support for the special
4247 value "auto". If used, a suitable matching partition on the booted OS
4248 is found as source to copy blocks from. This is useful when
4249 implementing replicating installers, that are booted from one medium
4250 and then stream their own root partition onto the target medium.
4252 * systemd-repart's partition configuration files gained support for a
4253 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
4254 GPT partition flags for the created partitions: this is useful for
4255 marking newly created partitions as read-only, or as not being
4256 subject for automatic mounting from creation on.
4258 * The /etc/os-release file has been extended with two new (optional)
4259 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
4260 information for OS images that are updated comprehensively and
4261 atomically as one image. Two new specifiers %M, %A now resolve to
4262 these two fields in the various configuration options that resolve
4265 * portablectl gained a new switch --extension= for enabling portable
4266 service images with extensions that follow the extension image
4267 concept introduced with v248, and thus allows layering multiple
4268 images when setting up the root filesystem of the service.
4270 * systemd-coredump will now extract ELF build-id information from
4271 processes dumping core and include it in the coredump report.
4272 Moreover, it will look for ELF .note.package sections with
4273 distribution packaging meta-information about the crashing process.
4274 This is useful to directly embed the rpm or deb (or any other)
4275 package name and version in ELF files, making it easy to match
4276 coredump reports with the specific package for which the software was
4277 compiled. This is particularly useful on environments with ELF files
4278 from multiple vendors, different distributions and versions, as is
4279 common today in our containerized and sand-boxed world. For further
4282 https://systemd.io/COREDUMP_PACKAGE_METADATA
4284 * A new udev hardware database has been added for FireWire devices
4287 * The "net_id" built-in of udev has been updated with three
4288 backwards-incompatible changes:
4290 - PCI hotplug slot names on s390 systems are now parsed as
4291 hexadecimal numbers. They were incorrectly parsed as decimal
4292 previously, or ignored if the name was not a valid decimal
4295 - PCI onboard indices up to 65535 are allowed. Previously, numbers
4296 above 16383 were rejected. This primarily impacts s390 systems,
4297 where values up to 65535 are used.
4299 - Invalid characters in interface names are replaced with "_".
4301 The new version of the net naming scheme is "v249". The previous
4302 scheme can be selected via the "net.naming_scheme=v247" kernel
4303 command line parameter.
4305 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
4306 NULL bus object, for which they will return false. Or in other words,
4307 an unallocated bus connection is neither ready nor open.
4309 * The sd-device API acquired a new API function
4310 sd_device_get_usec_initialized() that returns the monotonic time when
4311 the udev device first appeared in the database.
4313 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
4314 sd_device_get_trigger_uuid(). The former is similar to
4315 sd_device_trigger() but returns a randomly generated UUID that is
4316 associated with the synthetic uevent generated by the call. This UUID
4317 may be read from the sd_device object a monitor eventually receives,
4318 via the sd_device_get_trigger_uuid(). This interface requires kernel
4319 4.13 or above to work, and allows tracking a synthetic uevent through
4320 the entire device management stack. The "udevadm trigger --settle"
4321 logic has been updated to make use of this concept if available to
4322 wait precisely for the uevents it generates. "udevadm trigger" also
4323 gained a new parameter --uuid that prints the UUID for each generated
4326 * sd-device also gained new APIs sd_device_new_from_ifname() and
4327 sd_device_new_from_ifindex() for allocating an sd-device object for
4328 the specified network interface. The former accepts an interface name
4329 (either a primary or an alternative name), the latter an interface
4332 * The native Journal protocol has been documented. Clients may talk
4333 this as alternative to the classic BSD syslog protocol for locally
4334 delivering log records to the Journal. The protocol has been stable
4335 for a long time and in fact been implemented already in a variety
4336 of alternative client libraries. This documentation makes the support
4339 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
4341 * A new BPFProgram= setting has been added to service files. It may be
4342 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
4343 file, or a bind mount or symlink to one. This may be used to upload
4344 and manage BPF programs externally and then hook arbitrary systemd
4347 * The "home.arpa" domain that has been officially declared as the
4348 choice for domain for local home networks per RFC 8375 has been added
4349 to the default NTA list of resolved, since DNSSEC is generally not
4350 available on private domains.
4352 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
4354 * A new ManageForeignRoutingPolicyRules= setting has been added to
4355 .network files which may be used to exclude foreign-created routing
4356 policy rules from systemd-networkd management.
4358 * systemd-network-wait-online gained two new switches -4 and -6 that
4359 may be used to tweak whether to wait for only IPv4 or only IPv6
4362 * .network files gained a new RequiredFamilyForOnline= setting to
4363 fine-tune whether to require an IPv4 or IPv6 address in order to
4364 consider an interface "online".
4366 * networkctl will now show an over-all "online" state in the per-link
4369 * In .network files a new OutgoingInterface= setting has been added to
4370 specify the output interface in bridge FDB setups.
4372 * In .network files the Multipath group ID may now be configured for
4373 [NextHop] entries, via the new Group= setting.
4375 * The DHCP server logic configured in .network files gained a new
4376 setting RelayTarget= that turns the server into a DHCP server relay.
4377 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
4378 to further tweak the DHCP relay behaviour.
4380 * The DHCP server logic also gained a new ServerAddress= setting in
4381 .network files that explicitly specifies the server IP address to
4382 use. If not specified, the address is determined automatically, as
4385 * The DHCP server logic in systemd-networkd gained support for static
4386 DHCP leases, configurable via the [DHCPServerStaticLease]
4387 section. This allows explicitly mapping specific MAC addresses to
4388 fixed IP addresses and vice versa.
4390 * The RestrictAddressFamilies= setting in service files now supports a
4391 new special value "none". If specified sockets of all address
4392 families will be made unavailable to services configured that way.
4394 * systemd-fstab-generator and systemd-repart have been updated to
4395 support booting from disks that carry only a /usr/ partition but no
4396 root partition yet, and where systemd-repart can add it in on the
4397 first boot. This is useful for implementing systems that ship with a
4398 single /usr/ file system, and whose root file system shall be set up
4399 and formatted on a LUKS-encrypted volume whose key is generated
4400 locally (and possibly enrolled in the TPM) during the first boot.
4402 * The [Address] section of .network files now accepts a new
4403 RouteMetric= setting that configures the routing metric to use for
4404 the prefix route created as effect of the address configuration.
4405 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
4406 gained matching settings for their prefix routes. (The option of the
4407 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
4408 it conceptually belongs there; the old option is still understood for
4411 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
4414 * A new udev property ID_NET_DHCP_BROADCAST on network interface
4415 devices is now honoured by systemd-networkd, controlling whether to
4416 issue DHCP offers via broadcasting. This is used to ensure that s390
4417 layer 3 network interfaces work out-of-the-box with systemd-networkd.
4419 * nss-myhostname and systemd-resolved will now synthesize address
4420 records for a new special hostname "_outbound". The name will always
4421 resolve to the local IP addresses most likely used for outbound
4422 connections towards the default routes. On multi-homed hosts this is
4423 useful to have a stable handle referring to "the" local IP address
4424 that matters most, to the point where this is defined.
4426 * The Discoverable Partition Specification has been updated with a new
4427 GPT partition flag "grow-file-system" defined for its partition
4428 types. Whenever partitions with this flag set are automatically
4429 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
4430 of systemd-nspawn or other tools; and as opposed to explicit mounting
4431 via /etc/fstab), the file system within the partition is
4432 automatically grown to the full size of the partition. If the file
4433 system size already matches the partition size this flag has no
4434 effect. Previously, this functionality has been available via the
4435 explicit x-systemd.growfs mount option, and this new flag extends
4436 this to automatically discovered mounts. A new GrowFileSystem=
4437 setting has been added to systemd-repart drop-in files that allows
4438 configuring this partition flag. This new flag defaults to on for
4439 partitions automatically created by systemd-repart, except if they
4440 are marked read-only. See the specification for further details:
4442 https://systemd.io/DISCOVERABLE_PARTITIONS
4444 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
4445 section. If enabled (which is the default), and an NTP server address
4446 is acquired through a DHCP lease on this interface an explicit route
4447 to this address is created on this interface to ensure that NTP
4448 traffic to the NTP server acquired on an interface is also routed
4449 through that interface. The pre-existing RoutesToDNS= setting that
4450 implements the same for DNS servers is now enabled by default.
4452 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
4453 been added that may be used to restrict the network interfaces
4454 sockets created by the service may be bound to. This is implemented
4457 * A new ConditionFirmware= setting has been added to unit files to
4458 conditionalize on certain firmware features. At the moment it may
4459 check whether running on a UEFI system, a device.tree system, or if
4460 the system is compatible with some specified device-tree feature.
4462 * A new ConditionOSRelease= setting has been added to unit files to
4463 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
4464 operators may be used to check if some field has some specific value
4465 or do an alphanumerical comparison. Equality comparisons are useful
4466 for fields like ID, but relative comparisons for fields like
4467 VERSION_ID or IMAGE_VERSION.
4469 * hostnamed gained a new Describe() D-Bus method that returns a JSON
4470 serialization of the host data it exposes. This is exposed via
4471 "hostnamectl --json=" to acquire a host identity description in JSON.
4472 It's our intention to add a similar features to most services and
4473 objects systemd manages, in order to simplify integration with
4474 program code that can consume JSON.
4476 * Similarly, networkd gained a Describe() method on its Manager and
4477 Link bus objects. This is exposed via "networkctl --json=".
4479 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
4480 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
4481 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
4482 that is used both to get the value (when no argument is given), and
4483 to set the value (when an argument is specified). The old names
4484 continue to be supported for compatibility.
4486 * systemd-detect-virt and ConditionVirtualization= are now able to
4487 correctly identify Amazon EC2 environments.
4489 * The LogLevelMax= setting of unit files now applies not only to log
4490 messages generated *by* the service, but also to log messages
4491 generated *about* the service by PID 1. To suppress logs concerning a
4492 specific service comprehensively, set this option to a high log
4495 * bootctl gained support for a new --make-machine-id-directory= switch
4496 that allows precise control on whether to create the top-level
4497 per-machine directory in the boot partition that typically contains
4498 Type 1 boot loader entries.
4500 * During build SBAT data to include in the systemd-boot EFI PE binaries
4501 may be specified now.
4503 * /etc/crypttab learnt a new option "headless". If specified any
4504 requests to query the user interactively for passwords or PINs will
4505 be skipped. This is useful on systems that are headless, i.e. where
4506 an interactive user is generally not present.
4508 * /etc/crypttab also learnt a new option "password-echo=" that allows
4509 configuring whether the encryption password prompt shall echo the
4510 typed password and if so, do so literally or via asterisks. (The
4511 default is the same behaviour as before: provide echo feedback via
4514 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
4515 systemd-homed has been updated to allow explicit configuration of the
4516 "user presence" and "user verification" checks, as well as whether a
4517 PIN is required for authentication, via the new switches
4518 --fido2-with-user-presence=, --fido2-with-user-verification=,
4519 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
4520 features are available, and may be enabled or disabled depends on the
4523 * systemd-nspawn's --private-user= switch now accepts the special value
4524 "identity" which configures a user namespacing environment with an
4525 identity mapping of 65535 UIDs. This means the container UID 0 is
4526 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
4527 this doesn't appear to be useful, however it does reduce the attack
4528 surface a bit, since the resulting container will possess process
4529 capabilities only within its namespace and not on the host.
4531 * systemd-nspawn's --private-user-chown switch has been replaced by a
4532 more generic --private-user-ownership= switch that accepts one of
4533 three values: "chown" is equivalent to the old --private-user-chown,
4534 and "off" is equivalent to the absence of the old switch. The value
4535 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
4536 of files and directories of the underlying image to the chosen UID
4537 range for the container. "auto" is equivalent to "map" if UID mapping
4538 mount are supported, otherwise it is equivalent to "chown". The short
4539 -U switch systemd-nspawn now implies --private-user-ownership=auto
4540 instead of the old --private-user-chown. Effectively this means: if
4541 the backing file system supports UID mapping mounts the feature is
4542 now used by default if -U is used. Generally, it's a good idea to use
4543 UID mapping mounts instead of recursive chown()ing, since it allows
4544 running containers off immutable images (since no modifications of
4545 the images need to take place), and share images between multiple
4546 instances. Moreover, the recursive chown()ing operation is slow and
4547 can be avoided. Conceptually it's also a good thing if transient UID
4548 range uses do not leak into persistent file ownership anymore. TLDR:
4549 finally, the last major drawback of user namespacing has been
4550 removed, and -U should always be used (unless you use btrfs, where
4551 UID mapped mounts do not exist; or your container actually needs
4552 privileges on the host).
4554 * nss-systemd now synthesizes user and group shadow records in addition
4555 to the main user and group records. Thus, hashed passwords managed by
4556 systemd-homed are now accessible via the shadow database.
4558 * The userdb logic (and thus nss-systemd, and so on) now read
4559 additional user/group definitions in JSON format from the drop-in
4560 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
4561 /usr/lib/userdb/. This is a simple and powerful mechanism for making
4562 additional users available to the system, with full integration into
4563 NSS including the shadow databases. Since the full JSON user/group
4564 record format is supported this may also be used to define users with
4565 resource management settings and other runtime settings that
4566 pam_systemd and systemd-logind enforce at login.
4568 * The userdbctl tool gained two new switches --with-dropin= and
4569 --with-varlink= which can be used to fine-tune the sources used for
4570 user database lookups.
4572 * systemd-nspawn gained a new switch --bind-user= for binding a host
4573 user account into the container. This does three things: the user's
4574 home directory is bind mounted from the host into the container,
4575 below the /run/userdb/home/ hierarchy. A free UID is picked in the
4576 container, and a user namespacing UID mapping to the host user's UID
4577 installed. And finally, a minimal JSON user and group record (along
4578 with its hashed password) is dropped into /run/host/userdb/. These
4579 records are picked up automatically by the userdb drop-in logic
4580 describe above, and allow the user to login with the same password as
4581 on the host. Effectively this means: if host and container run new
4582 enough systemd versions making a host user available to the container
4583 is trivially simple.
4585 * systemd-journal-gatewayd now supports the switches --user, --system,
4586 --merge, --file= that are equivalent to the same switches of
4587 journalctl, and permit exposing only the specified subset of the
4590 * The OnFailure= dependency between units is now augmented with a
4591 implicit reverse dependency OnFailureOf= (this new dependency cannot
4592 be configured directly it's only created as effect of an OnFailure=
4593 dependency in the reverse order — it's visible in "systemctl show"
4594 however). Similar, Slice= now has an reverse dependency SliceOf=,
4595 that is also not configurable directly, but useful to determine all
4596 units that are members of a slice.
4598 * A pair of new dependency types between units PropagatesStopTo= +
4599 StopPropagatedFrom= has been added, that allows propagation of unit
4600 stop events between two units. It operates similar to the existing
4601 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
4603 * A new dependency type OnSuccess= has been added (plus the reverse
4604 dependency OnSuccessOf=, which cannot be configured directly, but
4605 exists only as effect of the reverse OnSuccess=). It is similar to
4606 OnFailure=, but triggers in the opposite case: when a service exits
4607 cleanly. This allows "chaining up" of services where one or more
4608 services are started once another service has successfully completed.
4610 * A new dependency type Upholds= has been added (plus the reverse
4611 dependency UpheldBy=, which cannot be configured directly, but exists
4612 only as effect of Upholds=). This dependency type is a stronger form
4613 of Wants=: if a unit has an UpHolds= dependency on some other unit
4614 and the former is active then the latter is started whenever it is
4615 found inactive (and no job is queued for it). This is an alternative
4616 to Restart= inside service units, but less configurable, and the
4617 request to uphold a unit is not encoded in the unit itself but in
4618 another unit that intends to uphold it.
4620 * The systemd-ask-password tool now also supports reading passwords
4621 from the credentials subsystem, via the new --credential= switch.
4623 * The systemd-ask-password tool learnt a new switch --emoji= which may
4624 be used to explicit control whether the lock and key emoji (🔐) is
4625 shown in the password prompt on suitable TTYs.
4627 * The --echo switch of systemd-ask-password now optionally takes a
4628 parameter that controls character echo. It may either show asterisks
4629 (default, as before), turn echo off entirely, or echo the typed
4630 characters literally.
4632 * The systemd-ask-password tool also gained a new -n switch for
4633 suppressing output of a trailing newline character when writing the
4634 acquired password to standard output, similar to /bin/echo's -n
4637 * New documentation has been added that describes the organization of
4638 the systemd source code tree:
4640 https://systemd.io/ARCHITECTURE
4642 * Units using ConditionNeedsUpdate= will no longer be activated in
4645 * It is now possible to list a template unit in the WantedBy= or
4646 RequiredBy= settings of the [Install] section of another template
4647 unit, which will be instantiated using the same instance name.
4649 * A new MemoryAvailable property is available for units. If the unit,
4650 or the slices it is part of, have a memory limit set via MemoryMax=/
4651 MemoryHigh=, MemoryAvailable will indicate how much more memory the
4652 unit can claim before hitting the limits.
4654 * systemd-coredump will now try to stay below the cgroup memory limit
4655 placed on itself or one of the slices it runs under, if the storage
4656 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
4657 since files written on such filesystems count toward the cgroup memory
4658 limit. If there is not enough available memory in such cases to store
4659 the core file uncompressed, systemd-coredump will skip to compressed
4660 storage directly (if enabled) and it will avoid analyzing the core file
4661 to print backtrace and metadata in the journal.
4663 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
4664 of a path matches the configured expectations, and remove it if not.
4666 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
4667 specify which of the several available filesystem timestamps (access
4668 time, birth time, change time, modification time) to look at when
4669 deciding whether a path has aged enough to be cleaned.
4671 * A new IPv6StableSecretAddress= setting has been added to .network
4672 files, which takes an IPv6 address to use as secret for IPv6 address
4675 * The [DHCPServer] logic in .network files gained support for a new
4676 UplinkInterface= setting that permits configuration of the uplink
4677 interface name to propagate DHCP lease information from.
4679 * The WakeOnLan= setting in .link files now accepts a list of flags
4680 instead of a single one, to configure multiple wake-on-LAN policies.
4682 * User-space defined tracepoints (USDT) have been added to udev at
4683 strategic locations. This is useful for tracing udev behaviour and
4684 performance with bpftrace and similar tools.
4686 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
4687 setting a network timeout time.
4689 * If a system service is running in a new mount namespace (RootDirectory=
4690 and friends), all file systems will be mounted with MS_NOSUID by
4691 default, unless the system is running with SELinux enabled.
4693 * When enumerating time zones the timedatectl tool will now consult the
4694 'tzdata.zi' file shipped by the IANA time zone database package, in
4695 addition to 'zone1970.tab', as before. This makes sure time zone
4696 aliases are now correctly supported. Some distributions so far did
4697 not install this additional file, most do however. If you
4698 distribution does not install it yet, it might make sense to change
4701 * Intel HID rfkill event is no longer masked, since it's the only
4702 source of rfkill event on newer HP laptops. To have both backward and
4703 forward compatibility, userspace daemon needs to debounce duplicated
4704 events in a short time window.
4706 Contributions from: Aakash Singh, adrian5, Albert Brox,
4707 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
4708 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
4709 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
4710 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
4711 Christian Hesse, Daniel Schaefer, Dan Streetman,
4712 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
4713 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
4714 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
4715 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
4716 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
4717 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
4718 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
4719 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
4720 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
4721 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
4722 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
4723 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
4724 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
4725 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
4726 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
4727 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
4728 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
4729 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
4730 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
4731 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
4732 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
4733 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
4734 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
4735 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
4736 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
4738 — Edinburgh, 2021-07-07
4742 * A concept of system extension images is introduced. Such images may
4743 be used to extend the /usr/ and /opt/ directory hierarchies at
4744 runtime with additional files (even if the file system is read-only).
4745 When a system extension image is activated, its /usr/ and /opt/
4746 hierarchies and os-release information are combined via overlayfs
4747 with the file system hierarchy of the host OS.
4749 A new systemd-sysext tool can be used to merge, unmerge, list, and
4750 refresh system extension hierarchies. See
4751 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
4753 The systemd-sysext.service automatically merges installed system
4754 extensions during boot (before basic.target, but not in very early
4755 boot, since various file systems have to be mounted first).
4757 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
4758 supported system extension level.
4760 * A new ExtensionImages= unit setting can be used to apply the same
4761 system extension image concept from systemd-sysext to the namespaced
4762 file hierarchy of specific services, following the same rules and
4765 * Support for a new special "root=tmpfs" kernel command-line option has
4766 been added. When specified, a tmpfs is mounted on /, and mount.usr=
4767 should be used to point to the operating system implementation.
4769 * A new configuration file /etc/veritytab may be used to configure
4770 dm-verity integrity protection for block devices. Each line is in the
4771 format "volume-name data-device hash-device roothash options",
4772 similar to /etc/crypttab.
4774 * A new kernel command-line option systemd.verity.root_options= may be
4775 used to configure dm-verity behaviour for the root device.
4777 * The key file specified in /etc/crypttab (the third field) may now
4778 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
4779 acquired by connecting to that socket and reading from it. This
4780 allows the implementation of a service to provide key information
4781 dynamically, at the moment when it is needed.
4783 * When the hostname is set explicitly to "localhost", systemd-hostnamed
4784 will respect this. Previously such a setting would be mostly silently
4785 ignored. The goal is to honour configuration as specified by the
4788 * The fallback hostname that will be used by the system manager and
4789 systemd-hostnamed can now be configured in two new ways: by setting
4790 DEFAULT_HOSTNAME= in os-release(5), or by setting
4791 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
4792 also be configured during compilation. The environment variable is
4793 intended for testing and local overrides, the os-release(5) field is
4794 intended to allow customization by different variants of a
4795 distribution that share the same compiled packages.
4797 * The environment block of the manager itself may be configured through
4798 a new ManagerEnvironment= setting in system.conf or user.conf. This
4799 complements existing ways to set the environment block (the kernel
4800 command line for the system manager, the inherited environment and
4801 user@.service unit file settings for the user manager).
4803 * systemd-hostnamed now exports the default hostname and the source of
4804 the configured hostname ("static", "transient", or "default") as
4807 * systemd-hostnamed now exports the "HardwareVendor" and
4808 "HardwareModel" D-Bus properties, which are supposed to contain a
4809 pair of cleaned up, human readable strings describing the system's
4810 vendor and model. It's typically sourced from the firmware's DMI
4811 tables, but may be augmented from a new hwdb database. hostnamectl
4812 shows this in the status output.
4814 * Support has been added to systemd-cryptsetup for extracting the
4815 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
4816 metadata header. This allows the information how to open the
4817 encrypted device to be embedded directly in the device and obviates
4818 the need for configuration in an external file.
4820 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
4821 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
4822 pre-existing support for PKCS#11 security tokens).
4824 * systemd-repart may enroll encrypted partitions using TPM2
4825 hardware. This may be useful for example to create an encrypted /var
4826 partition bound to the machine on first boot.
4828 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
4829 and PKCS#11 security tokens to LUKS volumes, list and destroy
4832 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
4834 It also supports enrolling "recovery keys" and regular passphrases.
4836 * The libfido2 dependency is now based on dlopen(), so that the library
4837 is used at runtime when installed, but is not a hard runtime
4840 * systemd-cryptsetup gained support for two new options in
4841 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
4842 request synchronous processing of encryption/decryption IO.
4844 * The manager may be configured at compile time to use the fexecve()
4845 instead of the execve() system call when spawning processes. Using
4846 fexecve() closes a window between checking the security context of an
4847 executable and spawning it, but unfortunately the kernel displays
4848 stale information in the process' "comm" field, which impacts ps
4851 * The configuration option -Dcompat-gateway-hostname has been dropped.
4852 "_gateway" is now the only supported name.
4854 * The ConditionSecurity=tpm2 unit file setting may be used to check if
4855 the system has at least one TPM2 (tpmrm class) device.
4857 * A new ConditionCPUFeature= has been added that may be used to
4858 conditionalize units based on CPU features. For example,
4859 ConditionCPUFeature=rdrand will condition a unit so that it is only
4860 run when the system CPU supports the RDRAND opcode.
4862 * The existing ConditionControlGroupController= setting has been
4863 extended with two new values "v1" and "v2". "v2" means that the
4864 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
4865 hierarchy or the hybrid hierarchy are used.
4867 * A new PrivateIPC= setting on a unit file allows executed processes to
4868 be moved into a private IPC namespace, with separate System V IPC
4869 identifiers and POSIX message queues.
4871 A new IPCNamespacePath= allows the unit to be joined to an existing
4874 * The tables of system calls in seccomp filters are now automatically
4875 generated from kernel lists exported on
4876 https://fedora.juszkiewicz.com.pl/syscalls.html.
4878 The following architectures should now have complete lists:
4879 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
4880 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
4882 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
4883 on /run/ if it is not already a mount point. A writable /run/ has
4884 always been a requirement for a functioning system, but this was not
4885 guaranteed when using a read-only image.
4887 Users can always specify BindPaths= or InaccessiblePaths= as
4888 overrides, and they will take precedence. If the host's root mount
4889 point is used, there is no change in behaviour.
4891 * New bind mounts and file system image mounts may be injected into the
4892 mount namespace of a service (without restarting it). This is exposed
4893 respectively as 'systemctl bind <unit> <path>…' and
4894 'systemctl mount-image <unit> <image>…'.
4896 * The StandardOutput= and StandardError= settings can now specify files
4897 to be truncated for output (as "truncate:<path>").
4899 * The ExecPaths= and NoExecPaths= settings may be used to specify
4900 noexec for parts of the file system.
4902 * sd-bus has a new function sd_bus_open_user_machine() to open a
4903 connection to the session bus of a specific user in a local container
4904 or on the local host. This is exposed in the existing -M switch to
4905 systemctl and similar tools:
4907 systemctl --user -M lennart@foobar start foo
4909 This will connect to the user bus of a user "lennart" in container
4910 "foobar". If no container name is specified, the specified user on
4911 the host itself is connected to
4913 systemctl --user -M lennart@ start quux
4915 * sd-bus also gained a convenience function sd_bus_message_send() to
4916 simplify invocations of sd_bus_send(), taking only a single
4917 parameter: the message to send.
4919 * sd-event allows rate limits to be set on event sources, for dealing
4920 with high-priority event sources that might starve out others. See
4921 the new man page sd_event_source_set_ratelimit(3) for details.
4923 * systemd.link files gained a [Link] Promiscuous= switch, which allows
4924 the device to be raised in promiscuous mode.
4926 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
4927 number of TX and RX queues to be configured.
4929 New [Link] TransmitQueueLength= setting allows the size of the TX
4930 queue to be configured.
4932 New [Link] GenericSegmentOffloadMaxBytes= and
4933 GenericSegmentOffloadMaxSegments= allow capping the packet size and
4934 the number of segments accepted in Generic Segment Offload.
4936 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
4937 wireless routing protocol that operates on ISO/OSI Layer 2 only and
4938 uses ethernet frames to route/bridge packets. This encompasses a new
4939 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
4940 new settings in .netdev files, and a new BatmanAdvanced= setting in
4943 * systemd.network files gained a [Network] RouteTable= configuration
4944 switch to select the routing policy table.
4946 systemd.network files gained a [RoutingPolicyRule] Type=
4947 configuration switch (one of "blackhole, "unreachable", "prohibit").
4949 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
4950 RouteAllowList= settings to ignore/accept route advertisements from
4951 routers matching specified prefixes. The DenyList= setting has been
4952 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
4955 systemd.network files gained a [DHCPv6] UseAddress= setting to
4956 optionally ignore the address provided in the lease.
4958 systemd.network files gained a [DHCPv6PrefixDelegation]
4959 ManageTemporaryAddress= switch.
4961 systemd.network files gained a new ActivationPolicy= setting which
4962 allows configuring how the UP state of an interface shall be managed,
4963 i.e. whether the interface is always upped, always downed, or may be
4964 upped/downed by the user using "ip link set dev".
4966 * The default for the Broadcast= setting in .network files has slightly
4967 changed: the broadcast address will not be configured for wireguard
4970 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
4971 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
4972 configuration options for VLAN packet handling.
4974 * udev rules may now set log_level= option. This allows debug logs to
4975 be enabled for select events, e.g. just for a specific subsystem or
4976 even a single device.
4978 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
4979 DATA_PREPARED_ID properties for block devices with ISO9660 file
4982 * udev now exports decoded DMI information about installed memory slots
4983 as device properties under the /sys/class/dmi/id/ pseudo device.
4985 * /dev/ is not mounted noexec anymore. This didn't provide any
4986 significant security benefits and would conflict with the executable
4987 mappings used with /dev/sgx device nodes. The previous behaviour can
4988 be restored for individual services with NoExecPaths=/dev (or by allow-
4989 listing and excluding /dev from ExecPaths=).
4991 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
4992 and /dev/vhost-net are owned by the kvm group.
4994 * The hardware database has been extended with a list of fingerprint
4995 readers that correctly support USB auto-suspend using data from
4998 * systemd-resolved can now answer DNSSEC questions through the stub
4999 resolver interface in a way that allows local clients to do DNSSEC
5000 validation themselves. For a question with DO+CD set, it'll proxy the
5001 DNS query and respond with a mostly unmodified packet received from
5002 the upstream server.
5004 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
5005 resolved.conf. If true the service will provide caching even for DNS
5006 lookups made to an upstream DNS server on the 127.0.0.1/::1
5007 addresses. By default (and when the option is false) systemd-resolved
5008 will not cache such lookups, in order to avoid duplicate local
5009 caching, under the assumption the local upstream server caches
5012 * systemd-resolved now implements RFC5001 NSID in its local DNS
5013 stub. This may be used by local clients to determine whether they are
5014 talking to the DNS resolver stub or a different DNS server.
5016 * When resolving host names and other records resolvectl will now
5017 report where the data was acquired from (i.e. the local cache, the
5018 network, locally synthesized, …) and whether the network traffic it
5019 effected was encrypted or not. Moreover the tool acquired a number of
5020 new options --cache=, --synthesize=, --network=, --zone=,
5021 --trust-anchor=, --validate= that take booleans and may be used to
5022 tweak a lookup, i.e. whether it may be answered from cached
5023 information, locally synthesized information, information acquired
5024 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
5025 anchor, and whether DNSSEC validation shall be executed for the
5028 * systemd-nspawn gained a new --ambient-capability= setting
5029 (AmbientCapability= in .nspawn files) to configure ambient
5030 capabilities passed to the container payload.
5032 * systemd-nspawn gained the ability to configure the firewall using the
5033 nftables subsystem (in addition to the existing iptables
5034 support). Similarly, systemd-networkd's IPMasquerade= option now
5035 supports nftables as back-end, too. In both cases NAT on IPv6 is now
5036 supported too, in addition to IPv4 (the iptables back-end still is
5039 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
5040 retains its meaning, but has been deprecated. Please switch to either
5041 "ivp4" or "both" (if covering IPv6 is desired).
5043 * systemd-importd will now download .verity and .roothash.p7s files
5044 along with the machine image (as exposed via machinectl pull-raw).
5046 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
5047 setting to configure the time a unit's cgroup needs to exceed memory
5048 pressure limits before action will be taken, and a new
5049 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
5052 systemd-oomd is now considered fully supported (the usual
5053 backwards-compatibility promises apply). Swap is not required for
5054 operation, but it is still recommended.
5056 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
5057 configures the retry delay when trying to contact servers.
5059 * systemd-stdio-bridge gained --system/--user options to connect to the
5060 system bus (previous default) or the user session bus.
5062 * systemd-localed may now call locale-gen to generate missing locales
5063 on-demand (UTF-8-only). This improves integration with Debian-based
5064 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
5066 * systemctl --check-inhibitors=true may now be used to obey inhibitors
5067 even when invoked non-interactively. The old --ignore-inhibitors
5068 switch is now deprecated and replaced by --check-inhibitors=false.
5070 * systemctl import-environment will now emit a warning when called
5071 without any arguments (i.e. to import the full environment block of
5072 the called program). This command will usually be invoked from a
5073 shell, which means that it'll inherit a bunch of variables which are
5074 specific to that shell, and usually to the TTY the shell is connected
5075 to, and don't have any meaning in the global context of the system or
5076 user service manager. Instead, only specific variables should be
5077 imported into the manager environment block.
5079 Similarly, programs which update the manager environment block by
5080 directly calling the D-Bus API of the manager, should also push
5081 specific variables, and not the full inherited environment.
5083 * systemctl's status output now shows unit state with a more careful
5084 choice of Unicode characters: units in maintenance show a "○" symbol
5085 instead of the usual "●", failed units show "×", and services being
5088 * coredumpctl gained a --debugger-arguments= switch to pass arguments
5089 to the debugger. It also gained support for showing coredump info in
5090 a simple JSON format.
5092 * systemctl/loginctl/machinectl's --signal= option now accept a special
5093 value "list", which may be used to show a brief table with known
5094 process signals and their numbers.
5096 * networkctl now shows the link activation policy in status.
5098 * Various tools gained --pager/--no-pager/--json= switches to
5099 enable/disable the pager and provide JSON output.
5101 * Various tools now accept two new values for the SYSTEMD_COLORS
5102 environment variable: "16" and "256", to configure how many terminal
5103 colors are used in output.
5105 * less 568 or newer is now required for the auto-paging logic of the
5106 various tools. Hyperlink ANSI sequences in terminal output are now
5107 used even if a pager is used, and older versions of less are not able
5108 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
5109 disable this output again.
5111 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
5112 builds, non-merged-usr builds) are now officially deprecated. A
5113 warning is emitted during build. Support is slated to be removed in
5114 about a year (when the Debian Bookworm release development starts).
5116 * Systems with the legacy cgroup v1 hierarchy are now marked as
5117 "tainted", to make it clearer that using the legacy hierarchy is not
5120 * systemd-localed will now refuse to configure a keymap which is not
5121 installed in the file system. This is intended as a bug fix, but
5122 could break cases where systemd-localed was used to configure the
5123 keymap in advanced of it being installed. It is necessary to install
5124 the keymap file first.
5126 * The main git development branch has been renamed to 'main'.
5128 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
5129 for partitions, as in the vast majority of cases they contain none
5130 and are used internally by the bootloader (eg: uboot).
5132 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
5133 spawned processes to the PID of the process itself. This may be used
5134 by programs for detecting whether they were forked off by the service
5135 manager itself or are a process forked off further down the tree.
5137 * The sd-device API gained four new calls: sd_device_get_action() to
5138 determine the uevent add/remove/change/… action the device object has
5139 been seen for, sd_device_get_seqno() to determine the uevent sequence
5140 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
5141 object from stat(2) data of a device node, and sd_device_trigger() to
5142 write to the 'uevent' attribute of a device.
5144 * For most tools the --no-legend= switch has been replaced by
5145 --legend=no and --legend=yes, to force whether tables are shown with
5148 * Units acquired a new property "Markers" that takes a list of zero,
5149 one or two of the following strings: "needs-reload" and
5150 "needs-restart". These markers may be set via "systemctl
5151 set-property". Once a marker is set, "systemctl reload-or-restart
5152 --marked" may be invoked to execute the operation the units are
5153 marked for. This is useful for package managers that want to mark
5154 units for restart/reload while updating, but effect the actual
5155 operations at a later step at once.
5157 * The sd_bus_message_read_strv() API call of sd-bus may now also be
5158 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
5161 * bootctl will now report whether the UEFI firmware used a TPM2 device
5162 and measured the boot process into it.
5164 * systemd-tmpfiles learnt support for a new environment variable
5165 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
5166 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
5167 even if the root fs of the system is not itself a btrfs volume.
5169 * systemd-detect-virt/ConditionVirtualization= will now explicitly
5170 detect Docker/Podman environments where possible. Moreover, they
5171 should be able to generically detect any container manager as long as
5172 it assigns the container a cgroup.
5174 * portablectl gained a new "reattach" verb for detaching/reattaching a
5175 portable service image, useful for updating images on-the-fly.
5177 * Intel SGX enclave device nodes (which expose a security feature of
5178 newer Intel CPUs) will now be owned by a new system group "sgx".
5180 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
5181 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
5182 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
5183 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
5184 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
5185 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
5186 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
5187 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
5188 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
5189 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
5190 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
5191 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
5192 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
5193 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
5194 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
5195 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
5196 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
5197 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
5198 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
5199 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
5200 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
5201 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
5202 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
5203 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
5204 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
5205 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
5206 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
5207 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
5208 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
5209 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
5210 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
5211 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
5212 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
5213 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
5214 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
5215 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
5216 Zmicer Turok, Дамјан Георгиевски
5218 — Berlin, 2021-03-30
5222 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
5223 "bind" and "unbind" to the Linux device model. When this kernel
5224 change was made, systemd-udevd was only minimally updated to handle
5225 and propagate these new event types. The introduction of these new
5226 uevents (which are typically generated for USB devices and devices
5227 needing a firmware upload before being functional) resulted in a
5228 number of issues which we so far didn't address. We hoped the kernel
5229 maintainers would themselves address these issues in some form, but
5230 that did not happen. To handle them properly, many (if not most) udev
5231 rules files shipped in various packages need updating, and so do many
5232 programs that monitor or enumerate devices with libudev or sd-device,
5233 or otherwise process uevents. Please note that this incompatibility
5234 is not fault of systemd or udev, but caused by an incompatible kernel
5235 change that happened back in Linux 4.14, but is becoming more and
5236 more visible as the new uevents are generated by more kernel drivers.
5238 To minimize issues resulting from this kernel change (but not avoid
5239 them entirely) starting with systemd-udevd 247 the udev "tags"
5240 concept (which is a concept for marking and filtering devices during
5241 enumeration and monitoring) has been reworked: udev tags are now
5242 "sticky", meaning that once a tag is assigned to a device it will not
5243 be removed from the device again until the device itself is removed
5244 (i.e. unplugged). This makes sure that any application monitoring
5245 devices that match a specific tag is guaranteed to both see uevents
5246 where the device starts being relevant, and those where it stops
5247 being relevant (the latter now regularly happening due to the new
5248 "unbind" uevent type). The udev tags concept is hence now a concept
5249 tied to a *device* instead of a device *event* — unlike for example
5250 udev properties whose lifecycle (as before) is generally tied to a
5251 device event, meaning that the previously determined properties are
5252 forgotten whenever a new uevent is processed.
5254 With the newly redefined udev tags concept, sometimes it's necessary
5255 to determine which tags are the ones applied by the most recent
5256 uevent/database update, in order to discern them from those
5257 originating from earlier uevents/database updates of the same
5258 device. To accommodate for this a new automatic property CURRENT_TAGS
5259 has been added that works similar to the existing TAGS property but
5260 only lists tags set by the most recent uevent/database
5261 update. Similarly, the libudev/sd-device API has been updated with
5262 new functions to enumerate these 'current' tags, in addition to the
5263 existing APIs that now enumerate the 'sticky' ones.
5265 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
5266 essential that all udev rules files and applications are updated to
5267 handle the new events. Specifically:
5269 • All rule files that currently use a header guard similar to
5270 ACTION!="add|change",GOTO="xyz_end" should be updated to use
5271 ACTION=="remove",GOTO="xyz_end" instead, so that the
5272 properties/tags they add are also applied whenever "bind" (or
5273 "unbind") is seen. (This is most important for all physical device
5274 types — those for which "bind" and "unbind" are currently
5275 generated, for all other device types this change is still
5276 recommended but not as important — but certainly prepares for
5277 future kernel uevent type additions).
5279 • Similarly, all code monitoring devices that contains an 'if' branch
5280 discerning the "add" + "change" uevent actions from all other
5281 uevents actions (i.e. considering devices only relevant after "add"
5282 or "change", and irrelevant on all other events) should be reworked
5283 to instead negatively check for "remove" only (i.e. considering
5284 devices relevant after all event types, except for "remove", which
5285 invalidates the device). Note that this also means that devices
5286 should be considered relevant on "unbind", even though conceptually
5287 this — in some form — invalidates the device. Since the precise
5288 effect of "unbind" is not generically defined, devices should be
5289 considered relevant even after "unbind", however I/O errors
5290 accessing the device should then be handled gracefully.
5292 • Any code that uses device tags for deciding whether a device is
5293 relevant or not most likely needs to be updated to use the new
5294 udev_device_has_current_tag() API (or sd_device_has_current_tag()
5295 in case sd-device is used), to check whether the tag is set at the
5296 moment an uevent is seen (as opposed to the existing
5297 udev_device_has_tag() API which checks if the tag ever existed on
5298 the device, following the API concept redefinition explained
5301 We are very sorry for this breakage and the requirement to update
5302 packages using these interfaces. We'd again like to underline that
5303 this is not caused by systemd/udev changes, but result of a kernel
5306 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
5307 packages have not retriggered devices once the udev package (or any
5308 auxiliary package installing additional udev rules) is updated. We
5309 intend to work with major distributions to change this, so that
5310 "udevadm trigger -c change" is issued on such upgrades, ensuring that
5311 the updated ruleset is applied to the devices already discovered, so
5312 that (asynchronously) after the upgrade completed the udev database
5313 is consistent with the updated rule set. This means udev rules must
5314 be ready to be retriggered with a "change" action any time, and
5315 result in correct and complete udev database entries. While the
5316 majority of udev rule files known to us currently get this right,
5317 some don't. Specifically, there are udev rules files included in
5318 various packages that only set udev properties on the "add" action,
5319 but do not handle the "change" action. If a device matching those
5320 rules is retriggered with the "change" action (as is intended here)
5321 it would suddenly lose the relevant properties. This always has been
5322 problematic, but as soon as all udev devices are triggered on relevant
5323 package upgrades this will become particularly so. It is strongly
5324 recommended to fix offending rules so that they can handle a "change"
5325 action at any time, and acquire all necessary udev properties even
5326 then. Or in other words: the header guard mentioned above
5327 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
5328 this, as it makes sure rules are rerun on "change" correctly, and
5329 accumulate the correct and complete set of udev properties. udev rule
5330 definitions that cannot handle "change" events being triggered at
5331 arbitrary times should be considered buggy.
5333 * The MountAPIVFS= service file setting now defaults to on if
5334 RootImage= and RootDirectory= are used, which means that with those
5335 two settings /proc/, /sys/ and /dev/ are automatically properly set
5336 up for services. Previous behaviour may be restored by explicitly
5337 setting MountAPIVFS=off.
5339 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
5340 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
5341 latter it takes precedence over the former, similar to how most of
5342 systemd's own configuration is handled. Given that PAM stack
5343 definitions are primarily put together by OS vendors/distributions
5344 (though possibly overridden by users), this systemd release moves its
5345 own PAM stack configuration for the "systemd-user" PAM service (i.e.
5346 for the PAM session invoked by the per-user user@.service instance)
5347 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
5348 packages' vendor versions of their PAM stack definitions from
5349 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
5350 desired the location to which systemd installs its PAM stack
5351 configuration may be changed via the -Dpamconfdir Meson option.
5353 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
5354 libpwquality and libcryptsetup have been changed to be based on
5355 dlopen(): instead of regular dynamic library dependencies declared in
5356 the binary ELF headers, these libraries are now loaded on demand
5357 only, if they are available. If the libraries cannot be found the
5358 relevant operations will fail gracefully, or a suitable fallback
5359 logic is chosen. This is supposed to be useful for general purpose
5360 distributions, as it allows minimizing the list of dependencies the
5361 systemd packages pull in, permitting building of more minimal OS
5362 images, while still making use of these "weak" dependencies should
5363 they be installed. Since many package managers automatically
5364 synthesize package dependencies from ELF shared library dependencies,
5365 some additional manual packaging work has to be done now to replace
5366 those (slightly downgraded from "required" to "recommended" or
5367 whatever is conceptually suitable for the package manager). Note that
5368 this change does not alter build-time behaviour: as before the
5369 build-time dependencies have to be installed during build, even if
5370 they now are optional during runtime.
5372 * sd-event.h gained a new call sd_event_add_time_relative() for
5373 installing timers relative to the current time. This is mostly a
5374 convenience wrapper around the pre-existing sd_event_add_time() call
5375 which installs absolute timers.
5377 * sd-event event sources may now be placed in a new "exit-on-failure"
5378 mode, which may be controlled via the new
5379 sd_event_source_get_exit_on_failure() and
5380 sd_event_source_set_exit_on_failure() functions. If enabled, any
5381 failure returned by the event source handler functions will result in
5382 exiting the event loop (unlike the default behaviour of just
5383 disabling the event source but continuing with the event loop). This
5384 feature is useful to set for all event sources that define "primary"
5385 program behaviour (where failure should be fatal) in contrast to
5386 "auxiliary" behaviour (where failure should remain local).
5388 * Most event source types sd-event supports now accept a NULL handler
5389 function, in which case the event loop is exited once the event
5390 source is to be dispatched, using the userdata pointer — converted to
5391 a signed integer — as exit code of the event loop. Previously this
5392 was supported for IO and signal event sources already. Exit event
5393 sources still do not support this (simply because it makes little
5394 sense there, as the event loop is already exiting when they are
5397 * A new per-unit setting RootImageOptions= has been added which allows
5398 tweaking the mount options for any file system mounted as effect of
5399 the RootImage= setting.
5401 * Another new per-unit setting MountImages= has been added, that allows
5402 mounting additional disk images into the file system tree accessible
5405 * Timer units gained a new FixedRandomDelay= boolean setting. If
5406 enabled, the random delay configured with RandomizedDelaySec= is
5407 selected in a way that is stable on a given system (though still
5408 different for different units).
5410 * Socket units gained a new setting Timestamping= that takes "us", "ns"
5411 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
5414 * systemd-repart now generates JSON output when requested with the new
5417 * systemd-machined's OpenMachineShell() bus call will now pass
5418 additional policy metadata data fields to the PolicyKit
5419 authentication request.
5421 * systemd-tmpfiles gained a new -E switch, which is equivalent to
5422 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
5423 --exclude=/sys. It's particularly useful in combination with --root=,
5424 when operating on OS trees that do not have any of these four runtime
5425 directories mounted, as this means no files below these subtrees are
5426 created or modified, since those mount points should probably remain
5429 * systemd-tmpfiles gained a new --image= switch which is like --root=,
5430 but takes a disk image instead of a directory as argument. The
5431 specified disk image is mounted inside a temporary mount namespace
5432 and the tmpfiles.d/ drop-ins stored in the image are executed and
5433 applied to the image. systemd-sysusers similarly gained a new
5434 --image= switch, that allows the sysusers.d/ drop-ins stored in the
5435 image to be applied onto the image.
5437 * Similarly, the journalctl command also gained an --image= switch,
5438 which is a quick one-step solution to look at the log data included
5441 * journalctl's --output=cat option (which outputs the log content
5442 without any metadata, just the pure text messages) will now make use
5443 of terminal colors when run on a suitable terminal, similarly to the
5446 * JSON group records now support a "description" string that may be
5447 used to add a human-readable textual description to such groups. This
5448 is supposed to match the user's GECOS field which traditionally
5449 didn't have a counterpart for group records.
5451 * The "systemd-dissect" tool that may be used to inspect OS disk images
5452 and that was previously installed to /usr/lib/systemd/ has now been
5453 moved to /usr/bin/, reflecting its updated status of an officially
5454 supported tool with a stable interface. It gained support for a new
5455 --mkdir switch which when combined with --mount has the effect of
5456 creating the directory to mount the image to if it is missing
5457 first. It also gained two new commands --copy-from and --copy-to for
5458 copying files and directories in and out of an OS image without the
5459 need to manually mount it. It also acquired support for a new option
5460 --json= to generate JSON output when inspecting an OS image.
5462 * The cgroup2 file system is now mounted with the
5463 "memory_recursiveprot" mount option, supported since kernel 5.7. This
5464 means that the MemoryLow= and MemoryMin= unit file settings now apply
5465 recursively to whole subtrees.
5467 * systemd-homed now defaults to using the btrfs file system — if
5468 available — when creating home directories in LUKS volumes. This may
5469 be changed with the DefaultFileSystemType= setting in homed.conf.
5470 It's now the default file system in various major distributions and
5471 has the major benefit for homed that it can be grown and shrunk while
5472 mounted, unlike the other contenders ext4 and xfs, which can both be
5473 grown online, but not shrunk (in fact xfs is the technically most
5474 limited option here, as it cannot be shrunk at all).
5476 * JSON user records managed by systemd-homed gained support for
5477 "recovery keys". These are basically secondary passphrases that can
5478 unlock user accounts/home directories. They are computer-generated
5479 rather than user-chosen, and typically have greater entropy.
5480 homectl's --recovery-key= option may be used to add a recovery key to
5481 a user account. The generated recovery key is displayed as a QR code,
5482 so that it can be scanned to be kept in a safe place. This feature is
5483 particularly useful in combination with systemd-homed's support for
5484 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
5485 security tokens are lost. Recovery keys may be entered wherever the
5486 system asks for a password.
5488 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
5489 home directory which indicates that a home directory has not been
5490 deactivated cleanly when offline. This flag is useful to identify
5491 home directories for which the offline discard logic did not run when
5492 offlining, and where it would be a good idea to log in again to catch
5495 * systemctl gained a new parameter --timestamp= which may be used to
5496 change the style in which timestamps are output, i.e. whether to show
5497 them in local timezone or UTC, or whether to show µs granularity.
5499 * Alibaba's "pouch" container manager is now detected by
5500 systemd-detect-virt, ConditionVirtualization= and similar
5501 constructs. Similar, they now also recognize IBM PowerVM machine
5504 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
5505 place to use for propagating external mounts into the
5506 container. Similarly /run/host/notify is now used as the socket path
5507 for container payloads to communicate with the container manager
5508 using sd_notify(). The container manager now uses the
5509 /run/host/inaccessible/ directory to place "inaccessible" file nodes
5510 of all relevant types which may be used by the container payload as
5511 bind mount source to over-mount inodes to make them inaccessible.
5512 /run/host/container-manager will now be initialized with the same
5513 string as the $container environment variable passed to the
5514 container's PID 1. /run/host/container-uuid will be initialized with
5515 the same string as $container_uuid. This means the /run/host/
5516 hierarchy is now the primary way to make host resources available to
5517 the container. The Container Interface documents these new files and
5520 https://systemd.io/CONTAINER_INTERFACE
5522 * Support for the "ConditionNull=" unit file condition has been
5523 deprecated and undocumented for 6 years. systemd started to warn
5524 about its use 1.5 years ago. It has now been removed entirely.
5526 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
5527 a sd_bus_error struct and a list of error names, and checks if the
5528 error matches one of these names. It's a convenience wrapper that is
5529 useful in cases where multiple errors shall be handled the same way.
5531 * A new system call filter list "@known" has been added, that contains
5532 all system calls known at the time systemd was built.
5534 * Behaviour of system call filter allow lists has changed slightly:
5535 system calls that are contained in @known will result in EPERM by
5536 default, while those not contained in it result in ENOSYS. This
5537 should improve compatibility because known system calls will thus be
5538 communicated as prohibited, while unknown (and thus newer ones) will
5539 be communicated as not implemented, which hopefully has the greatest
5540 chance of triggering the right fallback code paths in client
5543 * "systemd-analyze syscall-filter" will now show two separate sections
5544 at the bottom of the output: system calls known during systemd build
5545 time but not included in any of the filter groups shown above, and
5546 system calls defined on the local kernel but known during systemd
5549 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
5550 systemd-nspawn all system call filter violations will be logged by
5551 the kernel (audit). This is useful for tracking down system calls
5552 invoked by container payloads that are prohibited by the container's
5553 system call filter policy.
5555 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
5556 systemd-nspawn (and other programs that use seccomp) all seccomp
5557 filtering is turned off.
5559 * Two new unit file settings ProtectProc= and ProcSubset= have been
5560 added that expose the hidepid= and subset= mount options of procfs.
5561 All processes of the unit will only see processes in /proc that are
5562 are owned by the unit's user. This is an important new sandboxing
5563 option that is recommended to be set on all system services. All
5564 long-running system services that are included in systemd itself set
5565 this option now. This option is only supported on kernel 5.8 and
5566 above, since the hidepid= option supported on older kernels was not a
5567 per-mount option but actually applied to the whole PID namespace.
5569 * Socket units gained a new boolean setting FlushPending=. If enabled
5570 all pending socket data/connections are flushed whenever the socket
5571 unit enters the "listening" state, i.e. after the associated service
5574 * The unit file setting NUMAMask= gained a new "all" value: when used,
5575 all existing NUMA nodes are added to the NUMA mask.
5577 * A new "credentials" logic has been added to system services. This is
5578 a simple mechanism to pass privileged data to services in a safe and
5579 secure way. It's supposed to be used to pass per-service secret data
5580 such as passwords or cryptographic keys but also associated less
5581 private information such as user names, certificates, and similar to
5582 system services. Each credential is identified by a short user-chosen
5583 name and may contain arbitrary binary data. Two new unit file
5584 settings have been added: SetCredential= and LoadCredential=. The
5585 former allows setting a credential to a literal string, the latter
5586 sets a credential to the contents of a file (or data read from a
5587 user-chosen AF_UNIX stream socket). Credentials are passed to the
5588 service via a special credentials directory, one file for each
5589 credential. The path to the credentials directory is passed in a new
5590 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
5591 are passed in the file system they may be easily referenced in
5592 ExecStart= command lines too, thus no explicit support for the
5593 credentials logic in daemons is required (though ideally daemons
5594 would look for the bits they need in $CREDENTIALS_DIRECTORY
5595 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
5596 backed by unswappable memory if privileges allow it, immutable if
5597 privileges allow it, is accessible only to the service's UID, and is
5598 automatically destroyed when the service stops.
5600 * systemd-nspawn supports the same credentials logic. It can both
5601 consume credentials passed to it via the aforementioned
5602 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
5603 to its payload. The service manager/PID 1 has been updated to match
5604 this: it can also accept credentials from the container manager that
5605 invokes it (in fact: any process that invokes it), and passes them on
5606 to its services. Thus, credentials can be propagated recursively down
5607 the tree: from a system's service manager to a systemd-nspawn
5608 service, to the service manager that runs as container payload and to
5609 the service it runs below. Credentials may also be added on the
5610 systemd-nspawn command line, using new --set-credential= and
5611 --load-credential= command line switches that match the
5612 aforementioned service settings.
5614 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
5615 the partition drop-ins which may be used to format/LUKS
5616 encrypt/populate any created partitions. The partitions are
5617 encrypted/formatted/populated before they are registered in the
5618 partition table, so that they appear atomically: either the
5619 partitions do not exist yet or they exist fully encrypted, formatted,
5620 and populated — there is no time window where they are
5621 "half-initialized". Thus the system is robust to abrupt shutdown: if
5622 the tool is terminated half-way during its operations on next boot it
5623 will start from the beginning.
5625 * systemd-repart's --size= operation gained a new "auto" value. If
5626 specified, and operating on a loopback file it is automatically sized
5627 to the minimal size the size constraints permit. This is useful to
5628 use "systemd-repart" as an image builder for minimally sized images.
5630 * systemd-resolved now gained a third IPC interface for requesting name
5631 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
5632 interface is now supported. The nss-resolve NSS module has been
5633 modified to use this new interface instead of D-Bus. Using Varlink
5634 has a major benefit over D-Bus: it works without a broker service,
5635 and thus already during earliest boot, before the dbus daemon has
5636 been started. This means name resolution via systemd-resolved now
5637 works at the same time systemd-networkd operates: from earliest boot
5638 on, including in the initrd.
5640 * systemd-resolved gained support for a new DNSStubListenerExtra=
5641 configuration file setting which may be used to specify additional IP
5642 addresses the built-in DNS stub shall listen on, in addition to the
5643 main one on 127.0.0.53:53.
5645 * Name lookups issued via systemd-resolved's D-Bus and Varlink
5646 interfaces (and thus also via glibc NSS if nss-resolve is used) will
5647 now honour a trailing dot in the hostname: if specified the search
5648 path logic is turned off. Thus "resolvectl query foo." is now
5649 equivalent to "resolvectl query --search=off foo.".
5651 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
5652 exposes how /etc/resolv.conf is currently managed: by resolved (and
5653 in which mode if so) or another subsystem. "resolvctl" will display
5654 this property in its status output.
5656 * The resolv.conf snippets systemd-resolved provides will now set "."
5657 as the search domain if no other search domain is known. This turns
5658 off the derivation of an implicit search domain by nss-dns for the
5659 hostname, when the hostname is set to an FQDN. This change is done to
5660 make nss-dns using resolv.conf provided by systemd-resolved behave
5661 more similarly to nss-resolve.
5663 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
5664 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
5665 "birth" time (btime) of a file in addition to the atime, mtime, and
5668 * systemd-analyze gained a new verb "capability" that lists all known
5669 capabilities by the systemd build and by the kernel.
5671 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
5672 advance the system clock to it at boot if it is noticed to be before
5673 that time. Previously, PID 1 would only advance the time to an epoch
5674 time that is set during build-time. With this new file OS builders
5675 can change this epoch timestamp on individual OS images without
5676 having to rebuild systemd.
5678 * systemd-logind will now listen to the KEY_RESTART key from the Linux
5679 input layer and reboot the system if it is pressed, similarly to how
5680 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
5681 was originally defined in the Multimedia context (to restart playback
5682 of a song or film), but is now primarily used in various embedded
5683 devices for "Reboot" buttons. Accordingly, systemd-logind will now
5684 honour it as such. This may configured in more detail via the new
5685 HandleRebootKey= and RebootKeyIgnoreInhibited=.
5687 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
5688 copying OS trees, for example in "systemd-nspawn --ephemeral",
5689 "systemd-nspawn --template=", "machinectl clone" and similar. This is
5690 useful when operating with OSTree images, which use hardlinks heavily
5691 throughout, and where such copies previously resulting in "exploding"
5694 * systemd-nspawn's --console= setting gained support for a new
5695 "autopipe" value, which is identical to "interactive" when invoked on
5696 a TTY, and "pipe" otherwise.
5698 * systemd-networkd's .network files gained support for explicitly
5699 configuring the multicast membership entries of bridge devices in the
5700 [BridgeMDB] section. It also gained support for the PIE queuing
5701 discipline in the [FlowQueuePIE] sections.
5703 * systemd-networkd's .netdev files may now be used to create "BareUDP"
5704 tunnels, configured in the new [BareUDP] setting.
5706 * systemd-networkd's Gateway= setting in .network files now accepts the
5707 special values "_dhcp4" and "_ipv6ra" to configure additional,
5708 locally defined, explicit routes to the gateway acquired via DHCP or
5709 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
5710 but still accepted for backwards compatibility.
5712 * systemd-networkd's [IPv6PrefixDelegation] section and
5713 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
5714 IPv6SendRA= (the old names are still accepted for backwards
5717 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
5718 boolean setting in [Network] section. If enabled, the delegated prefix
5719 gained by another link will be configured, and an address within the
5720 prefix will be assigned.
5722 * systemd-networkd's .network files gained the Announce= boolean setting
5723 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
5724 prefix will be announced through IPv6 router advertisement (IPv6 RA).
5725 The setting is enabled by default.
5727 * VXLAN tunnels may now be marked as independent of any underlying
5728 network interface via the new Independent= boolean setting.
5730 * systemctl gained support for two new verbs: "service-log-level" and
5731 "service-log-target" may be used on services that implement the
5732 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
5733 adjust the log level and target. All of systemd's long-running
5734 services support this now, but ideally all system services would
5735 implement this interface to make the system more uniformly
5738 * The SystemCallErrorNumber= unit file setting now accepts the new
5739 "kill" and "log" actions, in addition to arbitrary error number
5740 specifications as before. If "kill" the processes are killed on the
5741 event, if "log" the offending system call is audit logged.
5743 * A new SystemCallLog= unit file setting has been added that accepts a
5744 list of system calls that shall be logged about (audit).
5746 * The OS image dissection logic (as used by RootImage= in unit files or
5747 systemd-nspawn's --image= switch) has gained support for identifying
5748 and mounting explicit /usr/ partitions, which are now defined in the
5749 discoverable partition specification. This should be useful for
5750 environments where the root file system is
5751 generated/formatted/populated dynamically on first boot and combined
5752 with an immutable /usr/ tree that is supplied by the vendor.
5754 * In the final phase of shutdown, within the systemd-shutdown binary
5755 we'll now try to detach MD devices (i.e software RAID) in addition to
5756 loopback block devices and DM devices as before. This is supposed to
5757 be a safety net only, in order to increase robustness if things go
5758 wrong. Storage subsystems are expected to properly detach their
5759 storage volumes during regular shutdown already (or in case of
5760 storage backing the root file system: in the initrd hook we return to
5763 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
5764 will now log the thread ID in their log output. This is useful when
5765 working with heavily threaded programs.
5767 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
5768 not use the RDRAND CPU instruction. This is useful in environments
5769 such as replay debuggers where non-deterministic behaviour is not
5772 * The autopaging logic in systemd's various tools (such as systemctl)
5773 has been updated to turn on "secure" mode in "less"
5774 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
5775 detected. This disables invoking external programs from the pager,
5776 via the pipe logic. This behaviour may be overridden via the new
5777 $SYSTEMD_PAGERSECURE environment variable.
5779 * Units which have resource limits (.service, .mount, .swap, .slice,
5780 .socket, and .slice) gained new configuration settings
5781 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
5782 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
5783 limits and optional action taken by systemd-oomd.
5785 * A new service systemd-oomd has been added. It monitors resource
5786 contention for selected parts of the unit hierarchy using the PSI
5787 information reported by the kernel, and kills processes when memory
5788 or swap pressure is above configured limits. This service is only
5789 enabled by default in developer mode (see below) and should be
5790 considered a preview in this release. Behaviour details and option
5791 names are subject to change without the usual backwards-compatibility
5794 * A new helper oomctl has been added to introspect systemd-oomd state.
5795 It is only enabled by default in developer mode and should be
5796 considered a preview without the usual backwards-compatibility
5799 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
5800 enabled, systemd reads the system UID boundaries from /etc/login.defs
5801 at runtime, instead of using the built-in values selected during
5802 build. This is an option to improve compatibility for upgrades from
5803 old systems. It's strongly recommended not to make use of this
5804 functionality on new systems (or even enable it during build), as it
5805 makes something runtime-configurable that is mostly an implementation
5806 detail of the OS, and permits avoidable differences in deployments
5807 that create all kinds of problems in the long run.
5809 * New meson option '-Dmode=developer|release' has been added. When
5810 'developer', additional checks and features are enabled that are
5811 relevant during upstream development, e.g. verification that
5812 semi-automatically-generated documentation has been properly updated
5813 following API changes. Those checks are considered hints for
5814 developers and are not actionable in downstream builds. In addition,
5815 extra features that are not ready for general consumption may be
5816 enabled in developer mode. It is thus recommended to set
5817 '-Dmode=release' in end-user and distro builds.
5819 * systemd-cryptsetup gained support for processing detached LUKS
5820 headers specified on the kernel command line via the header=
5821 parameter of the luks.options= kernel command line option. The same
5822 device/path syntax as for key files is supported for header files
5825 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
5826 slot index data for devices that are connected through a PCI bridge
5827 where the _SUN index is associated with the bridge instead of the
5828 network device itself. Previously this would create ambiguous device
5829 naming if multiple network interfaces were connected to the same PCI
5830 bridge. Since this is a naming scheme incompatibility on systems that
5831 possess hardware like this it has been introduced as new naming
5832 scheme "v247". The previous scheme can be selected via the
5833 "net.naming_scheme=v245" kernel command line parameter.
5835 * ConditionFirstBoot= semantics have been modified to be safe towards
5836 abnormal system power-off during first boot. Specifically, the
5837 "systemd-machine-id-commit.service" service now acts as boot
5838 milestone indicating when the first boot process is sufficiently
5839 complete in order to not consider the next following boot also a
5840 first boot. If the system is reset before this unit is reached the
5841 first time, the next boot will still be considered a first boot; once
5842 it has been reached, no further boots will be considered a first
5843 boot. The "first-boot-complete.target" unit now acts as official hook
5844 point to order against this. If a service shall be run on every boot
5845 until the first boot fully succeeds it may thus be ordered before
5846 this target unit (and pull it in) and carry ConditionFirstBoot=
5849 * bootctl's set-default and set-oneshot commands now accept the three
5850 special strings "@default", "@oneshot", "@current" in place of a boot
5851 entry id. These strings are resolved to the current default and
5852 oneshot boot loader entry, as well as the currently booted one. Thus
5853 a command "bootctl set-default @current" may be used to make the
5854 currently boot menu item the new default for all subsequent boots.
5856 * "systemctl edit" has been updated to show the original effective unit
5857 contents in commented form in the text editor.
5859 * Units in user mode are now segregated into three new slices:
5860 session.slice (units that form the core of graphical session),
5861 app.slice ("normal" user applications), and background.slice
5862 (low-priority tasks). Unless otherwise configured, user units are
5863 placed in app.slice. The plan is to add resource limits and
5864 protections for the different slices in the future.
5866 * New GPT partition types for RISCV32/64 for the root and /usr
5867 partitions, and their associated Verity partitions have been defined,
5868 and are now understood by systemd-gpt-auto-generator, and the OS
5869 image dissection logic.
5871 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
5872 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
5873 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
5874 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
5875 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
5876 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
5877 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
5878 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
5879 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
5880 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
5881 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
5882 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
5883 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
5884 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
5885 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
5886 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
5887 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
5888 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
5889 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
5890 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
5891 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
5892 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
5893 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
5894 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
5895 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
5896 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
5897 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
5898 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
5899 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
5900 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
5901 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
5902 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
5903 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
5905 – Warsaw, 2020-11-26
5909 * The service manager gained basic support for cgroup v2 freezer. Units
5910 can now be suspended or resumed either using new systemctl verbs,
5911 freeze and thaw respectively, or via D-Bus.
5913 * PID 1 may now automatically load pre-compiled AppArmor policies from
5914 /etc/apparmor/earlypolicy during early boot.
5916 * The CPUAffinity= setting in service unit files now supports a new
5917 special value "numa" that causes the CPU affinity masked to be set
5918 based on the NUMA mask.
5920 * systemd will now log about all left-over processes remaining in a
5921 unit when the unit is stopped. It will now warn about services using
5922 KillMode=none, as this is generally an unsafe thing to make use of.
5924 * Two new unit file settings
5925 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
5926 added. They may be used to check whether a specific file system path
5927 resides on a block device that is encrypted on the block level
5928 (i.e. using dm-crypt/LUKS).
5930 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
5931 has been added that may be used for simple environment checks. This
5932 is particularly useful when passing in environment variables from a
5933 container manager (or from PAM in case of the systemd --user
5936 * .service unit files now accept a new setting CoredumpFilter= which
5937 allows configuration of the memory sections coredumps of the
5938 service's processes shall include.
5940 * .mount units gained a new ReadWriteOnly= boolean option. If set
5941 it will not be attempted to mount a file system read-only if mounting
5942 in read-write mode doesn't succeed. An option x-systemd.rw-only is
5943 available in /etc/fstab to control the same.
5945 * .socket units gained a new boolean setting PassPacketInfo=. If
5946 enabled, the kernel will attach additional per-packet metadata to all
5947 packets read from the socket, as an ancillary message. This controls
5948 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
5949 depending on socket type.
5951 * .service units gained a new setting RootHash= which may be used to
5952 specify the root hash for verity enabled disk images which are
5953 specified in RootImage=. RootVerity= may be used to specify a path to
5954 the Verity data matching a RootImage= file system. (The latter is
5955 only useful for images that do not contain the Verity data embedded
5956 into the same image that carries a GPT partition table following the
5957 Discoverable Partition Specification). Similarly, systemd-nspawn
5958 gained a new switch --verity-data= that takes a path to a file with
5959 the verity data of the disk image supplied in --image=, if the image
5960 doesn't contain the verity data itself.
5962 * .service units gained a new setting RootHashSignature= which takes
5963 either a base64 encoded PKCS#7 signature of the root hash specified
5964 with RootHash=, or a path to a file to read the signature from. This
5965 allows validation of the root hash against public keys available in
5966 the kernel keyring, and is only supported on recent kernels
5967 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
5968 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
5969 this mechanism has also been added to systemd-veritysetup.
5971 * .service unit files gained two new options
5972 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
5973 tune behaviour if a start or stop timeout is hit, i.e. whether to
5974 terminate the service with SIGTERM, SIGABRT or SIGKILL.
5976 * Most options in systemd that accept hexadecimal values prefixed with
5977 0x in additional to the usual decimal notation now also support octal
5978 notation when the 0o prefix is used and binary notation if the 0b
5981 * Various command line parameters and configuration file settings that
5982 configure key or certificate files now optionally take paths to
5983 AF_UNIX sockets in the file system. If configured that way a stream
5984 connection is made to the socket and the required data read from
5985 it. This is a simple and natural extension to the existing regular
5986 file logic, and permits other software to provide keys or
5987 certificates via simple IPC services, for example when unencrypted
5988 storage on disk is not desired. Specifically, systemd-networkd's
5989 Wireguard and MACSEC key file settings as well as
5990 systemd-journal-gatewayd's and systemd-journal-remote's PEM
5991 key/certificate parameters support this now.
5993 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
5994 configuration files that support specifier expansion learnt six new
5995 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
5996 resolve to the various ID fields from /etc/os-release, %l resolves to
5997 the "short" hostname of the system, i.e. the hostname configured in
5998 the kernel truncated at the first dot.
6000 * Support for the .include syntax in unit files has been removed. The
6001 concept has been obsolete for 6 years and we started warning about
6002 its pending removal 2 years ago (also see NEWS file below). It's
6005 * StandardError= and StandardOutput= in unit files no longer support
6006 the "syslog" and "syslog-console" switches. They were long removed
6007 from the documentation, but will now result in warnings when used,
6008 and be converted to "journal" and "journal+console" automatically.
6010 * If the service setting User= is set to the "nobody" user, a warning
6011 message is now written to the logs (but the value is nonetheless
6012 accepted). Setting User=nobody is unsafe, since the primary purpose
6013 of the "nobody" user is to own all files whose owner cannot be mapped
6014 locally. It's in particular used by the NFS subsystem and in user
6015 namespacing. By running a service under this user's UID it might get
6016 read and even write access to all these otherwise unmappable files,
6017 which is quite likely a major security problem.
6019 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
6020 and others) now have a size and inode limits applied (50% of RAM for
6021 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
6022 that the implicit kernel default is 50% too, so there is no change
6023 in the size limit for /tmp and /dev/shm.
6025 * nss-mymachines lost support for resolution of users and groups, and
6026 now only does resolution of hostnames. This functionality is now
6027 provided by nss-systemd. Thus, the 'mymachines' entry should be
6028 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
6029 (and 'systemd' added if it is not already there).
6031 * A new kernel command line option systemd.hostname= has been added
6032 that allows controlling the hostname that is initialized early during
6035 * A kernel command line option "udev.blockdev_read_only" has been
6036 added. If specified all hardware block devices that show up are
6037 immediately marked as read-only by udev. This option is useful for
6038 making sure that a specific boot under no circumstances modifies data
6039 on disk. Use "blockdev --setrw" to undo the effect of this, per
6042 * A new boolean kernel command line option systemd.swap= has been
6043 added, which may be used to turn off automatic activation of swap
6044 devices listed in /etc/fstab.
6046 * New kernel command line options systemd.condition_needs_update= and
6047 systemd.condition_first_boot= have been added, which override the
6048 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
6051 * A new kernel command line option systemd.clock_usec= has been added
6052 that allows setting the system clock to the specified time in µs
6053 since Jan 1st, 1970 early during boot. This is in particular useful
6054 in order to make test cases more reliable.
6056 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
6057 systemd-coredump to save core files for suid processes. When saving
6058 the core file, systemd-coredump will use the effective uid and gid of
6059 the process that faulted.
6061 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
6062 now automatically set to "Y" at boot, in order to enable pstore
6063 generation for collection with systemd-pstore.
6065 * We provide a set of udev rules to enable auto-suspend on PCI and USB
6066 devices that were tested to correctly support it. Previously, this
6067 was distributed as a set of udev rules, but has now been replaced by
6068 by a set of hwdb entries (and a much shorter udev rule to take action
6069 if the device modalias matches one of the new hwdb entries).
6071 As before, entries are periodically imported from the database
6072 maintained by the ChromiumOS project. If you have a device that
6073 supports auto-suspend correctly and where it should be enabled by
6074 default, please submit a patch that adds it to the database (see
6075 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
6077 * systemd-udevd gained the new configuration option timeout_signal= as well
6078 as a corresponding kernel command line option udev.timeout_signal=.
6079 The option can be used to configure the UNIX signal that the main
6080 daemon sends to the worker processes on timeout. Setting the signal
6081 to SIGABRT is useful for debugging.
6083 * .link files managed by systemd-udevd gained options RxFlowControl=,
6084 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
6085 order to configure various flow control parameters. They also gained
6086 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
6087 frame ring buffer sizes.
6089 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
6090 enabled systemd-networkd manages all routes configured by other tools.
6092 * .network files managed by systemd-networkd gained a new section
6093 [SR-IOV], in order to configure SR-IOV capable network devices.
6095 * systemd-networkd's [IPv6Prefix] section in .network files gained a
6096 new boolean setting Assign=. If enabled an address from the prefix is
6097 automatically assigned to the interface.
6099 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
6100 controls delegated prefixes assigned by DHCPv6 client. The section
6101 has three settings: SubnetID=, Assign=, and Token=. The setting
6102 SubnetID= allows explicit configuration of the preferred subnet that
6103 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
6104 Assign= is enabled (which is the default) an address from any acquired
6105 delegated prefix is automatically chosen and assigned to the
6106 interface. The setting Token= specifies an optional address generation
6109 * systemd-networkd's [Network] section gained a new setting
6110 IPv4AcceptLocal=. If enabled the interface accepts packets with local
6113 * systemd-networkd gained support for configuring the HTB queuing
6114 discipline in the [HierarchyTokenBucket] and
6115 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
6116 be configured in the [PFIFO] section, "GRED" in
6117 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
6118 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
6119 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
6120 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
6121 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
6122 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
6124 * systemd-networkd gained support for a new Termination= setting in the
6125 [CAN] section for configuring the termination resistor. It also
6126 gained a new ListenOnly= setting for controlling whether to only
6127 listen on CAN interfaces, without interfering with traffic otherwise
6128 (which is useful for debugging/monitoring CAN network
6129 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
6130 been added to configure various CAN-FD aspects.
6132 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
6133 When enabled, DHCPv6 will be attempted right-away without requiring an
6134 Router Advertisement packet suggesting it first (i.e. without the 'M'
6135 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
6136 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
6137 the RA packets suggest it.
6139 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
6140 which may be used to turn off use of the gateway information provided
6141 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
6142 used to configure how to process leases that lack a lifetime option.
6144 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
6145 setting SendVendorOption= allowing configuration of additional vendor
6146 options to send in the DHCP requests/responses. The [DHCPv6] section
6147 gained a new SendOption= setting for sending arbitrary DHCP
6148 options. RequestOptions= has been added to request arbitrary options
6149 from the server. UserClass= has been added to set the DHCP user class
6152 * systemd-networkd's [DHCPServer] section gained a new set of options
6153 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
6154 information about these three protocols in the DHCP lease. It also
6155 gained support for including "MUD" URLs ("Manufacturer Usage
6156 Description"). Support for "MUD" URLs was also added to the LLDP
6157 stack, configurable in the [LLDP] section in .network files.
6159 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
6160 mode. Also, the sections now support a new setting SourceMACAddress=.
6162 * systemd-networkd's .netdev files now support a new setting
6163 VLANProtocol= in the [Bridge] section that allows configuration of
6164 the VLAN protocol to use.
6166 * systemd-networkd supports a new Group= setting in the [Link] section
6167 of the .network files, to control the link group.
6169 * systemd-networkd's [Network] section gained a new
6170 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
6171 link local address is generated.
6173 * A new default .network file is now shipped that matches TUN/TAP
6174 devices that begin with "vt-" in their name. Such interfaces will
6175 have IP routing onto the host links set up automatically. This is
6176 supposed to be used by VM managers to trivially acquire a network
6177 interface which is fully set up for host communication, simply by
6178 carefully picking an interface name to use.
6180 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
6181 which sets the route priority for routes specified by the DHCP server.
6183 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
6184 which configures the vendor class information sent to DHCP server.
6186 * The BlackList= settings in .network files' [DHCPv4] and
6187 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
6188 are still understood to provide compatibility.
6190 * networkctl gained the new "forcerenew" command for forcing all DHCP
6191 server clients to renew their lease. The interface "status" output
6192 will now show numerous additional fields of information about an
6193 interface. There are new "up" and "down" commands to bring specific
6194 interfaces up or down.
6196 * systemd-resolved's DNS= configuration option now optionally accepts a
6197 port number (after ":") and a host name (after "#"). When the host
6198 name is specified, the DNS-over-TLS certificate is validated to match
6199 the specified hostname. Additionally, in case of IPv6 addresses, an
6200 interface may be specified (after "%").
6202 * systemd-resolved may be configured to forward single-label DNS names.
6203 This is not standard-conformant, but may make sense in setups where
6204 public DNS servers are not used.
6206 * systemd-resolved's DNS-over-TLS support gained SNI validation.
6208 * systemd-nspawn's --resolv-conf= switch gained a number of new
6209 supported values. Specifically, options starting with "replace-" are
6210 like those prefixed "copy-" but replace any existing resolv.conf
6211 file. And options ending in "-uplink" and "-stub" can now be used to
6212 propagate other flavours of resolv.conf into the container (as
6213 defined by systemd-resolved).
6215 * The various programs included in systemd can now optionally output
6216 their log messages on stderr prefixed with a timestamp, controlled by
6217 the $SYSTEMD_LOG_TIME environment variable.
6219 * systemctl gained a new "-P" switch that is a shortcut for "--value
6222 * "systemctl list-units" and "systemctl list-machines" no longer hide
6223 their first output column with --no-legend. To hide the first column,
6226 * "systemctl reboot" takes the option "--reboot-argument=".
6227 The optional positional argument to "systemctl reboot" is now
6228 being deprecated in favor of this option.
6230 * systemd-run gained a new switch --slice-inherit. If specified the
6231 unit it generates is placed in the same slice as the systemd-run
6234 * systemd-journald gained support for zstd compression of large fields
6235 in journal files. The hash tables in journal files have been hardened
6236 against hash collisions. This is an incompatible change and means
6237 that journal files created with new systemd versions are not readable
6238 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
6239 environment variable for systemd-journald.service is set to 0 this
6240 new hardening functionality may be turned off, so that generated
6241 journal files remain compatible with older journalctl
6244 * journalctl will now include a clickable link in the default output for
6245 each log message for which a URL with further documentation is
6246 known. This is only supported on terminal emulators that support
6247 clickable hyperlinks, and is turned off if a pager is used (since
6248 "less" still doesn't support hyperlinks,
6249 unfortunately). Documentation URLs may be included in log messages
6250 either by including a DOCUMENTATION= journal field in it, or by
6251 associating a journal message catalog entry with the log message's
6252 MESSAGE_ID, which then carries a "Documentation:" tag.
6254 * journald.conf gained a new boolean setting Audit= that may be used to
6255 control whether systemd-journald will enable audit during
6258 * when systemd-journald's log stream is broken up into multiple lines
6259 because the PID of the sender changed this is indicated in the
6260 generated log records via the _LINE_BREAK=pid-change field.
6262 * journalctl's "-o cat" output mode will now show one or more journal
6263 fields specified with --output-fields= instead of unconditionally
6264 MESSAGE=. This is useful to retrieve a very specific set of fields
6265 without any decoration.
6267 * The sd-journal.h API gained two new functions:
6268 sd_journal_enumerate_available_unique() and
6269 sd_journal_enumerate_available_data() that operate like their
6270 counterparts that lack the _available_ in the name, but skip items
6271 that cannot be read and processed by the local implementation
6272 (i.e. are compressed in an unsupported format or such),
6274 * coredumpctl gained a new --file= switch, matching the same one in
6275 journalctl: a specific journal file may be specified to read the
6278 * coredumps collected by systemd-coredump may now be compressed using
6281 * systemd-binfmt gained a new switch --unregister for unregistering all
6282 registered entries at once. This is now invoked automatically at
6283 shutdown, so that binary formats registered with the "F" flag will
6284 not block clean file system unmounting.
6286 * systemd-notify's --pid= switch gained new values: "parent", "self",
6287 "auto" for controlling which PID to send to the service manager: the
6288 systemd-notify process' PID, or the one of the process invoking it.
6290 * systemd-logind's Session bus object learnt a new method call
6291 SetType() for temporarily updating the session type of an already
6292 allocated session. This is useful for upgrading tty sessions to
6293 graphical ones once a compositor is invoked.
6295 * systemd-socket-proxy gained a new switch --exit-idle-time= for
6296 configuring an exit-on-idle time.
6298 * systemd-repart's --empty= setting gained a new value "create". If
6299 specified a new empty regular disk image file is created under the
6300 specified name. Its size may be specified with the new --size=
6301 option. The latter is also supported without the "create" mode, in
6302 order to grow existing disk image files to the specified size. These
6303 two new options are useful when creating or manipulating disk images
6304 instead of operating on actual block devices.
6306 * systemd-repart drop-ins now support a new UUID= setting to control
6307 the UUID to assign to a newly created partition.
6309 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
6312 * systemd-repart's Label= setting now support the usual, simple
6313 specifier expansion.
6315 * systemd-homed's LUKS backend gained the ability to discard empty file
6316 system blocks automatically when the user logs out. This is enabled
6317 by default to ensure that home directories take minimal space when
6318 logged out but get full size guarantees when logged in. This may be
6319 controlled with the new --luks-offline-discard= switch to homectl.
6321 * If systemd-homed detects that /home/ is encrypted as a whole it will
6322 now default to the directory or subvolume backends instead of the
6323 LUKS backend, in order to avoid double encryption. The default
6324 storage and file system may now be configured explicitly, too, via
6325 the new /etc/systemd/homed.conf configuration file.
6327 * systemd-homed now supports unlocking home directories with FIDO2
6328 security tokens that support the 'hmac-secret' extension, in addition
6329 to the existing support for PKCS#11 security token unlocking
6330 support. Note that many recent hardware security tokens support both
6331 interfaces. The FIDO2 support is accessible via homectl's
6332 --fido2-device= option.
6334 * homectl's --pkcs11-uri= setting now accepts two special parameters:
6335 if "auto" is specified and only one suitable PKCS#11 security token
6336 is plugged in, its URL is automatically determined and enrolled for
6337 unlocking the home directory. If "list" is specified a brief table of
6338 suitable PKCS#11 security tokens is shown. Similar, the new
6339 --fido2-device= option also supports these two special values, for
6340 automatically selecting and listing suitable FIDO2 devices.
6342 * The /etc/crypttab tmp option now optionally takes an argument
6343 selecting the file system to use. Moreover, the default is now
6344 changed from ext2 to ext4.
6346 * There's a new /etc/crypttab option "keyfile-erase". If specified the
6347 key file listed in the same line is removed after use, regardless if
6348 volume activation was successful or not. This is useful if the key
6349 file is only acquired transiently at runtime and shall be erased
6350 before the system continues to boot.
6352 * There's also a new /etc/crypttab option "try-empty-password". If
6353 specified, before asking the user for a password it is attempted to
6354 unlock the volume with an empty password. This is useful for
6355 installing encrypted images whose password shall be set on first boot
6356 instead of at installation time.
6358 * systemd-cryptsetup will now attempt to load the keys to unlock
6359 volumes with automatically from files in
6360 /etc/cryptsetup-keys.d/<volume>.key and
6361 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
6363 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
6364 /etc/crypttab, during boot.
6366 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
6367 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
6370 * A new generator systemd-xdg-autostart-generator has been added. It
6371 generates systemd unit files from XDG autostart .desktop files, and
6372 may be used to let the systemd user instance manage services that are
6373 started automatically as part of the desktop session.
6375 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
6376 to query and change the firmware's 'Reboot Into Firmware Interface'
6379 * systemd-firstboot gained a new switch --kernel-command-line= that may
6380 be used to initialize the /etc/kernel/cmdline file of the image. It
6381 also gained a new switch --root-password-hashed= which is like
6382 --root-password= but accepts a pre-hashed UNIX password as
6383 argument. The new option --delete-root-password may be used to unset
6384 any password for the root user (dangerous!). The --root-shell= switch
6385 may be used to control the shell to use for the root account. A new
6386 --force option may be used to override any already set settings with
6387 the parameters specified on the command line (by default, the tool
6388 will not override what has already been set before, i.e. is purely
6391 * systemd-firstboot gained support for a new --image= switch, which is
6392 similar to --root= but accepts the path to a disk image file, on
6393 which it then operates.
6395 * A new sd-path.h API has been added to libsystemd. It provides a
6396 simple API for retrieving various search paths and primary
6397 directories for various resources.
6399 * A new call sd_notify_barrier() has been added to the sd-daemon.h
6400 API. The call will block until all previously sent sd_notify()
6401 messages have been processed by the service manager. This is useful
6402 to remove races caused by a process already having disappeared at the
6403 time a notification message is processed by the service manager,
6404 making correct attribution impossible. The systemd-notify tool will
6405 now make use of this call implicitly, but this can be turned off again
6406 via the new --no-block switch.
6408 * When sending a file descriptor (fd) to the service manager to keep
6409 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
6410 may be specified. If passed the service manager will refrain from
6411 poll()ing on the file descriptor. Traditionally (and when the
6412 parameter is not specified), the service manager will poll it for
6413 POLLHUP or POLLERR events, and immediately close the fds in that
6416 * The service manager (PID1) gained a new D-Bus method call
6417 SetShowStatus() which may be used to control whether it shall show
6418 boot-time status output on the console. This method has a similar
6419 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
6421 * The sd-bus API gained a number of convenience functions that take
6422 va_list arguments rather than "...". For example, there's now
6423 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
6424 it easier to build wrappers that accept variadic arguments and want
6425 to pass a ready va_list structure to sd-bus.
6427 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
6428 flag which alters how the userdata pointer to pass to the callbacks
6429 is determined. When the flag is set, the offset field is converted
6430 as-is into a pointer, without adding it to the object pointer the
6431 vtable is associated with.
6433 * sd-bus now exposes four new functions:
6434 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
6435 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
6436 validate strings to check if they qualify as various D-Bus concepts.
6438 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
6439 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
6440 that simplify adding argument names to D-Bus methods and signals.
6442 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
6444 * Various D-Bus APIs of systemd daemons now have man pages that
6445 document the methods, signals and properties.
6447 * The expectations on user/group name syntax are now documented in
6448 detail; documentation on how classic home directories may be
6449 converted into home directories managed by homed has been added;
6450 documentation regarding integration of homed/userdb functionality in
6451 desktops has been added:
6453 https://systemd.io/USER_NAMES
6454 https://systemd.io/CONVERTING_TO_HOMED
6455 https://systemd.io/USERDB_AND_DESKTOPS
6457 * Documentation for the on-disk Journal file format has been updated
6458 and has now moved to:
6460 https://systemd.io/JOURNAL_FILE_FORMAT
6462 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
6463 has been extended by a set of environment variables that expose
6464 select fields from the host's os-release file to the container
6465 payload. Similarly, host's os-release files can be mounted into the
6466 container underneath /run/host. Together, those mechanisms provide a
6467 standardized way to expose information about the host to the
6468 container payload. Both interfaces are implemented in systemd-nspawn.
6470 * All D-Bus services shipped in systemd now implement the generic
6471 LogControl1 D-Bus API which allows clients to change log level +
6472 target of the service during runtime.
6474 * Only relevant for developers: the mkosi.default symlink has been
6475 dropped from version control. Please create a symlink to one of the
6476 distribution-specific defaults in .mkosi/ based on your preference.
6478 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
6479 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
6480 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
6481 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
6482 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
6483 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
6484 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
6485 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
6486 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
6487 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
6488 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
6489 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
6490 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
6491 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
6492 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
6493 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
6494 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
6495 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
6496 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
6497 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
6498 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
6499 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
6500 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
6501 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
6502 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
6503 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
6504 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
6505 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
6506 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
6507 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
6508 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
6509 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
6510 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
6511 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
6512 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
6513 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
6514 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
6515 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
6517 – Warsaw, 2020-07-30
6521 * A new tool "systemd-repart" has been added, that operates as an
6522 idempotent declarative repartitioner for GPT partition tables.
6523 Specifically, a set of partitions that must or may exist can be
6524 configured via drop-in files, and during every boot the partition
6525 table on disk is compared with these files, creating missing
6526 partitions or growing existing ones based on configurable relative
6527 and absolute size constraints. The tool is strictly incremental,
6528 i.e. does not delete, shrink or move partitions, but only adds and
6529 grows them. The primary use-case is OS images that ship in minimized
6530 form, that on first boot are grown to the size of the underlying
6531 block device or augmented with additional partitions. For example,
6532 the root partition could be extended to cover the whole disk, or a
6533 swap or /home partitions could be added on first boot. It can also be
6534 used for systems that use an A/B update scheme but ship images with
6535 just the A partition, with B added on first boot. The tool is
6536 primarily intended to be run in the initrd, shortly before
6537 transitioning into the host OS, but can also be run after the
6538 transition took place. It automatically discovers the disk backing
6539 the root file system, and should hence not require any additional
6540 configuration besides the partition definition drop-ins. If no
6541 configuration drop-ins are present, no action is taken.
6543 * A new component "userdb" has been added, along with a small daemon
6544 "systemd-userdbd.service" and a client tool "userdbctl". The framework
6545 allows defining rich user and group records in a JSON format,
6546 extending on the classic "struct passwd" and "struct group"
6547 structures. Various components in systemd have been updated to
6548 process records in this format, including systemd-logind and
6549 pam-systemd. The user records are intended to be extensible, and
6550 allow setting various resource management, security and runtime
6551 parameters that shall be applied to processes and sessions of the
6552 user as they log in. This facility is intended to allow associating
6553 such metadata directly with user/group records so that they can be
6554 produced, extended and consumed in unified form. We hope that
6555 eventually frameworks such as sssd will generate records this way, so
6556 that for the first time resource management and various other
6557 per-user settings can be configured in LDAP directories and then
6558 provided to systemd (specifically to systemd-logind and pam-system)
6559 to apply on login. For further details see:
6561 https://systemd.io/USER_RECORD
6562 https://systemd.io/GROUP_RECORD
6563 https://systemd.io/USER_GROUP_API
6565 * A small new service systemd-homed.service has been added, that may be
6566 used to securely manage home directories with built-in encryption.
6567 The complete user record data is unified with the home directory,
6568 thus making home directories naturally migratable. Its primary
6569 back-end is based on LUKS volumes, but fscrypt, plain directories,
6570 and other storage schemes are also supported. This solves a couple of
6571 problems we saw with traditional ways to manage home directories, in
6572 particular when it comes to encryption. For further discussion of
6573 this, see the video of Lennart's talk at AllSystemsGo! 2019:
6575 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
6577 For further details about the format and expectations on home
6578 directories this new daemon makes, see:
6580 https://systemd.io/HOME_DIRECTORY
6582 * systemd-journald is now multi-instantiable. In addition to the main
6583 instance systemd-journald.service there's now a template unit
6584 systemd-journald@.service, with each instance defining a new named
6585 log 'namespace' (whose name is specified via the instance part of the
6586 unit name). A new unit file setting LogNamespace= has been added,
6587 taking such a namespace name, that assigns services to the specified
6588 log namespaces. As each log namespace is serviced by its own
6589 independent journal daemon, this functionality may be used to improve
6590 performance and increase isolation of applications, at the price of
6591 losing global message ordering. Each instance of journald has a
6592 separate set of configuration files, with possibly different disk
6593 usage limitations and other settings.
6595 journalctl now takes a new option --namespace= to show logs from a
6596 specific log namespace. The sd-journal.h API gained
6597 sd_journal_open_namespace() for opening the log stream of a specific
6598 log namespace. systemd-journald also gained the ability to exit on
6599 idle, which is useful in the context of log namespaces, as this means
6600 log daemons for log namespaces can be activated automatically on
6601 demand and will stop automatically when no longer used, minimizing
6604 * When systemd-tmpfiles copies a file tree using the 'C' line type it
6605 will now label every copied file according to the SELinux database.
6607 * When systemd/PID 1 detects it is used in the initrd it will now boot
6608 into initrd.target rather than default.target by default. This should
6609 make it simpler to build initrds with systemd as for many cases the
6610 only difference between a host OS image and an initrd image now is
6611 the presence of the /etc/initrd-release file.
6613 * A new kernel command line option systemd.cpu_affinity= is now
6614 understood. It's equivalent to the CPUAffinity= option in
6615 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
6616 itself and the default for all other processes.
6618 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
6619 equivalent), the SELinux database is now reloaded, ensuring that
6620 sockets and other file system objects are generated taking the new
6621 database into account.
6623 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
6624 "quiet" has been changed to imply that instead of
6625 "systemd.show-status=auto". In this mode, only messages about errors
6626 and significant delays in boot are shown on the console.
6628 * The sd-event.h API gained native support for the new Linux "pidfd"
6629 concept. This permits watching processes using file descriptors
6630 instead of PID numbers, which fixes a number of races and makes
6631 process supervision more robust and efficient. All of systemd's
6632 components will now use pidfds if the kernel supports it for process
6633 watching, with the exception of PID 1 itself, unfortunately. We hope
6634 to move PID 1 to exclusively using pidfds too eventually, but this
6635 requires some more kernel work first. (Background: PID 1 watches
6636 processes using waitid() with the P_ALL flag, and that does not play
6637 together nicely with pidfds yet.)
6639 * Closely related to this, the sd-event.h API gained two new calls
6640 sd_event_source_send_child_signal() (for sending a signal to a
6641 watched process) and sd_event_source_get_child_process_own() (for
6642 marking a process so that it is killed automatically whenever the
6643 event source watching it is freed).
6645 * systemd-networkd gained support for configuring Token Bucket Filter
6646 (TBF) parameters in its qdisc configuration support. Similarly,
6647 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
6648 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
6650 * systemd-networkd gained support for Intermediate Functional Block
6651 (IFB) network devices.
6653 * systemd-networkd gained support for configuring multi-path IP routes,
6654 using the new MultiPathRoute= setting in the [Route] section.
6656 * systemd-networkd's DHCPv4 client has been updated to support a new
6657 SendDecline= option. If enabled, duplicate address detection is done
6658 after a DHCP offer is received from the server. If a conflict is
6659 detected, the address is declined. The DHCPv4 client also gained
6660 support for a new RouteMTUBytes= setting that allows to configure the
6661 MTU size to be used for routes generated from DHCPv4 leases.
6663 * The PrefixRoute= setting in systemd-networkd's [Address] section of
6664 .network files has been deprecated, and replaced by AddPrefixRoute=,
6665 with its sense inverted.
6667 * The Gateway= setting of [Route] sections of .network files gained
6668 support for a special new value "_dhcp". If set, the configured
6669 static route uses the gateway host configured via DHCP.
6671 * New User= and SuppressPrefixLength= settings have been implemented
6672 for the [RoutingPolicyRule] section of .network files to configure
6673 source routing based on UID ranges and prefix length, respectively.
6675 * The Type= match property of .link files has been generalized to
6676 always match the device type shown by 'networkctl status', even for
6677 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
6680 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
6681 D-Bus message object as "sensitive". Those objects are erased from
6682 memory when they are freed. This concept is intended to be used for
6683 messages that contain security sensitive data. A new flag
6684 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
6685 in sd-bus vtables, causing any incoming and outgoing messages of
6686 those methods to be implicitly marked as "sensitive".
6688 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
6689 contents of a message (or parts thereof) to standard output for
6692 * systemd-sysusers gained support for creating users with the primary
6693 group named differently than the user.
6695 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
6696 gained support for growing XFS partitions. Previously it supported
6697 only ext4 and btrfs partitions.
6699 * The support for /etc/crypttab gained a new x-initrd.attach option. If
6700 set, the specified encrypted volume is unlocked already in the
6701 initrd. This concept corresponds to the x-initrd.mount option in
6704 * systemd-cryptsetup gained native support for unlocking encrypted
6705 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
6706 encryption of volumes to YubiKeys. This is exposed in the new
6707 pkcs11-uri= option in /etc/crypttab.
6709 * The /etc/fstab support in systemd now supports two new mount options
6710 x-systemd.{required,wanted}-by=, for explicitly configuring the units
6711 that the specified mount shall be pulled in by, in place of
6712 the usual local-fs.target/remote-fs.target.
6714 * The https://systemd.io/ web site has been relaunched, directly
6715 populated with most of the documentation included in the systemd
6716 repository. systemd also acquired a new logo, thanks to Tobias
6719 * systemd-udevd gained support for managing "alternative" network
6720 interface names, as supported by new Linux kernels. For the first
6721 time this permits assigning multiple (and longer!) names to a network
6722 interface. systemd-udevd will now by default assign the names
6723 generated via all supported naming schemes to each interface. This
6724 may be further tweaked with .link files and the AlternativeName= and
6725 AlternativeNamesPolicy= settings. Other components of systemd have
6726 been updated to support the new alternative names wherever
6727 appropriate. For example, systemd-nspawn will now generate
6728 alternative interface names for the host-facing side of container
6729 veth links based on the full container name without truncation.
6731 * systemd-nspawn interface naming logic has been updated in another way
6732 too: if the main interface name (i.e. as opposed to new-style
6733 "alternative" names) based on the container name is truncated, a
6734 simple hashing scheme is used to give different interface names to
6735 multiple containers whose names all begin with the same prefix. Since
6736 this changes the primary interface names pointing to containers if
6737 truncation happens, the old scheme may still be requested by
6738 selecting an older naming scheme, via the net.naming_scheme= kernel
6739 command line option.
6741 * PrivateUsers= in service files now works in services run by the
6742 systemd --user per-user instance of the service manager.
6744 * A new per-service sandboxing option ProtectClock= has been added that
6745 locks down write access to the system clock. It takes away device
6746 node access to /dev/rtc as well as the system calls that set the
6747 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
6748 Note that this option does not affect access to auxiliary services
6749 that allow changing the clock, for example access to
6752 * The systemd-id128 tool gained a new "show" verb for listing or
6753 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
6754 GPT partition table types.
6756 * The Discoverable Partitions Specification has been updated to support
6757 /var and /var/tmp partition discovery. Support for this has been
6758 added to systemd-gpt-auto-generator. For details see:
6760 https://systemd.io/DISCOVERABLE_PARTITIONS
6762 * "systemctl list-unit-files" has been updated to show a new column
6763 with the suggested enablement state based on the vendor preset files
6764 for the respective units.
6766 * "systemctl" gained a new option "--with-dependencies". If specified
6767 commands such as "systemctl status" or "systemctl cat" will now show
6768 all specified units along with all units they depend on.
6770 * networkctl gained support for showing per-interface logs in its
6773 * systemd-networkd-wait-online gained support for specifying the maximum
6774 operational state to wait for, and to wait for interfaces to
6777 * The [Match] section of .link and .network files now supports a new
6778 option PermanentMACAddress= which may be used to check against the
6779 permanent MAC address of a network device even if a randomized MAC
6782 * The [TrafficControlQueueingDiscipline] section in .network files has
6783 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
6784 dropped from the individual setting names.
6786 * Any .link and .network files that have an empty [Match] section (this
6787 also includes empty and commented-out files) will now be
6788 rejected. systemd-udev and systemd-networkd started warning about
6789 such files in version 243.
6791 * systemd-logind will now validate access to the operation of changing
6792 the virtual terminal via a polkit action. By default, only users
6793 with at least one session on a local VT are granted permission.
6795 * When systemd sets up PAM sessions that invoked service processes
6796 shall run in, the pam_setcred() API is now invoked, thus permitting
6797 PAM modules to set additional credentials for the processes.
6799 * portablectl attach/detach verbs now accept --now and --enable options
6800 to combine attachment with enablement and invocation, or detachment
6801 with stopping and disablement.
6803 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
6804 fixed, which in turn exposed bugs in unit configuration of services
6805 which have Type=oneshot and should only run once, but do not have
6806 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
6807 service may be started again after exiting successfully, for example
6808 as a dependency in another transaction. Affected services included
6809 some internal systemd services (most notably
6810 systemd-vconsole-setup.service, which was updated to have
6811 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
6812 plymouth has been suitably updated or patched before upgrading to
6813 this systemd release. See
6814 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
6815 additional discussion.
6817 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
6818 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
6819 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
6820 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
6821 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
6822 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
6823 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
6824 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
6825 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
6826 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
6827 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
6828 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
6829 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
6830 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
6831 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
6832 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
6833 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
6834 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
6835 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
6836 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
6837 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
6838 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
6839 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
6840 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
6843 – Warsaw, 2020-03-06
6847 * Support for the cpuset cgroups v2 controller has been added.
6848 Processes may be restricted to specific CPUs using the new
6849 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
6850 AllowedMemoryNodes= setting.
6852 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
6853 now be configured using a new RestartKillSignal= setting. This
6854 allows units which signals to request termination to implement
6855 different behaviour when stopping in preparation for a restart.
6857 * "systemctl clean" may now be used also for socket, mount, and swap
6860 * systemd will also read configuration options from the EFI variable
6861 SystemdOptions. This may be used to configure systemd behaviour when
6862 modifying the kernel command line is inconvenient, but configuration
6863 on disk is read too late, for example for the options related to
6864 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
6865 set the EFI variable.
6867 * systemd will now disable printk ratelimits in early boot. This should
6868 allow us to capture more logs from the early boot phase where normal
6869 storage is not available and the kernel ring buffer is used for
6870 logging. Configuration on the kernel command line has higher priority
6871 and overrides the systemd setting.
6873 systemd programs which log to /dev/kmsg directly use internal
6874 ratelimits to prevent runaway logging. (Normally this is only used
6875 during early boot, so in practice this change has very little
6878 * Unit files now support top level dropin directories of the form
6879 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
6880 that affects all corresponding unit files.
6882 * systemctl gained support for 'stop --job-mode=triggering' which will
6883 stop the specified unit and any units which could trigger it.
6885 * Unit status display now includes units triggering and triggered by
6886 the unit being shown.
6888 * The RuntimeMaxSec= setting is now supported by scopes, not just
6889 .service units. This is particularly useful for PAM sessions which
6890 create a scope unit for the user login. systemd.runtime_max_sec=
6891 setting may used with the pam_systemd module to limit the duration
6892 of the PAM session, for example for time-limited logins.
6894 * A new @pkey system call group is now defined to make it easier to
6895 allow-list memory protection syscalls for containers and services
6896 which need to use them.
6898 * systemd-udevd: removed the 30s timeout for killing stale workers on
6899 exit. systemd-udevd now waits for workers to finish. The hard-coded
6900 exit timeout of 30s was too short for some large installations, where
6901 driver initialization could be prematurely interrupted during initrd
6902 processing if the root file system had been mounted and init was
6903 preparing to switch root. If udevd is run without systemd and workers
6904 are hanging while udevd receives an exit signal, udevd will now exit
6905 when udev.event_timeout is reached for the last hanging worker. With
6906 systemd, the exit timeout can additionally be configured using
6907 TimeoutStopSec= in systemd-udevd.service.
6909 * udev now provides a program (fido_id) that identifies FIDO CTAP1
6910 ("U2F")/CTAP2 security tokens based on the usage declared in their
6911 report and descriptor and outputs suitable environment variables.
6912 This replaces the externally maintained allow lists of all known
6913 security tokens that were used previously.
6915 * Automatically generated autosuspend udev rules for allow-listed
6916 devices have been imported from the Chromium OS project. This should
6917 improve power saving with many more devices.
6919 * udev gained a new "CONST{key}=value" setting that allows matching
6920 against system-wide constants without forking a helper binary.
6921 Currently "arch" and "virt" keys are supported.
6923 * udev now opens CDROMs in non-exclusive mode when querying their
6924 capabilities. This should fix issues where other programs trying to
6925 use the CDROM cannot gain access to it, but carries a risk of
6926 interfering with programs writing to the disk, if they did not open
6927 the device in exclusive mode as they should.
6929 * systemd-networkd does not create a default route for IPv4 link local
6930 addressing anymore. The creation of the route was unexpected and was
6931 breaking routing in various cases, but people who rely on it being
6932 created implicitly will need to adjust. Such a route may be requested
6933 with DefaultRouteOnDevice=yes.
6935 Similarly, systemd-networkd will not assign a link-local IPv6 address
6936 when IPv6 link-local routing is not enabled.
6938 * Receive and transmit buffers may now be configured on links with
6939 the new RxBufferSize= and TxBufferSize= settings.
6941 * systemd-networkd may now advertise additional IPv6 routes. A new
6942 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
6945 * systemd-networkd may now configure "next hop" routes using the
6946 [NextHop] section and Gateway= and Id= settings.
6948 * systemd-networkd will now retain DHCP config on restarts by default
6949 (but this may be overridden using the KeepConfiguration= setting).
6950 The default for SendRelease= has been changed to true.
6952 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
6953 received from the server.
6955 The client will use the received SIP server list if UseSIP=yes is
6958 The client may be configured to request specific options from the
6959 server using a new RequestOptions= setting.
6961 The client may be configured to send arbitrary options to the server
6962 using a new SendOption= setting.
6964 A new IPServiceType= setting has been added to configure the "IP
6965 service type" value used by the client.
6967 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
6968 request prefix hints in the DHCPv6 solicitation.
6970 * The DHCPv4 server may be configured to send arbitrary options using
6971 a new SendOption= setting.
6973 * The DHCPv4 server may now be configured to emit SIP server list using
6974 the new EmitSIP= and SIP= settings.
6976 * systemd-networkd and networkctl may now renew DHCP leases on demand.
6977 networkctl has a new 'networkctl renew' verb.
6979 * systemd-networkd may now reconfigure links on demand. networkctl
6980 gained two new verbs: "reload" will reload the configuration, and
6981 "reconfigure DEVICE…" will reconfigure one or more devices.
6983 * .network files may now match on SSID and BSSID of a wireless network,
6984 i.e. the access point name and hardware address using the new SSID=
6985 and BSSID= options. networkctl will display the current SSID and
6986 BSSID for wireless links.
6988 .network files may also match on the wireless network type using the
6989 new WLANInterfaceType= option.
6991 * systemd-networkd now includes default configuration that enables
6992 link-local addressing when connected to an ad-hoc wireless network.
6994 * systemd-networkd may configure the Traffic Control queueing
6995 disciplines in the kernel using the new
6996 [TrafficControlQueueingDiscipline] section and Parent=,
6997 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
6998 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
6999 NetworkEmulatorDuplicateRate= settings.
7001 * systemd-tmpfiles gained a new w+ setting to append to files.
7003 * systemd-analyze dump will now report when the memory configuration in
7004 the kernel does not match what systemd has configured (usually,
7005 because some external program has modified the kernel configuration
7008 * systemd-analyze gained a new --base-time= switch instructs the
7009 'calendar' verb to resolve times relative to that timestamp instead
7010 of the present time.
7012 * journalctl --update-catalog now produces deterministic output (making
7013 reproducible image builds easier).
7015 * A new devicetree-overlay setting is now documented in the Boot Loader
7018 * The default value of the WatchdogSec= setting used in systemd
7019 services (the ones bundled with the project itself) may be set at
7020 configuration time using the -Dservice-watchdog= setting. If set to
7021 empty, the watchdogs will be disabled.
7023 * systemd-resolved validates IP addresses in certificates now when GnuTLS
7026 * libcryptsetup >= 2.0.1 is now required.
7028 * A configuration option -Duser-path= may be used to override the $PATH
7029 used by the user service manager. The default is again to use the same
7030 path as the system manager.
7032 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
7033 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
7036 * Service units gained a new sandboxing option ProtectKernelLogs= which
7037 makes sure the program cannot get direct access to the kernel log
7038 buffer anymore, i.e. the syslog() system call (not to be confused
7039 with the API of the same name in libc, which is not affected), the
7040 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
7041 inaccessible to the service. It's recommended to enable this setting
7042 for all services that should not be able to read from or write to the
7043 kernel log buffer, which are probably almost all.
7045 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
7046 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
7047 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
7048 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
7049 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
7050 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
7051 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
7052 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
7053 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
7054 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
7055 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
7056 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
7057 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
7058 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
7059 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
7060 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
7061 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
7062 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
7063 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
7064 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
7065 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
7066 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
7067 Zach Smith, Zbigniew Jędrzejewski-Szmek
7069 – Warsaw, 2019-11-29
7073 * This release enables unprivileged programs (i.e. requiring neither
7074 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
7075 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
7076 kernel for the whole UNIX group range, i.e. all processes. This
7077 change should be reasonably safe, as the kernel support for it was
7078 specifically implemented to allow safe access to ICMP Echo for
7079 processes lacking any privileges. If this is not desirable, it can be
7080 disabled again by setting the parameter to "1 0".
7082 * Previously, filters defined with SystemCallFilter= would have the
7083 effect that any calling of an offending system call would terminate
7084 the calling thread. This behaviour never made much sense, since
7085 killing individual threads of unsuspecting processes is likely to
7086 create more problems than it solves. With this release the default
7087 action changed from killing the thread to killing the whole
7088 process. For this to work correctly both a kernel version (>= 4.14)
7089 and a libseccomp version (>= 2.4.0) supporting this new seccomp
7090 action is required. If an older kernel or libseccomp is used the old
7091 behaviour continues to be used. This change does not affect any
7092 services that have no system call filters defined, or that use
7093 SystemCallErrorNumber= (and thus see EPERM or another error instead
7094 of being killed when calling an offending system call). Note that
7095 systemd documentation always claimed that the whole process is
7096 killed. With this change behaviour is thus adjusted to match the
7099 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
7100 4194304 by default, i.e. the full 22bit range the kernel allows, up
7101 from the old 16-bit range. This should improve security and
7102 robustness, as PID collisions are made less likely (though certainly
7103 still possible). There are rumours this might create compatibility
7104 problems, though at this moment no practical ones are known to
7105 us. Downstream distributions are hence advised to undo this change in
7106 their builds if they are concerned about maximum compatibility, but
7107 for everybody else we recommend leaving the value bumped. Besides
7108 improving security and robustness this should also simplify things as
7109 the maximum number of allowed concurrent tasks was previously bounded
7110 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
7111 only a single knob is left ("kernel.threads-max"). There have been
7112 concerns that usability is affected by this change because larger PID
7113 numbers are harder to type, but we believe the change from 5 digits
7114 to 7 digits doesn't hamper usability.
7116 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
7117 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
7118 hierarchically set default memory protection values for a particular
7119 subtree of the unit hierarchy.
7121 * Memory protection directives can now take a value of zero, allowing
7122 explicit opting out of a default value propagated by an ancestor.
7124 * systemd now defaults to the "unified" cgroup hierarchy setup during
7125 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
7126 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
7127 change reflects the fact that cgroupsv2 support has matured
7128 substantially in both systemd and in the kernel, and is clearly the
7129 way forward. Downstream production distributions might want to
7130 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
7131 their builds as unfortunately the popular container managers have not
7132 caught up with the kernel API changes.
7134 * Man pages are not built by default anymore (html pages were already
7135 disabled by default), to make development builds quicker. When
7136 building systemd for a full installation with documentation, meson
7137 should be called with -Dman=true and/or -Dhtml=true as appropriate.
7138 The default was changed based on the assumption that quick one-off or
7139 repeated development builds are much more common than full optimized
7140 builds for installation, and people need to pass various other
7141 options to when doing "proper" builds anyway, so the gain from making
7142 development builds quicker is bigger than the one time disruption for
7145 Two scripts are created in the *build* directory to generate and
7146 preview man and html pages on demand, e.g.:
7148 build/man/man systemctl
7149 build/man/html systemd.index
7151 * libidn2 is used by default if both libidn2 and libidn are installed.
7152 Please use -Dlibidn=true if libidn is preferred.
7154 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
7155 big-endian machines. Before, bytes were written and read in native
7156 machine order as exposed by the native libc __cpu_mask interface.
7157 Now, little-endian order is always used (CPUs 0–7 are described by
7158 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
7159 This change fixes D-Bus calls that cross endianness boundary.
7161 The presentation format used for CPUAffinity= by "systemctl show" and
7162 "systemd-analyze dump" is changed to present CPU indices instead of
7163 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
7164 shown as CPUAffinity=03000000000000000000000000000… (on
7165 little-endian) or CPUAffinity=00000000000000300000000000000… (on
7166 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
7167 input format. The maximum integer that will be printed in the new
7168 format is 8191 (four digits), while the old format always used a very
7169 long number (with the length varying by architecture), so they can be
7170 unambiguously distinguished.
7172 * /usr/sbin/halt.local is no longer supported. Implementation in
7173 distributions was inconsistent and it seems this functionality was
7176 To replace this functionality, users should:
7177 - either define a new unit and make it a dependency of final.target
7178 (systemctl add-wants final.target my-halt-local.service)
7179 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
7180 and ensure that it accepts "halt", "poweroff", "reboot", and
7181 "kexec" as an argument, see the description in systemd-shutdown(8).
7183 * When a [Match] section in .link or .network file is empty (contains
7184 no match patterns), a warning will be emitted. Please add any "match
7185 all" pattern instead, e.g. OriginalName=* or Name=* in case all
7186 interfaces should really be matched.
7188 * A new setting NUMAPolicy= may be used to set process memory
7189 allocation policy. This setting can be specified in
7190 /etc/systemd/system.conf and hence will set the default policy for
7191 PID1. The default policy can be overridden on a per-service
7192 basis. The related setting NUMAMask= is used to specify NUMA node
7193 mask that should be associated with the selected policy.
7195 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
7196 generates when processes it manages are reaching their memory limits,
7197 and will place their units in a special state, and optionally kill or
7198 stop the whole unit.
7200 * The service manager will now expose bus properties for the IO
7201 resources used by units. This information is also shown in "systemctl
7202 status" now (for services that have IOAccounting=yes set). Moreover,
7203 the IO accounting data is included in the resource log message
7204 generated whenever a unit stops.
7206 * Units may now configure an explicit timeout to wait for when killed
7207 with SIGABRT, for example when a service watchdog is hit. Previously,
7208 the regular TimeoutStopSec= timeout was applied in this case too —
7209 now a separate timeout may be set using TimeoutAbortSec=.
7211 * Services may now send a special WATCHDOG=trigger message with
7212 sd_notify() to trigger an immediate "watchdog missed" event, and thus
7213 trigger service termination. This is useful both for testing watchdog
7214 handling, but also for defining error paths in services, that shall
7215 be handled the same way as watchdog events.
7217 * There are two new per-unit settings IPIngressFilterPath= and
7218 IPEgressFilterPath= which allow configuration of a BPF program
7219 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
7220 to apply to the IP packet ingress/egress path of all processes of a
7221 unit. This is useful to allow running systemd services with BPF
7222 programs set up externally.
7224 * systemctl gained a new "clean" verb for removing the state, cache,
7225 runtime or logs directories of a service while it is terminated. The
7226 new verb may also be used to remove the state maintained on disk for
7227 timer units that have Persistent= configured.
7229 * During the last phase of shutdown systemd will now automatically
7230 increase the log level configured in the "kernel.printk" sysctl so
7231 that any relevant loggable events happening during late shutdown are
7232 made visible. Previously, loggable events happening so late during
7233 shutdown were generally lost if the "kernel.printk" sysctl was set to
7234 high thresholds, as regular logging daemons are terminated at that
7235 time and thus nothing is written to disk.
7237 * If processes terminated during the last phase of shutdown do not exit
7238 quickly systemd will now show their names after a short time, to make
7239 debugging easier. After a longer timeout they are forcibly killed,
7242 * journalctl (and the other tools that display logs) will now highlight
7243 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
7244 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
7245 are now shown in blue color, to separate them visually from regular
7246 logs. References to configuration files are now turned into clickable
7247 links on terminals that support that.
7249 * systemd-journald will now stop logging to /var/log/journal during
7250 shutdown when /var/ is on a separate mount, so that it can be
7251 unmounted safely during shutdown.
7253 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
7255 * systemd-resolved "Cache=" configuration option in resolved.conf has
7256 been extended to also accept the 'no-negative' value. Previously,
7257 only a boolean option was allowed (yes/no), having yes as the
7258 default. If this option is set to 'no-negative', negative answers are
7259 not cached while the old cache heuristics are used positive answers.
7260 The default remains unchanged.
7262 * The predictable naming scheme for network devices now supports
7263 generating predictable names for "netdevsim" devices.
7265 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
7268 Those two changes form a new net.naming_scheme= entry. Distributions
7269 which want to preserve naming stability may want to set the
7270 -Ddefault-net-naming-scheme= configuration option.
7272 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
7273 interfaces natively.
7275 * systemd-networkd's bridge FDB support now allows configuration of a
7276 destination address for each entry (Destination=), as well as the
7277 VXLAN VNI (VNI=), as well as an option to declare what an entry is
7278 associated with (AssociatedWith=).
7280 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
7281 option for configuring the maximum number of DHCP lease requests. It
7282 also learnt a new BlackList= option for deny-listing DHCP servers (a
7283 similar setting has also been added to the IPv6 RA client), as well
7284 as a SendRelease= option for configuring whether to send a DHCP
7285 RELEASE message when terminating.
7287 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
7288 separately in the [DHCPv4] and [DHCPv6] sections.
7290 * systemd-networkd's DHCP support will now optionally create an
7291 implicit host route to the DNS server specified in the DHCP lease, in
7292 addition to the routes listed explicitly in the lease. This should
7293 ensure that in multi-homed systems DNS traffic leaves the systems on
7294 the interface that acquired the DNS server information even if other
7295 routes such as default routes exist. This behaviour may be turned on
7296 with the new RoutesToDNS= option.
7298 * systemd-networkd's VXLAN support gained a new option
7299 GenericProtocolExtension= for enabling VXLAN Generic Protocol
7300 Extension support, as well as IPDoNotFragment= for setting the IP
7301 "Don't fragment" bit on outgoing packets. A similar option has been
7302 added to the GENEVE support.
7304 * In systemd-networkd's [Route] section you may now configure
7305 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
7306 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
7307 propagation. The Type= setting now supports local, broadcast,
7308 anycast, multicast, any, xresolve routes, too.
7310 * systemd-networkd's [Network] section learnt a new option
7311 DefaultRouteOnDevice= for automatically configuring a default route
7312 onto the network device.
7314 * systemd-networkd's bridging support gained two new options ProxyARP=
7315 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
7316 MulticastRouter= for configuring multicast routing behaviour. A new
7317 option MulticastIGMPVersion= may be used to change bridge's multicast
7318 Internet Group Management Protocol (IGMP) version.
7320 * systemd-networkd's FooOverUDP support gained the ability to configure
7321 local and peer IP addresses via Local= and Peer=. A new option
7322 PeerPort= may be used to configure the peer's IP port.
7324 * systemd-networkd's TUN support gained a new setting VnetHeader= for
7325 tweaking Generic Segment Offload support.
7327 * The address family for policy rules may be specified using the new
7328 Family= option in the [RoutingPolicyRule] section.
7330 * networkctl gained a new "delete" command for removing virtual network
7331 devices, as well as a new "--stats" switch for showing device
7334 * networkd.conf gained a new setting SpeedMeter= and
7335 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
7336 measured speed may be shown by 'networkctl status'.
7338 * "networkctl status" now displays MTU and queue lengths, and more
7339 detailed information about VXLAN and bridge devices.
7341 * systemd-networkd's .network and .link files gained a new Property=
7342 setting in the [Match] section, to match against devices with
7343 specific udev properties.
7345 * systemd-networkd's tunnel support gained a new option
7346 AssignToLoopback= for selecting whether to use the loopback device
7347 "lo" as underlying device.
7349 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
7350 been renamed to LinkLayerAddress=, and it now allows configuration of
7353 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
7354 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
7355 IPv6 configuration (static or DHCPv6) was found for a given
7356 interface. It will not touch the sysctl otherwise.
7358 * The order of entries is $PATH used by the user manager instance was
7359 changed to put bin/ entries before the corresponding sbin/ entries.
7360 It is recommended to not rely on this order, and only ever have one
7361 binary with a given name in the system paths under /usr.
7363 * A new tool systemd-network-generator has been added that may generate
7364 .network, .netdev and .link files from IP configuration specified on
7365 the kernel command line in the format used by Dracut.
7367 * The CriticalConnection= setting in .network files is now deprecated,
7368 and replaced by a new KeepConfiguration= setting which allows more
7369 detailed configuration of the IP configuration to keep in place.
7371 * systemd-analyze gained a few new verbs:
7373 - "systemd-analyze timestamp" parses and converts timestamps. This is
7374 similar to the existing "systemd-analyze calendar" command which
7375 does the same for recurring calendar events.
7377 - "systemd-analyze timespan" parses and converts timespans (i.e.
7378 durations as opposed to points in time).
7380 - "systemd-analyze condition" will parse and test ConditionXYZ=
7383 - "systemd-analyze exit-status" will parse and convert exit status
7384 codes to their names and back.
7386 - "systemd-analyze unit-files" will print a list of all unit
7387 file paths and unit aliases.
7389 * SuccessExitStatus=, RestartPreventExitStatus=, and
7390 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
7391 is equivalent to "65"). Those exit status name mappings may be
7392 displayed with the systemd-analyze exit-status verb describe above.
7394 * systemd-logind now exposes a per-session SetBrightness() bus call,
7395 which may be used to securely change the brightness of a kernel
7396 brightness device, if it belongs to the session's seat. By using this
7397 call unprivileged clients can make changes to "backlight" and "leds"
7398 devices securely with strict requirements on session membership.
7399 Desktop environments may use this to generically make brightness
7400 changes to such devices without shipping private SUID binaries or
7401 udev rules for that purpose.
7403 * "udevadm info" gained a --wait-for-initialization switch to wait for
7404 a device to be initialized.
7406 * systemd-hibernate-resume-generator will now look for resumeflags= on
7407 the kernel command line, which is similar to rootflags= and may be
7408 used to configure device timeout for the hibernation device.
7410 * sd-event learnt a new API call sd_event_source_disable_unref() for
7411 disabling and unref'ing an event source in a single function. A
7412 related call sd_event_source_disable_unrefp() has been added for use
7413 with gcc's cleanup extension.
7415 * The sd-id128.h public API gained a new definition
7416 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
7419 * "busctl introspect" gained a new switch --xml-interface for dumping
7420 XML introspection data unmodified.
7422 * PID 1 may now show the unit name instead of the unit description
7423 string in its status output during boot. This may be configured in
7424 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
7425 kernel command line option systemd.status_unit_format=.
7427 * PID 1 now understands a new option KExecWatchdogSec= in
7428 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
7429 Previously watchdog functionality was only available for regular
7430 reboots. The new setting defaults to off, because we don't know in
7431 the general case if the watchdog will be reset after kexec (some
7432 drivers do reset it, but not all), and the new userspace might not be
7433 configured to handle the watchdog.
7435 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
7436 RebootWatchdogSec= to more clearly communicate what it is about. The
7437 old name is still accepted for compatibility.
7439 * The systemd.debug_shell kernel command line option now optionally
7440 takes a tty name to spawn the debug shell on, which allows a
7441 different tty to be selected than the built-in default.
7443 * Service units gained a new ExecCondition= setting which will run
7444 before ExecStartPre= and either continue execution of the unit (for
7445 clean exit codes), stop execution without marking the unit failed
7446 (for exit codes 1 through 254), or stop execution and fail the unit
7447 (for exit code 255 or abnormal termination).
7449 * A new service systemd-pstore.service has been added that pulls data
7450 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
7453 * timedatectl gained new verbs for configuring per-interface NTP
7454 service configuration for systemd-timesyncd.
7456 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
7457 2019. (You can set non-UTF-8 locales though, if you know their name.)
7459 * If variable assignments in sysctl.d/ files are prefixed with "-" any
7460 failures to apply them are now ignored.
7462 * systemd-random-seed.service now optionally credits entropy when
7463 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
7464 true for the service to enable this behaviour, but please consult the
7465 documentation first, since this comes with a couple of caveats.
7467 * systemd-random-seed.service is now a synchronization point for full
7468 initialization of the kernel's entropy pool. Services that require
7469 /dev/urandom to be correctly initialized should be ordered after this
7472 * The systemd-boot boot loader has been updated to optionally maintain
7473 a random seed file in the EFI System Partition (ESP). During the boot
7474 phase, this random seed is read and updated with a new seed
7475 cryptographically derived from it. Another derived seed is passed to
7476 the OS. The latter seed is then credited to the kernel's entropy pool
7477 very early during userspace initialization (from PID 1). This allows
7478 systems to boot up with a fully initialized kernel entropy pool from
7479 earliest boot on, and thus entirely removes all entropy pool
7480 initialization delays from systems using systemd-boot. Special care
7481 is taken to ensure different seeds are derived on system images
7482 replicated to multiple systems. "bootctl status" will show whether
7483 a seed was received from the boot loader.
7485 * bootctl gained two new verbs:
7487 - "bootctl random-seed" will generate the file in ESP and an EFI
7488 variable to allow a random seed to be passed to the OS as described
7491 - "bootctl is-installed" checks whether systemd-boot is currently
7494 * bootctl will warn if it detects that boot entries are misconfigured
7495 (for example if the kernel image was removed without purging the
7498 * A new document has been added describing systemd's use and support
7499 for the kernel's entropy pool subsystem:
7501 https://systemd.io/RANDOM_SEEDS
7503 * When the system is hibernated the swap device to write the
7504 hibernation image to is now automatically picked from all available
7505 swap devices, preferring the swap device with the highest configured
7506 priority over all others, and picking the device with the most free
7507 space if there are multiple devices with the highest priority.
7509 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
7510 option that permits selecting the timeout how long to wait for a
7511 device with an encryption key before asking for the password.
7513 * IOWeight= has learnt to properly set the IO weight when using the
7514 BFQ scheduler officially found in kernels 5.0+.
7516 * A new mailing list has been created for reporting of security issues:
7517 systemd-security@redhat.com. For mode details, see
7518 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
7520 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
7521 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
7522 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
7523 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
7524 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
7525 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
7526 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
7527 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
7528 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
7529 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
7530 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
7531 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
7532 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
7533 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
7534 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
7535 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
7536 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
7537 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
7538 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
7539 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
7540 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
7541 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
7542 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
7543 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
7544 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
7545 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
7546 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
7547 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
7548 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
7549 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
7551 – Camerino, 2019-09-03
7555 * In .link files, MACAddressPolicy=persistent (the default) is changed
7556 to cover more devices. For devices like bridges, tun, tap, bond, and
7557 similar interfaces that do not have other identifying information,
7558 the interface name is used as the basis for persistent seed for MAC
7559 and IPv4LL addresses. The way that devices that were handled
7560 previously is not changed, and this change is about covering more
7561 devices then previously by the "persistent" policy.
7563 MACAddressPolicy=random may be used to force randomized MACs and
7564 IPv4LL addresses for a device if desired.
7566 Hint: the log output from udev (at debug level) was enhanced to
7567 clarify what policy is followed and which attributes are used.
7568 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
7569 may be used to view this.
7571 Hint: if a bridge interface is created without any slaves, and gains
7572 a slave later, then now the bridge does not inherit slave's MAC.
7573 To inherit slave's MAC, for example, create the following file:
7575 # /etc/systemd/network/98-bridge-inherit-mac.link
7580 MACAddressPolicy=none
7583 * The .device units generated by systemd-fstab-generator and other
7584 generators do not automatically pull in the corresponding .mount unit
7585 as a Wants= dependency. This means that simply plugging in the device
7586 will not cause the mount unit to be started automatically. But please
7587 note that the mount unit may be started for other reasons, in
7588 particular if it is part of local-fs.target, and any unit which
7589 (transitively) depends on local-fs.target is started.
7591 * networkctl list/status/lldp now accept globbing wildcards for network
7592 interface names to match against all existing interfaces.
7594 * The $PIDFILE environment variable is set to point the absolute path
7595 configured with PIDFile= for processes of that service.
7597 * The fallback DNS server list was augmented with Cloudflare public DNS
7598 servers. Use `-Ddns-servers=` to set a different fallback.
7600 * A new special target usb-gadget.target will be started automatically
7601 when a USB Device Controller is detected (which means that the system
7602 is a USB peripheral).
7604 * A new unit setting CPUQuotaPeriodSec= assigns the time period
7605 relatively to which the CPU time quota specified by CPUQuota= is
7608 * A new unit setting ProtectHostname= may be used to prevent services
7609 from modifying hostname information (even if they otherwise would
7610 have privileges to do so).
7612 * A new unit setting NetworkNamespacePath= may be used to specify a
7613 namespace for service or socket units through a path referring to a
7614 Linux network namespace pseudo-file.
7616 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
7617 have an effect on .socket units: when used the listening socket is
7618 created within the configured network namespace instead of the host
7621 * ExecStart= command lines in unit files may now be prefixed with ':'
7622 in which case environment variable substitution is
7623 disabled. (Supported for the other ExecXYZ= settings, too.)
7625 * .timer units gained two new boolean settings OnClockChange= and
7626 OnTimezoneChange= which may be used to also trigger a unit when the
7627 system clock is changed or the local timezone is
7628 modified. systemd-run has been updated to make these options easily
7629 accessible from the command line for transient timers.
7631 * Two new conditions for units have been added: ConditionMemory= may be
7632 used to conditionalize a unit based on installed system
7633 RAM. ConditionCPUs= may be used to conditionalize a unit based on
7634 installed CPU cores.
7636 * The @default system call filter group understood by SystemCallFilter=
7637 has been updated to include the new rseq() system call introduced in
7640 * A new time-set.target has been added that indicates that the system
7641 time has been set from a local source (possibly imprecise). The
7642 existing time-sync.target is stronger and indicates that the time has
7643 been synchronized with a precise external source. Services where
7644 approximate time is sufficient should use the new target.
7646 * "systemctl start" (and related commands) learnt a new
7647 --show-transaction option. If specified brief information about all
7648 jobs queued because of the requested operation is shown.
7650 * systemd-networkd recognizes a new operation state 'enslaved', used
7651 (instead of 'degraded' or 'carrier') for interfaces which form a
7652 bridge, bond, or similar, and an new 'degraded-carrier' operational
7653 state used for the bond or bridge master interface when one of the
7654 enslaved devices is not operational.
7656 * .network files learnt the new IgnoreCarrierLoss= option for leaving
7657 networks configured even if the carrier is lost.
7659 * The RequiredForOnline= setting in .network files may now specify a
7660 minimum operational state required for the interface to be considered
7661 "online" by systemd-networkd-wait-online. Related to this
7662 systemd-networkd-wait-online gained a new option --operational-state=
7663 to configure the same, and its --interface= option was updated to
7664 optionally also take an operational state specific for an interface.
7666 * systemd-networkd-wait-online gained a new setting --any for waiting
7667 for only one of the requested interfaces instead of all of them.
7669 * systemd-networkd now implements L2TP tunnels.
7671 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
7672 may be used to cause autonomous and onlink prefixes received in IPv6
7673 Router Advertisements to be ignored.
7675 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
7676 file settings may be used to tweak bridge behaviour.
7678 * The new TripleSampling= option in .network files may be used to
7679 configure CAN triple sampling.
7681 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
7682 used to point to private or preshared key for a WireGuard interface.
7684 * /etc/crypttab now supports the same-cpu-crypt and
7685 submit-from-crypt-cpus options to tweak encryption work scheduling
7688 * systemd-tmpfiles will now take a BSD file lock before operating on a
7689 contents of directory. This may be used to temporarily exclude
7690 directories from aging by taking the same lock (useful for example
7691 when extracting a tarball into /tmp or /var/tmp as a privileged user,
7692 which might create files with really old timestamps, which
7693 nevertheless should not be deleted). For further details, see:
7695 https://systemd.io/TEMPORARY_DIRECTORIES
7697 * systemd-tmpfiles' h line type gained support for the
7698 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
7699 controlling project quota inheritance.
7701 * sd-boot and bootctl now implement support for an Extended Boot Loader
7702 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
7703 addition to the ESP partition mounted to /efi or /boot/efi.
7704 Configuration file fragments, kernels, initrds and other EFI images
7705 to boot will be loaded from both the ESP and XBOOTLDR partitions.
7706 The XBOOTLDR partition was previously described by the Boot Loader
7707 Specification, but implementation was missing in sd-boot. Support for
7708 this concept allows using the sd-boot boot loader in more
7709 conservative scenarios where the boot loader itself is placed in the
7710 ESP but the kernels to boot (and their metadata) in a separate
7713 * A system may now be booted with systemd.volatile=overlay on the
7714 kernel command line, which causes the root file system to be set up
7715 an overlayfs mount combining the root-only root directory with a
7716 writable tmpfs. In this setup, the underlying root device is not
7717 modified, and any changes are lost at reboot.
7719 * Similar, systemd-nspawn can now boot containers with a volatile
7720 overlayfs root with the new --volatile=overlay switch.
7722 * systemd-nspawn can now consume OCI runtime bundles using a new
7723 --oci-bundle= option. This implementation is fully usable, with most
7724 features in the specification implemented, but since this a lot of
7725 new code and functionality, this feature should most likely not
7726 be used in production yet.
7728 * systemd-nspawn now supports various options described by the OCI
7729 runtime specification on the command-line and in .nspawn files:
7730 --inaccessible=/Inaccessible= may be used to mask parts of the file
7731 system tree, --console=/--pipe may be used to configure how standard
7732 input, output, and error are set up.
7734 * busctl learned the `emit` verb to generate D-Bus signals.
7736 * systemd-analyze cat-config may be used to gather and display
7737 configuration spread over multiple files, for example system and user
7738 presets, tmpfiles.d, sysusers.d, udev rules, etc.
7740 * systemd-analyze calendar now takes an optional new parameter
7741 --iterations= which may be used to show a maximum number of iterations
7742 the specified expression will elapse next.
7744 * The sd-bus C API gained support for naming method parameters in the
7747 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
7748 the reboot() system call expects.
7750 * journalctl learnt a new --cursor-file= option that points to a file
7751 from which a cursor should be loaded in the beginning and to which
7752 the updated cursor should be stored at the end.
7754 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
7755 detected by systemd-detect-virt (and may also be used in
7756 ConditionVirtualization=).
7758 * The behaviour of systemd-logind may now be modified with environment
7759 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
7760 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
7761 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
7762 skip the relevant operation completely (when set to false), or to
7763 create a flag file in /run/systemd (when set to true), instead of
7764 actually commencing the real operation when requested. The presence
7765 of /run/systemd/reboot-to-firmware-setup,
7766 /run/systemd/reboot-to-boot-loader-menu, and
7767 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
7768 boot loader implementations to replace some steps logind performs
7769 during reboot with their own operations.
7771 * systemctl can be used to request a reboot into the boot loader menu
7772 or a specific boot loader entry with the new --boot-load-menu= and
7773 --boot-loader-entry= options to a reboot command. (This requires a
7774 boot loader that supports this, for example sd-boot.)
7776 * kernel-install will no longer unconditionally create the output
7777 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
7778 snippets, but will do only if the machine-specific parent directory
7779 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
7780 to create this parent directory during sd-boot installation.
7782 This makes it easier to use kernel-install with plugins which support
7783 a different layout of the bootloader partitions (for example grub2).
7785 * During package installation (with `ninja install`), we would create
7786 symlinks for getty@tty1.service, systemd-networkd.service,
7787 systemd-networkd.socket, systemd-resolved.service,
7788 remote-cryptsetup.target, remote-fs.target,
7789 systemd-networkd-wait-online.service, and systemd-timesyncd.service
7790 in /etc, as if `systemctl enable` was called for those units, to make
7791 the system usable immediately after installation. Now this is not
7792 done anymore, and instead calling `systemctl preset-all` is
7793 recommended after the first installation of systemd.
7795 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
7796 is built on seccomp. When turned on creation of SUID/SGID files is
7799 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
7800 implied if DynamicUser= is turned on for a service. This hardens
7801 these services, so that they neither can benefit from nor create
7802 SUID/SGID executables. This is a minor compatibility breakage, given
7803 that when DynamicUser= was first introduced SUID/SGID behaviour was
7804 unaffected. However, the security benefit of these two options is
7805 substantial, and the setting is still relatively new, hence we opted
7806 to make it mandatory for services with dynamic users.
7808 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
7809 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
7810 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
7811 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
7812 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
7813 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
7814 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
7815 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
7816 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
7817 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
7818 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
7819 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
7820 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
7821 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
7822 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
7823 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
7824 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
7825 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
7827 — Warsaw, 2019-04-11
7831 * The default locale can now be configured at compile time. Otherwise,
7832 a suitable default will be selected automatically (one of C.UTF-8,
7833 en_US.UTF-8, and C).
7835 * The version string shown by systemd and other tools now includes the
7836 git commit hash when built from git. An override may be specified
7837 during compilation, which is intended to be used by distributions to
7838 include the package release information.
7840 * systemd-cat can now filter standard input and standard error streams
7841 for different syslog priorities using the new --stderr-priority=
7844 * systemd-journald and systemd-journal-remote reject entries which
7845 contain too many fields (CVE-2018-16865) and set limits on the
7846 process' command line length (CVE-2018-16864).
7848 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
7851 * A new network device NamePolicy "keep" is implemented for link files,
7852 and used by default in 99-default.link (the fallback configuration
7853 provided by systemd). With this policy, if the network device name
7854 was already set by userspace, the device will not be renamed again.
7855 This matches the naming scheme that was implemented before
7856 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
7857 is also enabled by default, even if not specified. Effectively, this
7858 means that if naming-scheme >= 240 is specified, network devices will
7859 be renamed according to the configuration, even if they have been
7860 renamed already, if "keep" is not specified as the naming policy in
7861 the .link file. The 99-default.link file provided by systemd includes
7862 "keep" for backwards compatibility, but it is recommended for user
7863 installed .link files to *not* include it.
7865 The "kernel" policy, which keeps kernel names declared to be
7866 "persistent", now works again as documented.
7868 * kernel-install script now optionally takes the paths to one or more
7869 initrd files, and passes them to all plugins.
7871 * The mincore() system call has been dropped from the @system-service
7872 system call filter group, as it is pretty exotic and may potentially
7873 used for side-channel attacks.
7875 * -fPIE is dropped from compiler and linker options. Please specify
7876 -Db_pie=true option to meson to build position-independent
7877 executables. Note that the meson option is supported since meson-0.49.
7879 * The fs.protected_regular and fs.protected_fifos sysctls, which were
7880 added in Linux 4.19 to make some data spoofing attacks harder, are
7881 now enabled by default. While this will hopefully improve the
7882 security of most installations, it is technically a backwards
7883 incompatible change; to disable these sysctls again, place the
7884 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
7886 fs.protected_regular = 0
7887 fs.protected_fifos = 0
7889 Note that the similar hardlink and symlink protection has been
7890 enabled since v199, and may be disabled likewise.
7892 * The files read from the EnvironmentFile= setting in unit files now
7893 parse backslashes inside quotes literally, matching the behaviour of
7896 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
7897 now automatically become NOPs when run in a chroot() environment.
7899 * The tmpfiles.d/ "C" line type will now copy directory trees not only
7900 when the destination is so far missing, but also if it already exists
7901 as a directory and is empty. This is useful to cater for systems
7902 where directory trees are put together from multiple separate mount
7903 points but otherwise empty.
7905 * A new function sd_bus_close_unref() (and the associated
7906 sd_bus_close_unrefp()) has been added to libsystemd, that combines
7907 sd_bus_close() and sd_bus_unref() in one.
7909 * udevadm control learnt a new option for --ping for testing whether a
7910 systemd-udevd instance is running and reacting.
7912 * udevadm trigger learnt a new option for --wait-daemon for waiting
7913 systemd-udevd daemon to be initialized.
7915 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
7916 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
7917 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
7918 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
7919 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
7920 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
7921 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
7922 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
7923 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
7924 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
7925 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
7926 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
7927 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
7928 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
7929 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
7930 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
7931 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
7933 — Berlin, 2019-02-14
7937 * NoNewPrivileges=yes has been set for all long-running services
7938 implemented by systemd. Previously, this was problematic due to
7939 SELinux (as this would also prohibit the transition from PID1's label
7940 to the service's label). This restriction has since been lifted, but
7941 an SELinux policy update is required.
7942 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
7944 * DynamicUser=yes is dropped from systemd-networkd.service,
7945 systemd-resolved.service and systemd-timesyncd.service, which was
7946 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
7947 and since v236 for systemd-timesyncd.service. The users and groups
7948 systemd-network, systemd-resolve and systemd-timesync are created
7949 by systemd-sysusers again. Distributors or system administrators
7950 may need to create these users and groups if they not exist (or need
7951 to re-enable DynamicUser= for those units) while upgrading systemd.
7952 Also, the clock file for systemd-timesyncd may need to move from
7953 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
7955 * When unit files are loaded from disk, previously systemd would
7956 sometimes (depending on the unit loading order) load units from the
7957 target path of symlinks in .wants/ or .requires/ directories of other
7958 units. This meant that unit could be loaded from different paths
7959 depending on whether the unit was requested explicitly or as a
7960 dependency of another unit, not honouring the priority of directories
7961 in search path. It also meant that it was possible to successfully
7962 load and start units which are not found in the unit search path, as
7963 long as they were requested as a dependency and linked to from
7964 .wants/ or .requires/. The target paths of those symlinks are not
7965 used for loading units anymore and the unit file must be found in
7968 * A new service type has been added: Type=exec. It's very similar to
7969 Type=simple but ensures the service manager will wait for both fork()
7970 and execve() of the main service binary to complete before proceeding
7971 with follow-up units. This is primarily useful so that the manager
7972 propagates any errors in the preparation phase of service execution
7973 back to the job that requested the unit to be started. For example,
7974 consider a service that has ExecStart= set to a file system binary
7975 that doesn't exist. With Type=simple starting the unit would be
7976 considered instantly successful, as only fork() has to complete
7977 successfully and the manager does not wait for execve(), and hence
7978 its failure is seen "too late". With the new Type=exec service type
7979 starting the unit will fail, as the manager will wait for the
7980 execve() and notice its failure, which is then propagated back to the
7983 NOTE: with the next release 241 of systemd we intend to change the
7984 systemd-run tool to default to Type=exec for transient services
7985 started by it. This should be mostly safe, but in specific corner
7986 cases might result in problems, as the systemd-run tool will then
7987 block on NSS calls (such as user name look-ups due to User=) done
7988 between the fork() and execve(), which under specific circumstances
7989 might cause problems. It is recommended to specify "-p Type=simple"
7990 explicitly in the few cases where this applies. For regular,
7991 non-transient services (i.e. those defined with unit files on disk)
7992 we will continue to default to Type=simple.
7994 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
7995 userspace processes is set to 1024 (soft) and 4096
7996 (hard). Previously, systemd passed this on unmodified to all
7997 processes it forked off. With this systemd release the hard limit
7998 systemd passes on is increased to 512K, overriding the kernel's
7999 defaults and substantially increasing the number of simultaneous file
8000 descriptors unprivileged userspace processes can allocate. Note that
8001 the soft limit remains at 1024 for compatibility reasons: the
8002 traditional UNIX select() call cannot deal with file descriptors >=
8003 1024 and increasing the soft limit globally might thus result in
8004 programs unexpectedly allocating a high file descriptor and thus
8005 failing abnormally when attempting to use it with select() (of
8006 course, programs shouldn't use select() anymore, and prefer
8007 poll()/epoll, but the call unfortunately remains undeservedly popular
8008 at this time). This change reflects the fact that file descriptor
8009 handling in the Linux kernel has been optimized in more recent
8010 kernels and allocating large numbers of them should be much cheaper
8011 both in memory and in performance than it used to be. Programs that
8012 want to take benefit of the increased limit have to "opt-in" into
8013 high file descriptors explicitly by raising their soft limit. Of
8014 course, when they do that they must acknowledge that they cannot use
8015 select() anymore (and neither can any shared library they use — or
8016 any shared library used by any shared library they use and so on).
8017 Which default hard limit is most appropriate is of course hard to
8018 decide. However, given reports that ~300K file descriptors are used
8019 in real-life applications we believe 512K is sufficiently high as new
8020 default for now. Note that there are also reports that using very
8021 high hard limits (e.g. 1G) is problematic: some software allocates
8022 large arrays with one element for each potential file descriptor
8023 (Java, …) — a high hard limit thus triggers excessively large memory
8024 allocations in these applications. Hopefully, the new default of 512K
8025 is a good middle ground: higher than what real-life applications
8026 currently need, and low enough for avoid triggering excessively large
8027 allocations in problematic software. (And yes, somebody should fix
8030 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
8031 to the highest possible values, as separate accounting of file
8032 descriptors is no longer necessary, as memcg tracks them correctly as
8033 part of the memory accounting anyway. Thus, from the four limits on
8034 file descriptors currently enforced (fs.file-max, fs.nr_open,
8035 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
8036 and keep only the latter two. A set of build-time options
8037 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
8038 has been added to revert this change in behaviour, which might be
8039 an option for systems that turn off memcg in the kernel.
8041 * When no /etc/locale.conf file exists (and hence no locale settings
8042 are in place), systemd will now use the "C.UTF-8" locale by default,
8043 and set LANG= to it. This locale is supported by various
8044 distributions including Fedora, with clear indications that upstream
8045 glibc is going to make it available too. This locale enables UTF-8
8046 mode by default, which appears appropriate for 2018.
8048 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
8049 default. This effectively switches the RFC3704 Reverse Path filtering
8050 from Strict mode to Loose mode. This is more appropriate for hosts
8051 that have multiple links with routes to the same networks (e.g.
8052 a client with a Wi-Fi and Ethernet both connected to the internet).
8054 Consult the kernel documentation for details on this sysctl:
8055 https://docs.kernel.org/networking/ip-sysctl.html
8057 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
8060 * CPUAccounting=yes no longer enables the CPU controller when using
8061 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
8062 statistics are now provided independently from the CPU controller.
8064 * Support for disabling a particular cgroup controller within a sub-tree
8065 has been added through the DisableControllers= directive.
8067 * cgroup_no_v1=all on the kernel command line now also implies
8068 using the unified cgroup hierarchy, unless one explicitly passes
8069 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
8071 * The new "MemoryMin=" unit file property may now be used to set the
8072 memory usage protection limit of processes invoked by the unit. This
8073 controls the cgroup v2 memory.min attribute. Similarly, the new
8074 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
8075 cgroup v2 io.latency cgroup property for configuring per-service I/O
8078 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
8079 to the cgroup v1 "devices" cgroup controller.
8081 * systemd-escape now is able to combine --unescape with --template. It
8082 also learnt a new option --instance for extracting and unescaping the
8083 instance part of a unit name.
8085 * sd-bus now provides the sd_bus_message_readv() which is similar to
8086 sd_bus_message_read() but takes a va_list object. The pair
8087 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
8088 has been added for configuring the default method call timeout to
8089 use. sd_bus_error_move() may be used to efficiently move the contents
8090 from one sd_bus_error structure to another, invalidating the
8091 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
8092 be used to control whether a bus connection object is automatically
8093 flushed when an sd-event loop is exited.
8095 * When processing classic BSD syslog log messages, journald will now
8096 save the original time-stamp string supplied in the new
8097 SYSLOG_TIMESTAMP= journal field. This permits consumers to
8098 reconstruct the original BSD syslog message more correctly.
8100 * StandardOutput=/StandardError= in service files gained support for
8101 new "append:…" parameters, for connecting STDOUT/STDERR of a service
8102 to a file, and appending to it.
8104 * The signal to use as last step of killing of unit processes is now
8105 configurable. Previously it was hard-coded to SIGKILL, which may now
8106 be overridden with the new KillSignal= setting. Note that this is the
8107 signal used when regular termination (i.e. SIGTERM) does not suffice.
8108 Similarly, the signal used when aborting a program in case of a
8109 watchdog timeout may now be configured too (WatchdogSignal=).
8111 * The XDG_SESSION_DESKTOP environment variable may now be configured in
8112 the pam_systemd argument line, using the new desktop= switch. This is
8113 useful to initialize it properly from a display manager without
8114 having to touch C code.
8116 * Most configuration options that previously accepted percentage values
8117 now also accept permille values with the '‰' suffix (instead of '%').
8119 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
8122 * systemd-resolved's configuration file resolved.conf gained a new
8123 option ReadEtcHosts= which may be used to turn off processing and
8124 honoring /etc/hosts entries.
8126 * The "--wait" switch may now be passed to "systemctl
8127 is-system-running", in which case the tool will synchronously wait
8128 until the system finished start-up.
8130 * hostnamed gained a new bus call to determine the DMI product UUID.
8132 * On x86-64 systemd will now prefer using the RDRAND processor
8133 instruction over /dev/urandom whenever it requires randomness that
8134 neither has to be crypto-grade nor should be reproducible. This
8135 should substantially reduce the amount of entropy systemd requests
8136 from the kernel during initialization on such systems, though not
8137 reduce it to zero. (Why not zero? systemd still needs to allocate
8138 UUIDs and such uniquely, which require high-quality randomness.)
8140 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
8141 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
8142 for forcing the "Other Information" bit in IPv6 RA messages. The
8143 bonding logic gained four new options AdActorSystemPriority=,
8144 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
8145 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
8146 shuffling of flows. The tunnel logic gained a new
8147 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
8148 Deployment. The policy rule logic gained four new options IPProtocol=,
8149 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
8150 support for the MulticastToUnicast= option. networkd also gained
8151 support for configuring static IPv4 ARP or IPv6 neighbor entries.
8153 * .preset files (as read by 'systemctl preset') may now be used to
8154 instantiate services.
8156 * /etc/crypttab now understands the sector-size= option to configure
8157 the sector size for an encrypted partition.
8159 * Key material for encrypted disks may now be placed on a formatted
8160 medium, and referenced from /etc/crypttab by the UUID of the file
8161 system, followed by "=" suffixed by the path to the key file.
8163 * The "collect" udev component has been removed without replacement, as
8164 it is neither used nor maintained.
8166 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
8167 LogsDirectory=, ConfigurationDirectory= settings are used in a
8168 service the executed processes will now receive a set of environment
8169 variables containing the full paths of these directories.
8170 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
8171 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
8172 are used. Note that these options may be used multiple times per
8173 service in which case the resulting paths will be concatenated and
8174 separated by colons.
8176 * Predictable interface naming has been extended to cover InfiniBand
8177 NICs. They will be exposed with an "ib" prefix.
8179 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
8180 which case the respective line failing is ignored.
8182 * .link files may now be used to configure the equivalent to the
8183 "ethtool advertise" commands.
8185 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
8186 alternative to libudev.h. Previously, the latter was just an internal
8187 wrapper around the former, but now these two APIs are exposed
8190 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
8191 which calculates an app-specific boot ID similar to how
8192 sd_id128_get_machine_app_specific() generates an app-specific machine
8195 * A new tool systemd-id128 has been added that can be used to determine
8196 and generate various 128-bit IDs.
8198 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
8201 * systemd-hibernate-resume-generator will now honor the "noresume"
8202 kernel command line option, in which case it will bypass resuming
8203 from any hibernated image.
8205 * The systemd-sleep.conf configuration file gained new options
8206 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
8207 AllowHybridSleep= for prohibiting specific sleep modes even if the
8208 kernel exports them.
8210 * portablectl is now officially supported and has thus moved to
8213 * bootctl learnt the two new commands "set-default" and "set-oneshot"
8214 for setting the default boot loader item to boot to (either
8215 persistently or only for the next boot). This is currently only
8216 compatible with sd-boot, but may be implemented on other boot loaders
8217 too, that follow the boot loader interface. The updated interface is
8218 now documented here:
8220 https://systemd.io/BOOT_LOADER_INTERFACE
8222 * A new kernel command line option systemd.early_core_pattern= is now
8223 understood which may be used to influence the core_pattern PID 1
8224 installs during early boot.
8226 * busctl learnt two new options -j and --json= for outputting method
8227 call replies, properties and monitoring output in JSON.
8229 * journalctl's JSON output now supports simple ANSI coloring as well as
8230 a new "json-seq" mode for generating RFC7464 output.
8232 * Unit files now support the %g/%G specifiers that resolve to the UNIX
8233 group/GID of the service manager runs as, similar to the existing
8234 %u/%U specifiers that resolve to the UNIX user/UID.
8236 * systemd-logind learnt a new global configuration option
8237 UserStopDelaySec= that may be set in logind.conf. It specifies how
8238 long the systemd --user instance shall remain started after a user
8239 logs out. This is useful to speed up repetitive re-connections of the
8240 same user, as it means the user's service manager doesn't have to be
8241 stopped/restarted on each iteration, but can be reused between
8242 subsequent options. This setting defaults to 10s. systemd-logind also
8243 exports two new properties on its Manager D-Bus objects indicating
8244 whether the system's lid is currently closed, and whether the system
8247 * systemd gained support for a generic boot counting logic, which
8248 generically permits automatic reverting to older boot loader entries
8249 if newer updated ones don't work. The boot loader side is implemented
8250 in sd-boot, but is kept open for other boot loaders too. For details
8253 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
8255 * The SuccessAction=/FailureAction= unit file settings now learnt two
8256 new parameters: "exit" and "exit-force", which result in immediate
8257 exiting of the service manager, and are only useful in systemd --user
8258 and container environments.
8260 * Unit files gained support for a pair of options
8261 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
8262 exit status to use as service manager exit status when
8263 SuccessAction=/FailureAction= is set to exit or exit-force.
8265 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
8266 options may now be used to configure the log rate limiting applied by
8267 journald per-service.
8269 * systemd-analyze gained a new verb "timespan" for parsing and
8270 normalizing time span values (i.e. strings like "5min 7s 8us").
8272 * systemd-analyze also gained a new verb "security" for analyzing the
8273 security and sand-boxing settings of services in order to determine an
8274 "exposure level" for them, indicating whether a service would benefit
8275 from more sand-boxing options turned on for them.
8277 * "systemd-analyze syscall-filter" will now also show system calls
8278 supported by the local kernel but not included in any of the defined
8281 * .nspawn files now understand the Ephemeral= setting, matching the
8282 --ephemeral command line switch.
8284 * sd-event gained the new APIs sd_event_source_get_floating() and
8285 sd_event_source_set_floating() for controlling whether a specific
8286 event source is "floating", i.e. destroyed along with the even loop
8289 * Unit objects on D-Bus gained a new "Refs" property that lists all
8290 clients that currently have a reference on the unit (to ensure it is
8293 * The JoinControllers= option in system.conf is no longer supported, as
8294 it didn't work correctly, is hard to support properly, is legacy (as
8295 the concept only exists on cgroup v1) and apparently wasn't used.
8297 * Journal messages that are generated whenever a unit enters the failed
8298 state are now tagged with a unique MESSAGE_ID. Similarly, messages
8299 generated whenever a service process exits are now made recognizable,
8300 too. A tagged message is also emitted whenever a unit enters the
8301 "dead" state on success.
8303 * systemd-run gained a new switch --working-directory= for configuring
8304 the working directory of the service to start. A shortcut -d is
8305 equivalent, setting the working directory of the service to the
8306 current working directory of the invoking program. The new --shell
8307 (or just -S) option has been added for invoking the $SHELL of the
8308 caller as a service, and implies --pty --same-dir --wait --collect
8309 --service-type=exec. Or in other words, "systemd-run -S" is now the
8310 quickest way to quickly get an interactive in a fully clean and
8311 well-defined system service context.
8313 * machinectl gained a new verb "import-fs" for importing an OS tree
8314 from a directory. Moreover, when a directory or tarball is imported
8315 and single top-level directory found with the OS itself below the OS
8316 tree is automatically mangled and moved one level up.
8318 * systemd-importd will no longer set up an implicit btrfs loop-back
8319 file system on /var/lib/machines. If one is already set up, it will
8320 continue to be used.
8322 * A new generator "systemd-run-generator" has been added. It will
8323 synthesize a unit from one or more program command lines included in
8324 the kernel command line. This is very useful in container managers
8327 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
8329 This will run "systemd-nspawn" on an image, invoke the specified
8330 command line and immediately shut down the container again, returning
8331 the command line's exit code.
8333 * The block device locking logic is now documented:
8335 https://systemd.io/BLOCK_DEVICE_LOCKING
8337 * loginctl and machinectl now optionally output the various tables in
8338 JSON using the --output= switch. It is our intention to add similar
8339 support to systemctl and all other commands.
8341 * udevadm's query and trigger verb now optionally take a .device unit
8344 * systemd-udevd's network naming logic now understands a new
8345 net.naming_scheme= kernel command line switch, which may be used to
8346 pick a specific version of the naming scheme. This helps stabilizing
8347 interface names even as systemd/udev are updated and the naming logic
8350 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
8351 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
8352 initialize one to all 0xFF.
8354 * After loading the SELinux policy systemd will now recursively relabel
8355 all files and directories listed in
8356 /run/systemd/relabel-extra.d/*.relabel (which should be simple
8357 newline separated lists of paths) in addition to the ones it already
8358 implicitly relabels in /run, /dev and /sys. After the relabelling is
8359 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
8360 removed. This is useful to permit initrds (i.e. code running before
8361 the SELinux policy is in effect) to generate files in the host
8362 filesystem safely and ensure that the correct label is applied during
8363 the transition to the host OS.
8365 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
8366 mknod() handling in user namespaces. Previously mknod() would always
8367 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
8368 but device nodes generated that way cannot be opened, and attempts to
8369 open them result in EPERM. This breaks the "graceful fallback" logic
8370 in systemd's PrivateDevices= sand-boxing option. This option is
8371 implemented defensively, so that when systemd detects it runs in a
8372 restricted environment (such as a user namespace, or an environment
8373 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
8374 where device nodes cannot be created the effect of PrivateDevices= is
8375 bypassed (following the logic that 2nd-level sand-boxing is not
8376 essential if the system systemd runs in is itself already sand-boxed
8377 as a whole). This logic breaks with 4.18 in container managers where
8378 user namespacing is used: suddenly PrivateDevices= succeeds setting
8379 up a private /dev/ file system containing devices nodes — but when
8380 these are opened they don't work.
8382 At this point it is recommended that container managers utilizing
8383 user namespaces that intend to run systemd in the payload explicitly
8384 block mknod() with seccomp or similar, so that the graceful fallback
8387 We are very sorry for the breakage and the requirement to change
8388 container configurations for newer kernels. It's purely caused by an
8389 incompatible kernel change. The relevant kernel developers have been
8390 notified about this userspace breakage quickly, but they chose to
8393 * PermissionsStartOnly= setting is deprecated (but is still supported
8394 for backwards compatibility). The same functionality is provided by
8395 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
8398 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
8399 pam_systemd anymore.
8401 * The naming scheme for network devices was changed to always rename
8402 devices, even if they were already renamed by userspace. The "kernel"
8403 policy was changed to only apply as a fallback, if no other naming
8406 * The requirements to build systemd is bumped to meson-0.46 and
8409 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
8410 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
8411 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
8412 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
8413 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
8414 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
8415 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
8416 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
8417 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
8418 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
8419 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
8420 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
8421 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
8422 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
8423 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
8424 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
8425 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
8426 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
8427 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
8428 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
8429 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
8430 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
8431 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
8432 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
8433 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
8434 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
8435 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
8436 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
8437 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
8438 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
8439 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
8440 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
8441 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
8442 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
8443 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
8444 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
8445 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
8446 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
8447 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
8448 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
8449 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
8450 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
8451 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
8452 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
8453 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
8455 — Warsaw, 2018-12-21
8459 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
8460 builtin will name network interfaces differently than in previous
8461 versions for virtual network interfaces created with SR-IOV and NPAR
8462 and for devices where the PCI network controller device does not have
8463 a slot number associated.
8465 SR-IOV virtual devices are now named based on the name of the parent
8466 interface, with a suffix of "v<N>", where <N> is the virtual device
8467 number. Previously those virtual devices were named as if completely
8470 The ninth and later NPAR virtual devices will be named following the
8471 scheme used for the first eight NPAR partitions. Previously those
8472 devices were not renamed and the kernel default (eth<n>) was used.
8474 "net_id" will also generate names for PCI devices where the PCI
8475 network controller device does not have an associated slot number
8476 itself, but one of its parents does. Previously those devices were
8477 not renamed and the kernel default (eth<n>) was used.
8479 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
8480 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
8481 the unit. So, it is expected that the default behavior of
8482 systemd-logind is not changed. However, if distribution packagers or
8483 administrators disabled or modified IPAddressDeny= setting by a
8484 drop-in config file, then it may be necessary to update the file to
8485 re-enable AF_INET and AF_INET6 to support network user name services,
8488 * When the RestrictNamespaces= unit property is specified multiple
8489 times, then the specified types are merged now. Previously, only the
8490 last assignment was used. So, if distribution packagers or
8491 administrators modified the setting by a drop-in config file, then it
8492 may be necessary to update the file.
8494 * When OnFailure= is used in combination with Restart= on a service
8495 unit, then the specified units will no longer be triggered on
8496 failures that result in restarting. Previously, the specified units
8497 would be activated each time the unit failed, even when the unit was
8498 going to be restarted automatically. This behaviour contradicted the
8499 documentation. With this release the code is adjusted to match the
8502 * systemd-tmpfiles will now print a notice whenever it encounters
8503 tmpfiles.d/ lines referencing the /var/run/ directory. It will
8504 recommend reworking them to use the /run/ directory instead (for
8505 which /var/run/ is simply a symlinked compatibility alias). This way
8506 systemd-tmpfiles can properly detect line conflicts and merge lines
8507 referencing the same file by two paths, without having to access
8510 * systemctl disable/unmask/preset/preset-all cannot be used with
8511 --runtime. Previously this was allowed, but resulted in unintuitive
8512 behaviour that wasn't useful. systemctl disable/unmask will now undo
8513 both runtime and persistent enablement/masking, i.e. it will remove
8514 any relevant symlinks both in /run and /etc.
8516 * Note that all long-running system services shipped with systemd will
8517 now default to a system call allow list (rather than a deny list, as
8518 before). In particular, systemd-udevd will now enforce one too. For
8519 most cases this should be safe, however downstream distributions
8520 which disabled sandboxing of systemd-udevd (specifically the
8521 MountFlags= setting), might want to disable this security feature
8522 too, as the default allow-listing will prohibit all mount, swap,
8523 reboot and clock changing operations from udev rules.
8525 * sd-boot acquired new loader configuration settings to optionally turn
8526 off Windows and MacOS boot partition discovery as well as
8527 reboot-into-firmware menu items. It is also able to pick a better
8528 screen resolution for HiDPI systems, and now provides loader
8529 configuration settings to change the resolution explicitly.
8531 * systemd-resolved now supports DNS-over-TLS. It's still
8532 turned off by default, use DNSOverTLS=opportunistic to turn it on in
8533 resolved.conf. We intend to make this the default as soon as couple
8534 of additional techniques for optimizing the initial latency caused by
8535 establishing a TLS/TCP connection are implemented.
8537 * systemd-resolved.service and systemd-networkd.service now set
8538 DynamicUser=yes. The users systemd-resolve and systemd-network are
8539 not created by systemd-sysusers anymore.
8541 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
8542 that embed a network facing module into any process using getpwuid()
8543 or related call: the dynamic allocation of the user ID for
8544 systemd-resolved.service means the service manager has to check NSS
8545 if the user name is already taken when forking off the service. Since
8546 the user in the common case won't be defined in /etc/passwd the
8547 lookup is likely to trigger nss-ldap which in turn might use NSS to
8548 ask systemd-resolved for hostname lookups. This will hence result in
8549 a deadlock: a user name lookup in order to start
8550 systemd-resolved.service will result in a hostname lookup for which
8551 systemd-resolved.service needs to be started already. There are
8552 multiple ways to work around this problem: pre-allocate the
8553 "systemd-resolve" user on such systems, so that nss-ldap won't be
8554 triggered; or use a different NSS package that doesn't do networking
8555 in-process but provides a local asynchronous name cache; or configure
8556 the NSS package to avoid lookups for UIDs in the range `pkg-config
8557 systemd --variable=dynamicuidmin` … `pkg-config systemd
8558 --variable=dynamicuidmax`, so that it does not consider itself
8559 authoritative for the same UID range systemd allocates dynamic users
8562 * The systemd-resolve tool has been renamed to resolvectl (it also
8563 remains available under the old name, for compatibility), and its
8564 interface is now verb-based, similar in style to the other <xyz>ctl
8565 tools, such as systemctl or loginctl.
8567 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
8568 compatibility. It may be symlinked under the 'resolvconf' name, in
8569 which case it will take arguments and input compatible with the
8570 Debian and FreeBSD resolvconf tool.
8572 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
8573 where the system initially suspends, and after a timeout resumes and
8576 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
8577 set the client will only send a DUID as client identifier. (EDIT: the
8578 option was broken, and was dropped in v255.)
8580 * The nss-systemd glibc NSS module will now enumerate dynamic users and
8581 groups in effect. Previously, it could resolve UIDs/GIDs to user
8582 names/groups and vice versa, but did not support enumeration.
8584 * journald's Compress= configuration setting now optionally accepts a
8585 byte threshold value. All journal objects larger than this threshold
8586 will be compressed, smaller ones will not. Previously this threshold
8587 was not configurable and set to 512.
8589 * A new system.conf setting NoNewPrivileges= is now available which may
8590 be used to turn off acquisition of new privileges system-wide
8591 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
8592 for all its children). Note that turning this option on means setuid
8593 binaries and file system capabilities lose their special powers.
8594 While turning on this option is a big step towards a more secure
8595 system, doing so is likely to break numerous pre-existing UNIX tools,
8596 in particular su and sudo.
8598 * A new service systemd-time-sync-wait.service has been added. If
8599 enabled it will delay the time-sync.target unit at boot until time
8600 synchronization has been received from the network. This
8601 functionality is useful on systems lacking a local RTC or where it is
8602 acceptable that the boot process shall be delayed by external network
8605 * When hibernating, systemd will now inform the kernel of the image
8606 write offset, on kernels new enough to support this. This means swap
8607 files should work for hibernation now.
8609 * When loading unit files, systemd will now look for drop-in unit files
8610 extensions in additional places. Previously, for a unit file name
8611 "foo-bar-baz.service" it would look for dropin files in
8612 "foo-bar-baz.service.d/*.conf". Now, it will also look in
8613 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
8614 service name truncated after all inner dashes. This scheme allows
8615 writing drop-ins easily that apply to a whole set of unit files at
8616 once. It's particularly useful for mount and slice units (as their
8617 naming is prefix based), but is also useful for service and other
8618 units, for packages that install multiple unit files at once,
8619 following a strict naming regime of beginning the unit file name with
8620 the package's name. Two new specifiers are now supported in unit
8621 files to match this: %j and %J are replaced by the part of the unit
8622 name following the last dash.
8624 * Unit files and other configuration files that support specifier
8625 expansion now understand another three new specifiers: %T and %V will
8626 resolve to /tmp and /var/tmp respectively, or whatever temporary
8627 directory has been set for the calling user. %E will expand to either
8628 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
8630 * The ExecStart= lines of unit files are no longer required to
8631 reference absolute paths. If non-absolute paths are specified the
8632 specified binary name is searched within the service manager's
8633 built-in $PATH, which may be queried with 'systemd-path
8634 search-binaries-default'. It's generally recommended to continue to
8635 use absolute paths for all binaries specified in unit files.
8637 * Units gained a new load state "bad-setting", which is used when a
8638 unit file was loaded, but contained fatal errors which prevent it
8639 from being started (for example, a service unit has been defined
8640 lacking both ExecStart= and ExecStop= lines).
8642 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
8643 support alternative debuggers, for example lldb. The old name
8644 continues to be available however, for compatibility reasons. Use the
8645 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
8646 to pick an alternative debugger instead of the default gdb.
8648 * systemctl and the other tools will now output escape sequences that
8649 generate proper clickable hyperlinks in various terminal emulators
8650 where useful (for example, in the "systemctl status" output you can
8651 now click on the unit file name to quickly open it in the
8652 editor/viewer of your choice). Note that not all terminal emulators
8653 support this functionality yet, but many do. Unfortunately, the
8654 "less" pager doesn't support this yet, hence this functionality is
8655 currently automatically turned off when a pager is started (which
8656 happens quite often due to auto-paging). We hope to remove this
8657 limitation as soon as "less" learns these escape sequences. This new
8658 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
8659 environment variable. For details on these escape sequences see:
8660 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
8662 * networkd's .network files now support a new IPv6MTUBytes= option for
8663 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
8664 option in the [Route] section to configure the MTU to use for
8665 specific routes. It also gained support for configuration of the DHCP
8666 "UserClass" option through the new UserClass= setting. It gained
8667 three new options in the new [CAN] section for configuring CAN
8668 networks. The MULTICAST and ALLMULTI interface flags may now be
8669 controlled explicitly with the new Multicast= and AllMulticast=
8672 * networkd will now automatically make use of the kernel's route
8673 expiration feature, if it is available.
8675 * udevd's .link files now support setting the number of receive and
8676 transmit channels, using the RxChannels=, TxChannels=,
8677 OtherChannels=, CombinedChannels= settings.
8679 * Support for UDPSegmentationOffload= has been removed, given its
8680 limited support in hardware, and waning software support.
8682 * networkd's .netdev files now support creating "netdevsim" interfaces.
8684 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
8685 to query the unit belonging to a specific kernel control group.
8687 * systemd-analyze gained a new verb "cat-config", which may be used to
8688 dump the contents of any configuration file, with all its matching
8689 drop-in files added in, and honouring the usual search and masking
8690 logic applied to systemd configuration files. For example use
8691 "systemd-analyze cat-config systemd/system.conf" to get the complete
8692 system configuration file of systemd how it would be loaded by PID 1
8693 itself. Similar to this, various tools such as systemd-tmpfiles or
8694 systemd-sysusers, gained a new option "--cat-config", which does the
8695 corresponding operation for their own configuration settings. For
8696 example, "systemd-tmpfiles --cat-config" will now output the full
8697 list of tmpfiles.d/ lines in place.
8699 * timedatectl gained three new verbs: "show" shows bus properties of
8700 systemd-timedated, "timesync-status" shows the current NTP
8701 synchronization state of systemd-timesyncd, and "show-timesync"
8702 shows bus properties of systemd-timesyncd.
8704 * systemd-timesyncd gained a bus interface on which it exposes details
8707 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
8708 understood by systemd-timedated. It takes a colon-separated list of
8709 unit names of NTP client services. The list is used by
8710 "timedatectl set-ntp".
8712 * systemd-nspawn gained a new --rlimit= switch for setting initial
8713 resource limits for the container payload. There's a new switch
8714 --hostname= to explicitly override the container's hostname. A new
8715 --no-new-privileges= switch may be used to control the
8716 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
8717 --oom-score-adjust= switch controls the OOM scoring adjustment value
8718 for the payload. The new --cpu-affinity= switch controls the CPU
8719 affinity of the container payload. The new --resolv-conf= switch
8720 allows more detailed control of /etc/resolv.conf handling of the
8721 container. Similarly, the new --timezone= switch allows more detailed
8722 control of /etc/localtime handling of the container.
8724 * systemd-detect-virt gained a new --list switch, which will print a
8725 list of all currently known VM and container environments.
8727 * Support for "Portable Services" has been added, see
8728 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
8729 experimental, but this is expected to change soon. Reflecting this
8730 experimental state, the "portablectl" binary is not installed into
8731 /usr/bin yet. The binary has to be called with the full path
8732 /usr/lib/systemd/portablectl instead.
8734 * journalctl's and systemctl's -o switch now knows a new log output
8735 mode "with-unit". The output it generates is very similar to the
8736 regular "short" mode, but displays the unit name instead of the
8737 syslog tag for each log line. Also, the date is shown with timezone
8738 information. This mode is probably more useful than the classic
8739 "short" output mode for most purposes, except where pixel-perfect
8740 compatibility with classic /var/log/messages formatting is required.
8742 * A new --dump-bus-properties switch has been added to the systemd
8743 binary, which may be used to dump all supported D-Bus properties.
8744 (Options which are still supported, but are deprecated, are *not*
8747 * sd-bus gained a set of new calls:
8748 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
8749 enable/disable the "floating" state of a bus slot object,
8750 i.e. whether the slot object pins the bus it is allocated for into
8751 memory or if the bus slot object gets disconnected when the bus goes
8752 away. sd_bus_open_with_description(),
8753 sd_bus_open_user_with_description(),
8754 sd_bus_open_system_with_description() may be used to allocate bus
8755 objects and set their description string already during allocation.
8757 * sd-event gained support for watching inotify events from the event
8758 loop, in an efficient way, sharing inotify handles between multiple
8759 users. For this a new function sd_event_add_inotify() has been added.
8761 * sd-event and sd-bus gained support for calling special user-supplied
8762 destructor functions for userdata pointers associated with
8763 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
8764 functions sd_bus_slot_set_destroy_callback,
8765 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
8766 sd_bus_track_get_destroy_callback,
8767 sd_event_source_set_destroy_callback,
8768 sd_event_source_get_destroy_callback have been added.
8770 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
8772 * PID 1 will now automatically reschedule .timer units whenever the
8773 local timezone changes. (They previously got rescheduled
8774 automatically when the system clock changed.)
8776 * New documentation has been added to document cgroups delegation,
8777 portable services and the various code quality tools we have set up:
8779 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
8780 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
8781 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
8783 * The Boot Loader Specification has been added to the source tree.
8785 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
8787 While moving it into our source tree we have updated it and further
8788 changes are now accepted through the usual github PR workflow.
8790 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
8791 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
8792 earlier PAM modules. The data in these fields is used to initialize
8793 the session scope's resource properties. Thus external PAM modules
8794 may now configure per-session limits, for example sourced from
8795 external user databases.
8797 * socket units with Accept=yes will now maintain a "refused" counter in
8798 addition to the existing "accepted" counter, counting connections
8799 refused due to the enforced limits.
8801 * The "systemd-path search-binaries-default" command may now be use to
8802 query the default, built-in $PATH PID 1 will pass to the services it
8805 * A new unit file setting PrivateMounts= has been added. It's a boolean
8806 option. If enabled the unit's processes are invoked in their own file
8807 system namespace. Note that this behaviour is also implied if any
8808 other file system namespacing options (such as PrivateTmp=,
8809 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
8810 primarily useful for services that do not use any of the other file
8811 system namespacing options. One such service is systemd-udevd.service
8812 where this is now used by default.
8814 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
8815 when the system is booted in UEFI "secure mode".
8817 * A new unit "system-update-pre.target" is added, which defines an
8818 optional synchronization point for offline system updates, as
8819 implemented by the pre-existing "system-update.target" unit. It
8820 allows ordering services before the service that executes the actual
8821 update process in a generic way.
8823 * Systemd now emits warnings whenever .include syntax is used.
8825 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
8826 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
8827 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
8828 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
8829 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
8830 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
8831 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
8832 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
8833 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
8834 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
8835 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
8836 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
8837 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
8838 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
8839 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
8840 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
8841 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
8842 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
8843 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
8844 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
8845 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
8846 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
8847 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
8848 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
8849 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
8850 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
8851 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
8852 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
8853 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8855 — Berlin, 2018-06-22
8859 * The MemoryAccounting= unit property now defaults to on. After
8860 discussions with the upstream control group maintainers we learnt
8861 that the negative impact of cgroup memory accounting on current
8862 kernels is finally relatively minimal, so that it should be safe to
8863 enable this by default without affecting system performance. Besides
8864 memory accounting only task accounting is turned on by default, all
8865 other forms of resource accounting (CPU, IO, IP) remain off for now,
8866 because it's not clear yet that their impact is small enough to move
8867 from opt-in to opt-out. We recommend downstreams to leave memory
8868 accounting on by default if kernel 4.14 or higher is primarily
8869 used. On very resource constrained systems or when support for old
8870 kernels is a necessity, -Dmemory-accounting-default=false can be used
8871 to revert this change.
8873 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
8874 %udev_rules_update) and the journal catalog (%journal_catalog_update)
8875 from the upgrade scriptlets of individual packages now do nothing.
8876 Transfiletriggers have been added which will perform those updates
8877 once at the end of the transaction.
8879 Similar transfiletriggers have been added to execute any sysctl.d
8880 and binfmt.d rules. Thus, it should be unnecessary to provide any
8881 scriptlets to execute this configuration from package installation
8884 * systemd-sysusers gained a mode where the configuration to execute is
8885 specified on the command line, but this configuration is not executed
8886 directly, but instead it is merged with the configuration on disk,
8887 and the result is executed. This is useful for package installation
8888 scripts which want to create the user before installing any files on
8889 disk (in case some of those files are owned by that user), while
8890 still allowing local admin overrides.
8892 This functionality is exposed to rpm scriptlets through a new
8893 %sysusers_create_package macro. Old %sysusers_create and
8894 %sysusers_create_inline macros are deprecated.
8896 A transfiletrigger for sysusers.d configuration is now installed,
8897 which means that it should be unnecessary to call systemd-sysusers from
8898 package installation scripts, unless the package installs any files
8899 owned by those newly-created users, in which case
8900 %sysusers_create_package should be used.
8902 * Analogous change has been done for systemd-tmpfiles: it gained a mode
8903 where the command-line configuration is merged with the configuration
8904 on disk. This is exposed as the new %tmpfiles_create_package macro,
8905 and %tmpfiles_create is deprecated. A transfiletrigger is installed
8906 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
8907 from package installation scripts.
8909 * sysusers.d configuration for a user may now also specify the group
8910 number, in addition to the user number ("u username 123:456"), or
8911 without the user number ("u username -:456").
8913 * Configution items for systemd-sysusers can now be specified as
8914 positional arguments when the new --inline switch is used.
8916 * The login shell of users created through sysusers.d may now be
8917 specified (previously, it was always /bin/sh for root and
8918 /sbin/nologin for other users).
8920 * systemd-analyze gained a new --global switch to look at global user
8921 configuration. It also gained a unit-paths verb to list the unit load
8922 paths that are compiled into systemd (which can be used with
8923 --systemd, --user, or --global).
8925 * udevadm trigger gained a new --settle/-w option to wait for any
8926 triggered events to finish (but just those, and not any other events
8927 which are triggered meanwhile).
8929 * The action that systemd-logind takes when the lid is closed and the
8930 machine is connected to external power can now be configured using
8931 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
8932 was determined by HandleLidSwitch=, and, for backwards compatibility,
8933 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
8935 * journalctl will periodically call sd_journal_process() to make it
8936 resilient against inotify queue overruns when journal files are
8937 rotated very quickly.
8939 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
8940 sd_bus_get_n_queued_write — may be used to check the number of
8941 pending bus messages.
8943 * systemd gained a new
8944 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
8945 which can be used to migrate foreign processes to scope and service
8946 units. The primary user for this new API is systemd itself: the
8947 systemd --user instance uses this call of the systemd --system
8948 instance to migrate processes if it itself gets the request to
8949 migrate processes and the kernel refuses this due to access
8950 restrictions. Thanks to this "systemd-run --scope --user …" works
8951 again in pure cgroup v2 environments when invoked from the user
8954 * A new TemporaryFileSystem= setting can be used to mask out part of
8955 the real file system tree with tmpfs mounts. This may be combined
8956 with BindPaths= and BindReadOnlyPaths= to hide files or directories
8957 not relevant to the unit, while still allowing some paths lower in
8958 the tree to be accessed.
8960 ProtectHome=tmpfs may now be used to hide user home and runtime
8961 directories from units, in a way that is mostly equivalent to
8962 "TemporaryFileSystem=/home /run/user /root".
8964 * Non-service units are now started with KeyringMode=shared by default.
8965 This means that mount and swapon and other mount tools have access
8966 to keys in the main keyring.
8968 * /sys/fs/bpf is now mounted automatically.
8970 * QNX virtualization is now detected by systemd-detect-virt and may
8971 be used in ConditionVirtualization=.
8973 * IPAccounting= may now be enabled also for slice units.
8975 * A new -Dsplit-bin= build configuration switch may be used to specify
8976 whether bin and sbin directories are merged, or if they should be
8977 included separately in $PATH and various listings of executable
8978 directories. The build configuration scripts will try to autodetect
8979 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
8980 system, but distributions are encouraged to configure this
8983 * A new -Dok-color= build configuration switch may be used to change
8984 the colour of "OK" status messages.
8986 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
8987 PrivateNetwork=yes was buggy in previous versions of systemd. This
8988 means that after the upgrade and daemon-reexec, any such units must
8991 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
8992 will not exclude read-only files owned by root from cleanup.
8994 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
8995 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
8996 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
8997 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
8998 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
8999 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
9000 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
9001 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
9002 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
9003 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
9004 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
9005 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
9006 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
9007 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
9008 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
9009 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
9011 — Warsaw, 2018-03-05
9015 * Some keyboards come with a zoom see-saw or rocker which until now got
9016 mapped to the Linux "zoomin/out" keys in hwdb. However, these
9017 keycodes are not recognized by any major desktop. They now produce
9018 Up/Down key events so that they can be used for scrolling.
9020 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
9021 slightly: previously, if an argument was specified for lines of this
9022 type (i.e. the right-most column was set) this string was appended to
9023 existing files each time systemd-tmpfiles was run. This behaviour was
9024 different from what the documentation said, and not particularly
9025 useful, as repeated systemd-tmpfiles invocations would not be
9026 idempotent and grow such files without bounds. With this release
9027 behaviour has been altered to match what the documentation says:
9028 lines of this type only have an effect if the indicated files don't
9029 exist yet, and only then the argument string is written to the file.
9031 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
9032 systemd-tmpfiles behaviour: previously, read-only files owned by root
9033 were always excluded from the file "aging" algorithm (i.e. the
9034 automatic clean-up of directories like /tmp based on
9035 atime/mtime/ctime). We intend to drop this restriction, and age files
9036 by default even when owned by root and read-only. This behaviour was
9037 inherited from older tools, but there have been requests to remove
9038 it, and it's not obvious why this restriction was made in the first
9039 place. Please speak up now, if you are aware of software that requires
9040 this behaviour, otherwise we'll remove the restriction in v238.
9042 * A new environment variable $SYSTEMD_OFFLINE is now understood by
9043 systemctl. It takes a boolean argument. If on, systemctl assumes it
9044 operates on an "offline" OS tree, and will not attempt to talk to the
9045 service manager. Previously, this mode was implicitly enabled if a
9046 chroot() environment was detected, and this new environment variable
9047 now provides explicit control.
9049 * .path and .socket units may now be created transiently, too.
9050 Previously only service, mount, automount and timer units were
9051 supported as transient units. The systemd-run tool has been updated
9052 to expose this new functionality, you may hence use it now to bind
9053 arbitrary commands to path or socket activation on-the-fly from the
9054 command line. Moreover, almost all properties are now exposed for the
9055 unit types that already supported transient operation.
9057 * The systemd-mount command gained support for a new --owner= parameter
9058 which takes a user name, which is then resolved and included in uid=
9059 and gid= mount options string of the file system to mount.
9061 * A new unit condition ConditionControlGroupController= has been added
9062 that checks whether a specific cgroup controller is available.
9064 * Unit files, udev's .link files, and systemd-networkd's .netdev and
9065 .network files all gained support for a new condition
9066 ConditionKernelVersion= for checking against specific kernel
9069 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
9070 support for configuring device flags in the Flags= setting. In the
9071 same files, the [Tunnel] section gained support for configuring
9072 AllowLocalRemote=. The [Route] section in .network files gained
9073 support for configuring InitialCongestionWindow=,
9074 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
9075 understands RapidCommit=.
9077 * systemd-networkd's DHCPv6 support gained support for Prefix
9080 * sd-bus gained support for a new "watch-bind" feature. When this
9081 feature is enabled, an sd_bus connection may be set up to connect to
9082 an AF_UNIX socket in the file system as soon as it is created. This
9083 functionality is useful for writing early-boot services that
9084 automatically connect to the system bus as soon as it is started,
9085 without ugly time-based polling. systemd-networkd and
9086 systemd-resolved have been updated to make use of this
9087 functionality. busctl exposes this functionality in a new
9088 --watch-bind= command line switch.
9090 * sd-bus will now optionally synthesize a local "Connected" signal as
9091 soon as a D-Bus connection is set up fully. This message mirrors the
9092 already existing "Disconnected" signal which is synthesized when the
9093 connection is terminated. This signal is generally useful but
9094 particularly handy in combination with the "watch-bind" feature
9095 described above. Synthesizing of this message has to be requested
9096 explicitly through the new API call sd_bus_set_connected_signal(). In
9097 addition a new call sd_bus_is_ready() has been added that checks
9098 whether a connection is fully set up (i.e. between the "Connected" and
9099 "Disconnected" signals).
9101 * sd-bus gained two new calls sd_bus_request_name_async() and
9102 sd_bus_release_name_async() for asynchronously registering bus
9103 names. Similar, there is now sd_bus_add_match_async() for installing
9104 a signal match asynchronously. All of systemd's own services have
9105 been updated to make use of these calls. Doing these operations
9106 asynchronously has two benefits: it reduces the risk of deadlocks in
9107 case of cyclic dependencies between bus services, and it speeds up
9108 service initialization since synchronization points for bus
9109 round-trips are removed.
9111 * sd-bus gained two new calls sd_bus_match_signal() and
9112 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
9113 and sd_bus_add_match_async() but instead of taking a D-Bus match
9114 string take match fields as normal function parameters.
9116 * sd-bus gained two new calls sd_bus_set_sender() and
9117 sd_bus_message_set_sender() for setting the sender name of outgoing
9118 messages (either for all outgoing messages or for just one specific
9119 one). These calls are only useful in direct connections as on
9120 brokered connections the broker fills in the sender anyway,
9121 overwriting whatever the client filled in.
9123 * sd-event gained a new pseudo-handle that may be specified on all API
9124 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
9125 used this refers to the default event loop object of the calling
9126 thread. Note however that this does not implicitly allocate one —
9127 which has to be done prior by using sd_event_default(). Similarly
9128 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
9129 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
9130 to the default bus of the specified type of the calling thread. Here
9131 too this does not implicitly allocate bus connection objects, this
9132 has to be done prior with sd_bus_default() and friends.
9134 * sd-event gained a new call pair
9135 sd_event_source_{get|set}_io_fd_own(). This may be used to request
9136 automatic closure of the file descriptor an IO event source watches
9137 when the event source is destroyed.
9139 * systemd-networkd gained support for natively configuring WireGuard
9142 * In previous versions systemd synthesized user records both for the
9143 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
9144 internally. In order to simplify distribution-wide renames of the
9145 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
9146 new transitional flag file has been added: if
9147 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
9148 user and group record within the systemd codebase is disabled.
9150 * systemd-notify gained a new --uid= option for selecting the source
9151 user/UID to use for notification messages sent to the service
9154 * journalctl gained a new --grep= option to list only entries in which
9155 the message matches a certain pattern. By default matching is case
9156 insensitive if the pattern is lowercase, and case sensitive
9157 otherwise. Option --case-sensitive=yes|no can be used to override
9158 this an specify case sensitivity or case insensitivity.
9160 * There's now a "systemd-analyze service-watchdogs" command for printing
9161 the current state of the service runtime watchdog, and optionally
9162 enabling or disabling the per-service watchdogs system-wide if given a
9163 boolean argument (i.e. the concept you configure in WatchdogSec=), for
9164 debugging purposes. There's also a kernel command line option
9165 systemd.service_watchdogs= for controlling the same.
9167 * Two new "log-level" and "log-target" options for systemd-analyze were
9168 added that merge the now deprecated get-log-level, set-log-level and
9169 get-log-target, set-log-target pairs. The deprecated options are still
9170 understood for backwards compatibility. The two new options print the
9171 current value when no arguments are given, and set them when a
9172 level/target is given as an argument.
9174 * sysusers.d's "u" lines now optionally accept both a UID and a GID
9175 specification, separated by a ":" character, in order to create users
9176 where UID and GID do not match.
9178 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
9179 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
9180 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
9181 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
9182 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
9183 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
9184 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
9185 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
9186 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
9187 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
9188 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
9189 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
9190 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
9191 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
9192 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
9193 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
9194 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
9195 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
9196 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
9197 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
9204 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
9205 in v235 has been extended to also set the dummy.ko module option
9206 numdummies=0, preventing the kernel from automatically creating
9207 dummy0. All dummy interfaces must now be explicitly created.
9209 * Unknown '%' specifiers in configuration files are now rejected. This
9210 applies to units and tmpfiles.d configuration. Any percent characters
9211 that are followed by a letter or digit that are not supposed to be
9212 interpreted as the beginning of a specifier should be escaped by
9213 doubling ("%%"). (So "size=5%" is still accepted, as well as
9214 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
9215 valid specifiers today.)
9217 * systemd-resolved now maintains a new dynamic
9218 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
9219 recommended to make /etc/resolv.conf a symlink to it. This file
9220 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
9221 includes dynamically acquired search domains, achieving more correct
9222 DNS resolution by software that bypasses local DNS APIs such as NSS.
9224 * The "uaccess" udev tag has been dropped from /dev/kvm and
9225 /dev/dri/renderD*. These devices now have the 0666 permissions by
9226 default (but this may be changed at build-time). /dev/dri/renderD*
9227 will now be owned by the "render" group along with /dev/kfd.
9229 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
9230 systemd-journal-gatewayd.service and
9231 systemd-journal-upload.service. This means "nss-systemd" must be
9232 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
9233 services are resolved properly.
9235 * In /etc/fstab two new mount options are now understood:
9236 x-systemd.makefs and x-systemd.growfs. The former has the effect that
9237 the configured file system is formatted before it is mounted, the
9238 latter that the file system is resized to the full block device size
9239 after it is mounted (i.e. if the file system is smaller than the
9240 partition it resides on, it's grown). This is similar to the fsck
9241 logic in /etc/fstab, and pulls in systemd-makefs@.service and
9242 systemd-growfs@.service as necessary, similar to
9243 systemd-fsck@.service. Resizing is currently only supported on ext4
9246 * In systemd-networkd, the IPv6 RA logic now optionally may announce
9247 DNS server and domain information.
9249 * Support for the LUKS2 on-disk format for encrypted partitions has
9250 been added. This requires libcryptsetup2 during compilation and
9253 * The systemd --user instance will now signal "readiness" when its
9254 basic.target unit has been reached, instead of when the run queue ran
9255 empty for the first time.
9257 * Tmpfiles.d with user configuration are now also supported.
9258 systemd-tmpfiles gained a new --user switch, and snippets placed in
9259 ~/.config/user-tmpfiles.d/ and corresponding directories will be
9260 executed by systemd-tmpfiles --user running in the new
9261 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
9262 running in the user session.
9264 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
9265 %S resolves to the top-level state directory (/var/lib for the system
9266 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
9267 top-level cache directory (/var/cache for the system instance,
9268 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
9269 logs directory (/var/log for the system instance,
9270 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
9271 existing %t specifier, that resolves to the top-level runtime
9272 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
9275 * journalctl learnt a new parameter --output-fields= for limiting the
9276 set of journal fields to output in verbose and JSON output modes.
9278 * systemd-timesyncd's configuration file gained a new option
9279 RootDistanceMaxSec= for setting the maximum root distance of servers
9280 it'll use, as well as the new options PollIntervalMinSec= and
9281 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
9283 * bootctl gained a new command "list" for listing all available boot
9284 menu items on systems that follow the boot loader specification.
9286 * systemctl gained a new --dry-run switch that shows what would be done
9287 instead of doing it, and is currently supported by the shutdown and
9290 * ConditionSecurity= can now detect the TOMOYO security module.
9292 * Unit file [Install] sections are now also respected in unit drop-in
9293 files. This is intended to be used by drop-ins under /usr/lib/.
9295 * systemd-firstboot may now also set the initial keyboard mapping.
9297 * Udev "changed" events for devices which are exposed as systemd
9298 .device units are now propagated to units specified in
9299 ReloadPropagatedFrom= as reload requests.
9301 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
9302 unit template name (i.e. a name in the form of 'foobar@.service',
9303 without the instance component between the '@' and - the '.'), then
9304 the escaped sysfs path of the device is automatically used as the
9307 * SystemCallFilter= in unit files has been extended so that an "errno"
9308 can be specified individually for each system call. Example:
9309 SystemCallFilter=~uname:EILSEQ.
9311 * The cgroup delegation logic has been substantially updated. Delegate=
9312 now optionally takes a list of controllers (instead of a boolean, as
9313 before), which lists the controllers to delegate at least.
9315 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
9317 * A new LogLevelMax= setting configures the maximum log level any
9318 process of the service may log at (i.e. anything with a lesser
9319 priority than what is specified is automatically dropped). A new
9320 LogExtraFields= setting allows configuration of additional journal
9321 fields to attach to all log records generated by any of the unit's
9324 * New StandardInputData= and StandardInputText= settings along with the
9325 new option StandardInput=data may be used to configure textual or
9326 binary data that shall be passed to the executed service process via
9327 standard input, encoded in-line in the unit file.
9329 * StandardInput=, StandardOutput= and StandardError= may now be used to
9330 connect stdin/stdout/stderr of executed processes directly with a
9331 file or AF_UNIX socket in the file system, using the new "file:" option.
9333 * A new unit file option CollectMode= has been added, that allows
9334 tweaking the garbage collection logic for units. It may be used to
9335 tell systemd to garbage collect units that have failed automatically
9336 (normally it only GCs units that exited successfully). systemd-run
9337 and systemd-mount expose this new functionality with a new -G option.
9339 * "machinectl bind" may now be used to bind mount non-directories
9340 (i.e. regularfiles, devices, fifos, sockets).
9342 * systemd-analyze gained a new verb "calendar" for validating and
9343 testing calendar time specifications to use for OnCalendar= in timer
9344 units. Besides validating the expression it will calculate the next
9345 time the specified expression would elapse.
9347 * In addition to the pre-existing FailureAction= unit file setting
9348 there's now SuccessAction=, for configuring a shutdown action to
9349 execute when a unit completes successfully. This is useful in
9350 particular inside containers that shall terminate after some workload
9351 has been completed. Also, both options are now supported for all unit
9352 types, not just services.
9354 * networkds's IP rule support gained two new options
9355 IncomingInterface= and OutgoingInterface= for configuring the incoming
9356 and outgoing interfaces of configured rules. systemd-networkd also
9357 gained support for "vxcan" network devices.
9359 * networkd gained a new setting RequiredForOnline=, taking a
9360 boolean. If set, systemd-wait-online will take it into consideration
9361 when determining that the system is up, otherwise it will ignore the
9362 interface for this purpose.
9364 * The sd_notify() protocol gained support for a new operation: with
9365 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
9366 store again, ahead of POLLHUP or POLLERR when they are removed
9369 * A new document doc/UIDS-GIDS.md has been added to the source tree,
9370 that documents the UID/GID range and assignment assumptions and
9371 requirements of systemd.
9373 * The watchdog device PID 1 will ping may now be configured through the
9374 WatchdogDevice= configuration file setting, or by setting the
9375 systemd.watchdog_service= kernel command line option.
9377 * systemd-resolved's gained support for registering DNS-SD services on
9378 the local network using MulticastDNS. Services may either be
9379 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
9380 the same dir below /run, /usr/lib), or through its D-Bus API.
9382 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
9383 extend the effective start, runtime, and stop time. The service must
9384 continue to send EXTEND_TIMEOUT_USEC within the period specified to
9385 prevent the service manager from making the service as timedout.
9387 * systemd-resolved's DNSSEC support gained support for RFC 8080
9388 (Ed25519 keys and signatures).
9390 * The systemd-resolve command line tool gained a new set of options
9391 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
9392 --set-nta= and --revert to configure per-interface DNS configuration
9393 dynamically during runtime. It's useful for pushing DNS information
9394 into systemd-resolved from DNS hook scripts that various interface
9395 managing software supports (such as pppd).
9397 * systemd-nspawn gained a new --network-namespace-path= command line
9398 option, which may be used to make a container join an existing
9399 network namespace, by specifying a path to a "netns" file.
9401 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
9402 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
9403 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
9404 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
9405 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
9406 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
9407 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
9408 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
9409 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
9410 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
9411 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
9412 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
9413 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
9414 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
9415 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
9416 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
9417 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
9418 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
9419 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
9420 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
9421 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
9422 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
9423 Jędrzejewski-Szmek, Zeal Jagannatha
9425 — Berlin, 2017-12-14
9429 * INCOMPATIBILITY: systemd-logind.service and other long-running
9430 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
9431 communication with the outside. This generally improves security of
9432 the system, and is in almost all cases a safe and good choice, as
9433 these services do not and should not provide any network-facing
9434 functionality. However, systemd-logind uses the glibc NSS API to
9435 query the user database. This creates problems on systems where NSS
9436 is set up to directly consult network services for user database
9437 lookups. In particular, this creates incompatibilities with the
9438 "nss-nis" module, which attempts to directly contact the NIS/YP
9439 network servers it is configured for, and will now consistently
9440 fail. In such cases, it is possible to turn off IP sandboxing for
9441 systemd-logind.service (set IPAddressDeny= in its [Service] section
9442 to the empty string, via a .d/ unit file drop-in). Downstream
9443 distributions might want to update their nss-nis packaging to include
9444 such a drop-in snippet, accordingly, to hide this incompatibility
9445 from the user. Another option is to make use of glibc's nscd service
9446 to proxy such network requests through a privilege-separated, minimal
9447 local caching daemon, or to switch to more modern technologies such
9448 sssd, whose NSS hook-ups generally do not involve direct network
9449 access. In general, we think it's definitely time to question the
9450 implementation choices of nss-nis, i.e. whether it's a good idea
9451 today to embed a network-facing loadable module into all local
9452 processes that need to query the user database, including the most
9453 trivial and benign ones, such as "ls". For more details about
9454 IPAddressDeny= see below.
9456 * A new modprobe.d drop-in is now shipped by default that sets the
9457 bonding module option max_bonds=0. This overrides the kernel default,
9458 to avoid conflicts and ambiguity as to whether or not bond0 should be
9459 managed by systemd-networkd or not. This resolves multiple issues
9460 with bond0 properties not being applied, when bond0 is configured
9461 with systemd-networkd. Distributors may choose to not package this,
9462 however in that case users will be prevented from correctly managing
9463 bond0 interface using systemd-networkd.
9465 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
9466 which print the logging level and target of the system manager. They
9467 complement the existing "set-log-level" and "set-log-target" verbs
9468 used to change those values.
9470 * journald.conf gained a new boolean setting ReadKMsg= which defaults
9471 to on. If turned off kernel log messages will not be read by
9472 systemd-journald or included in the logs. It also gained a new
9473 setting LineMax= for configuring the maximum line length in
9474 STDOUT/STDERR log streams. The new default for this value is 48K, up
9475 from the previous hardcoded 2048.
9477 * A new unit setting RuntimeDirectoryPreserve= has been added, which
9478 allows more detailed control of what to do with a runtime directory
9479 configured with RuntimeDirectory= (i.e. a directory below /run or
9480 $XDG_RUNTIME_DIR) after a unit is stopped.
9482 * The RuntimeDirectory= setting for units gained support for creating
9483 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
9484 one top-level directory.
9486 * Units gained new options StateDirectory=, CacheDirectory=,
9487 LogsDirectory= and ConfigurationDirectory= which are closely related
9488 to RuntimeDirectory= but manage per-service directories below
9489 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
9490 possible to write unit files which when activated automatically gain
9491 properly owned service specific directories in these locations, thus
9492 making unit files self-contained and increasing compatibility with
9493 stateless systems and factory reset where /etc or /var are
9494 unpopulated at boot. Matching these new settings there's also
9495 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
9496 ConfigurationDirectoryMode= for configuring the access mode of these
9497 directories. These settings are particularly useful in combination
9498 with DynamicUser=yes as they provide secure, properly-owned,
9499 writable, and stateful locations for storage, excluded from the
9500 sandbox that such services live in otherwise.
9502 * Automake support has been removed from this release. systemd is now
9505 * systemd-journald will now aggressively cache client metadata during
9506 runtime, speeding up log write performance under pressure. This comes
9507 at a small price though: as much of the metadata is read
9508 asynchronously from /proc/ (and isn't implicitly attached to log
9509 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
9510 metadata stored alongside a log entry might be slightly
9511 out-of-date. Previously it could only be slightly newer than the log
9512 message. The time window is small however, and given that the kernel
9513 is unlikely to be improved anytime soon in this regard, this appears
9516 * nss-myhostname/systemd-resolved will now by default synthesize an
9517 A/AAAA resource record for the "_gateway" hostname, pointing to the
9518 current default IP gateway. Previously it did that for the "gateway"
9519 name, hampering adoption, as some distributions wanted to leave that
9520 hostname open for local use. The old behaviour may still be
9521 requested at build time.
9523 * systemd-networkd's [Address] section in .network files gained a new
9524 Scope= setting for configuring the IP address scope. The [Network]
9525 section gained a new boolean setting ConfigureWithoutCarrier= that
9526 tells systemd-networkd to ignore link sensing when configuring the
9527 device. The [DHCP] section gained a new Anonymize= boolean option for
9528 turning on a number of options suggested in RFC 7844. A new
9529 [RoutingPolicyRule] section has been added for configuring the IP
9530 routing policy. The [Route] section has gained support for a new
9531 Type= setting which permits configuring
9532 blackhole/unreachable/prohibit routes.
9534 * The [VRF] section in .netdev files gained a new Table= setting for
9535 configuring the routing table to use. The [Tunnel] section gained a
9536 new Independent= boolean field for configuring tunnels independent of
9537 an underlying network interface. The [Bridge] section gained a new
9538 GroupForwardMask= option for configuration of propagation of link
9539 local frames between bridge ports.
9541 * The WakeOnLan= setting in .link files gained support for a number of
9542 new modes. A new TCP6SegmentationOffload= setting has been added for
9543 configuring TCP/IPv6 hardware segmentation offload.
9545 * The IPv6 RA sender implementation may now optionally send out RDNSS
9546 and RDNSSL records to supply DNS configuration to peers.
9548 * systemd-nspawn gained support for a new --system-call-filter= command
9549 line option for adding and removing entries in the default system
9550 call filter it applies. Moreover systemd-nspawn has been changed to
9551 implement a system call allow list instead of a deny list.
9553 * systemd-run gained support for a new --pipe command line option. If
9554 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
9555 are directly passed on to the activated transient service
9556 executable. This allows invoking arbitrary processes as systemd
9557 services (for example to take benefit of dependency management,
9558 accounting management, resource management or log management that is
9559 done automatically for services) — while still allowing them to be
9560 integrated in a classic UNIX shell pipeline.
9562 * When a service sends RELOAD=1 via sd_notify() and reload propagation
9563 using ReloadPropagationTo= is configured, a reload is now propagated
9564 to configured units. (Previously this was only done on explicitly
9565 requested reloads, using "systemctl reload" or an equivalent
9568 * For each service unit a restart counter is now kept: it is increased
9569 each time the service is restarted due to Restart=, and may be
9570 queried using "systemctl show -p NRestarts …".
9572 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
9573 @signal and @timer have been added, for usage with SystemCallFilter=
9574 in unit files and the new --system-call-filter= command line option
9575 of systemd-nspawn (see above).
9577 * ExecStart= lines in unit files gained two new modifiers: when a
9578 command line is prefixed with "!" the command will be executed as
9579 configured, except for the credentials applied by
9580 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
9581 "+", but does still apply namespacing options unlike "+". There's
9582 also "!!" now, which is mostly identical, but becomes a NOP on
9583 systems that support ambient capabilities. This is useful to write
9584 unit files that work with ambient capabilities where possible but
9585 automatically fall back to traditional privilege dropping mechanisms
9586 on systems where this is not supported.
9588 * ListenNetlink= settings in socket units now support RDMA netlink
9591 * A new unit file setting LockPersonality= has been added which permits
9592 locking down the chosen execution domain ("personality") of a service
9595 * A new special target "getty-pre.target" has been added, which is
9596 ordered before all text logins, and may be used to order services
9597 before textual logins acquire access to the console.
9599 * systemd will now attempt to load the virtio-rng.ko kernel module very
9600 early on if a VM environment supporting this is detected. This should
9601 improve entropy during early boot in virtualized environments.
9603 * A _netdev option is now supported in /etc/crypttab that operates in a
9604 similar way as the same option in /etc/fstab: it permits configuring
9605 encrypted devices that need to be ordered after the network is up.
9606 Following this logic, two new special targets
9607 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
9608 added that are to cryptsetup.target what remote-fs.target and
9609 remote-fs-pre.target are to local-fs.target.
9611 * Service units gained a new UnsetEnvironment= setting which permits
9612 unsetting specific environment variables for services that are
9613 normally passed to it (for example in order to mask out locale
9614 settings for specific services that can't deal with it).
9616 * Units acquired a new boolean option IPAccounting=. When turned on, IP
9617 traffic accounting (packet count as well as byte count) is done for
9618 the service, and shown as part of "systemctl status" or "systemd-run
9621 * Service units acquired two new options IPAddressAllow= and
9622 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
9623 for configuring a simple IP access control list for all sockets of
9624 the unit. These options are available also on .slice and .socket
9625 units, permitting flexible access list configuration for individual
9626 services as well as groups of services (as defined by a slice unit),
9627 including system-wide. Note that IP ACLs configured this way are
9628 enforced on every single IPv4 and IPv6 socket created by any process
9629 of the service unit, and apply to ingress as well as egress traffic.
9631 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
9632 structured log message is generated each time the unit is stopped,
9633 containing information about the consumed resources of this
9636 * A new setting KeyringMode= has been added to unit files, which may be
9637 used to control how the kernel keyring is set up for executed
9640 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
9641 "systemctl kexec" and "systemctl exit" are now always asynchronous in
9642 behaviour (that is: these commands return immediately after the
9643 operation was enqueued instead of waiting for the operation to
9644 complete). Previously, "systemctl poweroff" and "systemctl reboot"
9645 were asynchronous on systems using systemd-logind (i.e. almost
9646 always, and like they were on sysvinit), and the other three commands
9647 were unconditionally synchronous. With this release this is cleaned
9648 up, and callers will see the same asynchronous behaviour on all
9649 systems for all five operations.
9651 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
9654 * .timer units now accept calendar specifications in other timezones
9655 than UTC or the local timezone.
9657 * The tmpfiles snippet var.conf has been changed to create
9658 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
9659 the "utmp" group already, and it appears to be generally understood
9660 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
9661 databases. Previously this was implemented correctly for all these
9662 databases excepts btmp, which has been opened up like this now
9663 too. Note that while the other databases are world-readable
9664 (i.e. 0644), btmp is not and remains more restrictive.
9666 * The systemd-resolve tool gained a new --reset-server-features
9667 switch. When invoked like this systemd-resolved will forget
9668 everything it learnt about the features supported by the configured
9669 upstream DNS servers, and restarts the feature probing logic on the
9670 next resolver look-up for them at the highest feature level
9673 * The status dump systemd-resolved sends to the logs upon receiving
9674 SIGUSR1 now also includes information about all DNS servers it is
9675 configured to use, and the features levels it probed for them.
9677 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
9678 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
9679 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
9680 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
9681 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
9682 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
9683 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
9684 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
9685 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
9686 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
9687 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
9688 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
9689 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
9690 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
9691 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
9692 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
9693 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
9694 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
9695 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
9696 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9698 — Berlin, 2017-10-06
9702 * Meson is now supported as build system in addition to Automake. It is
9703 our plan to remove Automake in one of our next releases, so that
9704 Meson becomes our exclusive build system. Hence, please start using
9705 the Meson build system in your downstream packaging. There's plenty
9706 of documentation around how to use Meson, the extremely brief
9709 ./autogen.sh && ./configure && make && sudo make install
9713 meson build && ninja -C build && sudo ninja -C build install
9715 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
9716 which permits configuring a timeout on the time a job is
9717 running. This is particularly useful for setting timeouts on jobs for
9720 * Unit files gained two new options ConditionUser= and ConditionGroup=
9721 for conditionalizing units based on the identity of the user/group
9722 running a systemd user instance.
9724 * systemd-networkd now understands a new FlowLabel= setting in the
9725 [VXLAN] section of .network files, as well as a Priority= in
9726 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
9727 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
9728 gained support for configuration of GENEVE links, and IPv6 address
9729 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
9731 * .link files now understand a new Port= setting.
9733 * systemd-networkd's DHCP support gained support for DHCP option 119
9734 (domain search list).
9736 * systemd-networkd gained support for serving IPv6 address ranges using
9737 the Router Advertisement protocol. The new .network configuration
9738 section [IPv6Prefix] may be used to configure the ranges to
9739 serve. This is implemented based on a new, minimal, native server
9740 implementation of RA.
9742 * journalctl's --output= switch gained support for a new parameter
9743 "short-iso-precise" for a mode where timestamps are shown as precise
9746 * systemd-udevd's "net_id" builtin may now generate stable network
9747 interface names from IBM PowerVM VIO devices as well as ACPI platform
9750 * MulticastDNS support in systemd-resolved may now be explicitly
9751 enabled/disabled using the new MulticastDNS= configuration file
9754 * systemd-resolved may now optionally use libidn2 instead of the libidn
9755 for processing internationalized domain names. Support for libidn2
9756 should be considered experimental and should not be enabled by
9759 * "machinectl pull-tar" and related call may now do verification of
9760 downloaded images using SUSE-style .sha256 checksum files in addition
9761 to the already existing support for validating using Ubuntu-style
9764 * sd-bus gained support for a new sd_bus_message_appendv() call which
9765 is va_list equivalent of sd_bus_message_append().
9767 * sd-boot gained support for validating images using SHIM/MOK.
9769 * The SMACK code learnt support for "onlycap".
9771 * systemd-mount --umount is now much smarter in figuring out how to
9772 properly unmount a device given its mount or device path.
9774 * The code to call libnss_dns as a fallback from libnss_resolve when
9775 the communication with systemd-resolved fails was removed. This
9776 fallback was redundant and interfered with the [!UNAVAIL=return]
9777 suffix. See nss-resolve(8) for the recommended configuration.
9779 * systemd-logind may now be restarted without losing state. It stores
9780 the file descriptors for devices it manages in the system manager
9781 using the FDSTORE= mechanism. Please note that further changes in
9782 other components may be required to make use of this (for example
9783 Xorg has code to listen for stops of systemd-logind and terminate
9784 itself when logind is stopped or restarted, in order to avoid using
9785 stale file descriptors for graphical devices, which is now
9786 counterproductive and must be reverted in order for restarts of
9787 systemd-logind to be safe. See
9788 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
9790 * All kernel-install plugins are called with the environment variable
9791 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
9792 /etc/machine-id. If the machine ID could not be determined,
9793 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
9794 anything in the entry directory (passed as the second argument) if
9795 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
9796 temporary directory is passed as the entry directory and removed
9797 after all the plugins exit.
9799 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
9800 will now use its value as the machine ID instead of the machine ID
9801 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
9802 /etc/machine-info and no machine ID is set in /etc/machine-id,
9803 kernel-install will try to store the current machine ID there as
9804 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
9805 will generate a new UUID, store it in /etc/machine-info as
9806 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
9808 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
9809 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
9810 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
9811 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
9812 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
9813 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
9814 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
9815 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
9816 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
9817 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
9818 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
9819 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
9820 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
9821 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
9822 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
9823 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
9824 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
9825 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
9826 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
9827 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
9828 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
9829 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
9830 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
9831 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
9832 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
9833 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
9834 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
9835 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
9836 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
9839 — Berlin, 2017-07-12
9843 * The "hybrid" control group mode has been modified to improve
9844 compatibility with "legacy" cgroups-v1 setups. Specifically, the
9845 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
9846 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
9847 cgroups-v1 hierarchy), the only externally visible change being that
9848 the cgroups-v2 hierarchy is also mounted, to
9849 /sys/fs/cgroup/unified. This should provide a large degree of
9850 compatibility with "legacy" cgroups-v1, while taking benefit of the
9851 better management capabilities of cgroups-v2.
9853 * The default control group setup mode may be selected both a boot-time
9854 via a set of kernel command line parameters (specifically:
9855 systemd.unified_cgroup_hierarchy= and
9856 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
9857 default selected on the configure command line
9858 (--with-default-hierarchy=). The upstream default is "hybrid"
9859 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
9860 this will change in a future systemd version to be "unified" (pure
9861 cgroups-v2 mode). The third option for the compile time option is
9862 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
9863 distributions to default to "hybrid" mode for release distributions,
9864 starting with v233. We recommend "unified" for development
9865 distributions (specifically: distributions such as Fedora's rawhide)
9866 as that's where things are headed in the long run. Use "legacy" for
9867 greatest stability and compatibility only.
9869 * Note one current limitation of "unified" and "hybrid" control group
9870 setup modes: the kernel currently does not permit the systemd --user
9871 instance (i.e. unprivileged code) to migrate processes between two
9872 disconnected cgroup subtrees, even if both are managed and owned by
9873 the user. This effectively means "systemd-run --user --scope" doesn't
9874 work when invoked from outside of any "systemd --user" service or
9875 scope. Specifically, it is not supported from session scopes. We are
9876 working on fixing this in a future systemd version. (See #3388 for
9877 further details about this.)
9879 * DBus policy files are now installed into /usr rather than /etc. Make
9880 sure your system has dbus >= 1.9.18 running before upgrading to this
9881 version, or override the install path with --with-dbuspolicydir= .
9883 * All python scripts shipped with systemd (specifically: the various
9884 tests written in Python) now require Python 3.
9886 * systemd unit tests can now run standalone (without the source or
9887 build directories), and can be installed into /usr/lib/systemd/tests/
9888 with 'make install-tests'.
9890 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
9891 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
9894 * Support for the %c, %r, %R specifiers in unit files has been
9895 removed. Specifiers are not supposed to be dependent on configuration
9896 in the unit file itself (so that they resolve the same regardless
9897 where used in the unit files), but these specifiers were influenced
9898 by the Slice= option.
9900 * The shell invoked by debug-shell.service now defaults to /bin/sh in
9901 all cases. If distributions want to use a different shell for this
9902 purpose (for example Fedora's /sbin/sushell) they need to specify
9903 this explicitly at configure time using --with-debug-shell=.
9905 * The confirmation spawn prompt has been reworked to offer the
9908 (c)ontinue, proceed without asking anymore
9909 (D)ump, show the state of the unit
9910 (f)ail, don't execute the command and pretend it failed
9912 (i)nfo, show a short summary of the unit
9913 (j)obs, show jobs that are in progress
9914 (s)kip, don't execute the command and pretend it succeeded
9915 (y)es, execute the command
9917 The 'n' choice for the confirmation spawn prompt has been removed,
9918 because its meaning was confusing.
9920 The prompt may now also be redirected to an alternative console by
9921 specifying the console as parameter to systemd.confirm_spawn=.
9923 * Services of Type=notify require a READY=1 notification to be sent
9924 during startup. If no such message is sent, the service now fails,
9925 even if the main process exited with a successful exit code.
9927 * Services that fail to start up correctly now always have their
9928 ExecStopPost= commands executed. Previously, they'd enter "failed"
9929 state directly, without executing these commands.
9931 * The option MulticastDNS= of network configuration files has acquired
9932 an actual implementation. With MulticastDNS=yes a host can resolve
9933 names of remote hosts and reply to mDNS A and AAAA requests.
9935 * When units are about to be started an additional check is now done to
9936 ensure that all dependencies of type BindsTo= (when used in
9937 combination with After=) have been started.
9939 * systemd-analyze gained a new verb "syscall-filter" which shows which
9940 system call groups are defined for the SystemCallFilter= unit file
9941 setting, and which system calls they contain.
9943 * A new system call filter group "@filesystem" has been added,
9944 consisting of various file system related system calls. Group
9945 "@reboot" has been added, covering reboot, kexec and shutdown related
9946 calls. Finally, group "@swap" has been added covering swap
9947 configuration related calls.
9949 * A new unit file option RestrictNamespaces= has been added that may be
9950 used to restrict access to the various process namespace types the
9951 Linux kernel provides. Specifically, it may be used to take away the
9952 right for a service unit to create additional file system, network,
9953 user, and other namespaces. This sandboxing option is particularly
9954 relevant due to the high amount of recently discovered namespacing
9955 related vulnerabilities in the kernel.
9957 * systemd-udev's .link files gained support for a new AutoNegotiation=
9958 setting for configuring Ethernet auto-negotiation.
9960 * systemd-networkd's .network files gained support for a new
9961 ListenPort= setting in the [DHCP] section to explicitly configure the
9962 UDP client port the DHCP client shall listen on.
9964 * .network files gained a new Unmanaged= boolean setting for explicitly
9965 excluding one or more interfaces from management by systemd-networkd.
9967 * The systemd-networkd ProxyARP= option has been renamed to
9968 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
9969 renamed to ReduceARPProxy=. The old names continue to be available
9972 * systemd-networkd gained support for configuring IPv6 Proxy NDP
9973 addresses via the new IPv6ProxyNDPAddress= .network file setting.
9975 * systemd-networkd's bonding device support gained support for two new
9976 configuration options ActiveSlave= and PrimarySlave=.
9978 * The various options in the [Match] section of .network files gained
9979 support for negative matching.
9981 * New systemd-specific mount options are now understood in /etc/fstab:
9983 x-systemd.mount-timeout= may be used to configure the maximum
9984 permitted runtime of the mount command.
9986 x-systemd.device-bound may be set to bind a mount point to its
9987 backing device unit, in order to automatically remove a mount point
9988 if its backing device is unplugged. This option may also be
9989 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
9990 on the block device, which is now automatically set for all CDROM
9991 drives, so that mounted CDs are automatically unmounted when they are
9992 removed from the drive.
9994 x-systemd.after= and x-systemd.before= may be used to explicitly
9995 order a mount after or before another unit or mount point.
9997 * Enqueued start jobs for device units are now automatically garbage
9998 collected if there are no jobs waiting for them anymore.
10000 * systemctl list-jobs gained two new switches: with --after, for every
10001 queued job the jobs it's waiting for are shown; with --before the
10002 jobs which it's blocking are shown.
10004 * systemd-nspawn gained support for ephemeral boots from disk images
10005 (or in other words: --ephemeral and --image= may now be
10006 combined). Moreover, ephemeral boots are now supported for normal
10007 directories, even if the backing file system is not btrfs. Of course,
10008 if the file system does not support file system snapshots or
10009 reflinks, the initial copy operation will be relatively expensive, but
10010 this should still be suitable for many use cases.
10012 * Calendar time specifications in .timer units now support
10013 specifications relative to the end of a month by using "~" instead of
10014 "-" as separator between month and day. For example, "*-02~03" means
10015 "the third last day in February". In addition a new syntax for
10016 repeated events has been added using the "/" character. For example,
10017 "9..17/2:00" means "every two hours from 9am to 5pm".
10019 * systemd-socket-proxyd gained a new parameter --connections-max= for
10020 configuring the maximum number of concurrent connections.
10022 * sd-id128 gained a new API for generating unique IDs for the host in a
10023 way that does not leak the machine ID. Specifically,
10024 sd_id128_get_machine_app_specific() derives an ID based on the
10025 machine ID in a well-defined, non-reversible, stable way. This is
10026 useful whenever an identifier for the host is needed but where the
10027 identifier shall not be useful to identify the system beyond the
10028 scope of the application itself. (Internally this uses HMAC-SHA256 as
10029 keyed hash function using the machine ID as input.)
10031 * NotifyAccess= gained a new supported value "exec". When set
10032 notifications are accepted from all processes systemd itself invoked,
10033 including all control processes.
10035 * .nspawn files gained support for defining overlay mounts using the
10036 Overlay= and OverlayReadOnly= options. Previously this functionality
10037 was only available on the systemd-nspawn command line.
10039 * systemd-nspawn's --bind= and --overlay= options gained support for
10040 bind/overlay mounts whose source lies within the container tree by
10041 prefixing the source path with "+".
10043 * systemd-nspawn's --bind= and --overlay= options gained support for
10044 automatically allocating a temporary source directory in /var/tmp
10045 that is removed when the container dies. Specifically, if the source
10046 directory is specified as empty string this mechanism is selected. An
10047 example usage is --overlay=+/var::/var, which creates an overlay
10048 mount based on the original /var contained in the image, overlaid
10049 with a temporary directory in the host's /var/tmp. This way changes
10050 to /var are automatically flushed when the container shuts down.
10052 * systemd-nspawn --image= option does now permit raw file system block
10053 devices (in addition to images containing partition tables, as
10056 * The disk image dissection logic in systemd-nspawn gained support for
10057 automatically setting up LUKS encrypted as well as Verity protected
10058 partitions. When a container is booted from an encrypted image the
10059 passphrase is queried at start-up time. When a container with Verity
10060 data is started, the root hash is search in a ".roothash" file
10061 accompanying the disk image (alternatively, pass the root hash via
10062 the new --root-hash= command line option).
10064 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
10065 be used to dissect disk images the same way as systemd-nspawn does
10066 it, following the Bootable Partition Specification. It may even be
10067 used to mount disk images with complex partition setups (including
10068 LUKS and Verity partitions) to a local host directory, in order to
10069 inspect them. This tool is not considered public API (yet), and is
10070 thus not installed into /usr/bin. Please do not rely on its
10071 existence, since it might go away or be changed in later systemd
10074 * A new generator "systemd-verity-generator" has been added, similar in
10075 style to "systemd-cryptsetup-generator", permitting automatic setup of
10076 Verity root partitions when systemd boots up. In order to make use of
10077 this your partition setup should follow the Discoverable Partitions
10078 Specification, and the GPT partition ID of the root file system
10079 partition should be identical to the upper 128-bit of the Verity root
10080 hash. The GPT partition ID of the Verity partition protecting it
10081 should be the lower 128-bit of the Verity root hash. If the partition
10082 image follows this model it is sufficient to specify a single
10083 "roothash=" kernel command line argument to both configure which root
10084 image and verity partition to use as well as the root hash for
10085 it. Note that systemd-nspawn's Verity support follows the same
10086 semantics, meaning that disk images with proper Verity data in place
10087 may be booted in containers with systemd-nspawn as well as on
10088 physical systems via the verity generator. Also note that the "mkosi"
10089 tool available at https://github.com/systemd/mkosi has been updated
10090 to generate Verity protected disk images following this scheme. In
10091 fact, it has been updated to generate disk images that optionally
10092 implement a complete UEFI SecureBoot trust chain, involving a signed
10093 kernel and initrd image that incorporates such a root hash as well as
10094 a Verity-enabled root partition.
10096 * The hardware database (hwdb) udev supports has been updated to carry
10097 accelerometer quirks.
10099 * All system services are now run with a fresh kernel keyring set up
10100 for them. The invocation ID is stored by default in it, thus
10101 providing a safe, non-overridable way to determine the invocation
10102 ID of each service.
10104 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
10105 options for bind mounting arbitrary paths in a service-specific
10106 way. When these options are used, arbitrary host or service files and
10107 directories may be mounted to arbitrary locations in the service's
10110 * Documentation has been added that lists all of systemd's low-level
10111 environment variables:
10113 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
10115 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
10116 whether a specific socket file descriptor matches a specified socket
10119 * systemd-firstboot has been updated to check for the
10120 systemd.firstboot= kernel command line option. It accepts a boolean
10121 and when set to false the first boot questions are skipped.
10123 * systemd-fstab-generator has been updated to check for the
10124 systemd.volatile= kernel command line option, which either takes an
10125 optional boolean parameter or the special value "state". If used the
10126 system may be booted in a "volatile" boot mode. Specifically,
10127 "systemd.volatile" is used, the root directory will be mounted as
10128 tmpfs, and only /usr is mounted from the actual root file system. If
10129 "systemd.volatile=state" is used, the root directory will be mounted
10130 as usual, but /var is mounted as tmpfs. This concept provides similar
10131 functionality as systemd-nspawn's --volatile= option, but provides it
10132 on physical boots. Use this option for implementing stateless
10133 systems, or testing systems with all state and/or configuration reset
10134 to the defaults. (Note though that many distributions are not
10135 prepared to boot up without a populated /etc or /var, though.)
10137 * systemd-gpt-auto-generator gained support for LUKS encrypted root
10138 partitions. Previously it only supported LUKS encrypted partitions
10139 for all other uses, except for the root partition itself.
10141 * Socket units gained support for listening on AF_VSOCK sockets for
10142 communication in virtualized QEMU environments.
10144 * The "configure" script gained a new option --with-fallback-hostname=
10145 for specifying the fallback hostname to use if none is configured in
10146 /etc/hostname. For example, by specifying
10147 --with-fallback-hostname=fedora it is possible to default to a
10148 hostname of "fedora" on pristine installations.
10150 * systemd-cgls gained support for a new --unit= switch for listing only
10151 the control groups of a specific unit. Similar --user-unit= has been
10152 added for listing only the control groups of a specific user unit.
10154 * systemd-mount gained a new --umount switch for unmounting a mount or
10155 automount point (and all mount/automount points below it).
10157 * systemd will now refuse full configuration reloads (via systemctl
10158 daemon-reload and related calls) unless at least 16MiB of free space
10159 are available in /run. This is a safety precaution in order to ensure
10160 that generators can safely operate after the reload completed.
10162 * A new unit file option RootImage= has been added, which has a similar
10163 effect as RootDirectory= but mounts the service's root directory from
10164 a disk image instead of plain directory. This logic reuses the same
10165 image dissection and mount logic that systemd-nspawn already uses,
10166 and hence supports any disk images systemd-nspawn supports, including
10167 those following the Discoverable Partition Specification, as well as
10168 Verity enabled images. This option enables systemd to run system
10169 services directly off disk images acting as resource bundles,
10170 possibly even including full integrity data.
10172 * A new MountAPIVFS= unit file option has been added, taking a boolean
10173 argument. If enabled /proc, /sys and /dev (collectively called the
10174 "API VFS") will be mounted for the service. This is only relevant if
10175 RootDirectory= or RootImage= is used for the service, as these mounts
10176 are of course in place in the host mount namespace anyway.
10178 * systemd-nspawn gained support for a new --pivot-root= switch. If
10179 specified the root directory within the container image is pivoted to
10180 the specified mount point, while the original root disk is moved to a
10181 different place. This option enables booting of ostree images
10182 directly with systemd-nspawn.
10184 * The systemd build scripts will no longer complain if the NTP server
10185 addresses are not changed from the defaults. Google now supports
10186 these NTP servers officially. We still recommend downstreams to
10187 properly register an NTP pool with the NTP pool project though.
10189 * coredumpctl gained a new "--reverse" option for printing the list
10190 of coredumps in reverse order.
10192 * coredumpctl will now show additional information about truncated and
10193 inaccessible coredumps, as well as coredumps that are still being
10194 processed. It also gained a new --quiet switch for suppressing
10195 additional informational message in its output.
10197 * coredumpctl gained support for only showing coredumps newer and/or
10198 older than specific timestamps, using the new --since= and --until=
10199 options, reminiscent of journalctl's options by the same name.
10201 * The systemd-coredump logic has been improved so that it may be reused
10202 to collect backtraces in non-compiled languages, for example in
10203 scripting languages such as Python.
10205 * machinectl will now show the UID shift of local containers, if user
10206 namespacing is enabled for them.
10208 * systemd will now optionally run "environment generator" binaries at
10209 configuration load time. They may be used to add environment
10210 variables to the environment block passed to services invoked. One
10211 user environment generator is shipped by default that sets up
10212 environment variables based on files dropped into /etc/environment.d
10213 and ~/.config/environment.d/.
10215 * systemd-resolved now includes the new, recently published 2017 DNSSEC
10218 * hostnamed has been updated to report a new chassis type of
10219 "convertible" to cover "foldable" laptops that can both act as a
10220 tablet and as a laptop, such as various Lenovo Yoga devices.
10222 Contributions from: Adrián López, Alexander Galanin, Alexander
10223 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
10224 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
10225 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
10226 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
10227 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
10228 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
10229 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
10230 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
10231 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
10232 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
10233 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
10234 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
10235 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
10236 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
10237 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
10238 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
10239 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
10240 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
10241 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
10242 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
10243 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
10244 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
10245 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
10246 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
10247 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
10248 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
10249 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
10252 — Berlin, 2017-03-01
10256 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
10257 RestrictAddressFamilies= enabled. These sandboxing options should
10258 generally be compatible with the various external udev call-out
10259 binaries we are aware of, however there may be exceptions, in
10260 particular when exotic languages for these call-outs are used. In
10261 this case, consider turning off these settings locally.
10263 * The new RemoveIPC= option can be used to remove IPC objects owned by
10264 the user or group of a service when that service exits.
10266 * The new ProtectKernelModules= option can be used to disable explicit
10267 load and unload operations of kernel modules by a service. In
10268 addition access to /usr/lib/modules is removed if this option is set.
10270 * ProtectSystem= option gained a new value "strict", which causes the
10271 whole file system tree with the exception of /dev, /proc, and /sys,
10272 to be remounted read-only for a service.
10274 * The new ProtectKernelTunables= option can be used to disable
10275 modification of configuration files in /sys and /proc by a service.
10276 Various directories and files are remounted read-only, so access is
10277 restricted even if the file permissions would allow it.
10279 * The new ProtectControlGroups= option can be used to disable write
10280 access by a service to /sys/fs/cgroup.
10282 * Various systemd services have been hardened with
10283 ProtectKernelTunables=yes, ProtectControlGroups=yes,
10284 RestrictAddressFamilies=.
10286 * Support for dynamically creating users for the lifetime of a service
10287 has been added. If DynamicUser=yes is specified, user and group IDs
10288 will be allocated from the range 61184…65519 for the lifetime of the
10289 service. They can be resolved using the new nss-systemd.so NSS
10290 module. The module must be enabled in /etc/nsswitch.conf. Services
10291 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
10292 any resources allocated by the service will be cleaned up when the
10293 service exits. They also have ProtectHome=read-only and
10294 ProtectSystem=strict enabled, so they are not able to make any
10295 permanent modifications to the system.
10297 * The nss-systemd module also always resolves root and nobody, making
10298 it possible to have no /etc/passwd or /etc/group files in minimal
10299 container or chroot environments.
10301 * Services may be started with their own user namespace using the new
10302 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
10303 under which the service is running are mapped. All other users are
10306 * Support for the cgroup namespace has been added to systemd-nspawn. If
10307 supported by kernel, the container system started by systemd-nspawn
10308 will have its own view of the cgroup hierarchy. This new behaviour
10309 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
10311 * The new MemorySwapMax= option can be used to limit the maximum swap
10312 usage under the unified cgroup hierarchy.
10314 * Support for the CPU controller in the unified cgroup hierarchy has
10315 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
10316 options. This controller requires out-of-tree patches for the kernel
10317 and the support is provisional.
10319 * Mount and automount units may now be created transiently
10320 (i.e. dynamically at runtime via the bus API, instead of requiring
10321 unit files in the file system).
10323 * systemd-mount is a new tool which may mount file systems – much like
10324 mount(8), optionally pulling in additional dependencies through
10325 transient .mount and .automount units. For example, this tool
10326 automatically runs fsck on a backing block device before mounting,
10327 and allows the automount logic to be used dynamically from the
10328 command line for establishing mount points. This tool is particularly
10329 useful when dealing with removable media, as it will ensure fsck is
10330 run – if necessary – before the first access and that the file system
10331 is quickly unmounted after each access by utilizing the automount
10332 logic. This maximizes the chance that the file system on the
10333 removable media stays in a clean state, and if it isn't in a clean
10334 state is fixed automatically.
10336 * LazyUnmount=yes option for mount units has been added to expose the
10337 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
10340 * /efi will be used as the mount point of the EFI boot partition, if
10341 the directory is present, and the mount point was not configured
10342 through other means (e.g. fstab). If /efi directory does not exist,
10343 /boot will be used as before. This makes it easier to automatically
10344 mount the EFI partition on systems where /boot is used for something
10347 * When operating on GPT disk images for containers, systemd-nspawn will
10348 now mount the ESP to /boot or /efi according to the same rules as PID
10349 1 running on a host. This allows tools like "bootctl" to operate
10350 correctly within such containers, in order to make container images
10351 bootable on physical systems.
10353 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
10355 * Two new user session targets have been added to support running
10356 graphical sessions under the systemd --user instance:
10357 graphical-session.target and graphical-session-pre.target. See
10358 systemd.special(7) for a description of how those targets should be
10361 * The vconsole initialization code has been significantly reworked to
10362 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
10363 support unicode keymaps. Font and keymap configuration will now be
10364 copied to all allocated virtual consoles.
10366 * FreeBSD's bhyve virtualization is now detected.
10368 * Information recorded in the journal for core dumps now includes the
10369 contents of /proc/mountinfo and the command line of the process at
10370 the top of the process hierarchy (which is usually the init process
10373 * systemd-journal-gatewayd learned the --directory= option to serve
10374 files from the specified location.
10376 * journalctl --root=… can be used to peruse the journal in the
10377 /var/log/ directories inside of a container tree. This is similar to
10378 the existing --machine= option, but does not require the container to
10381 * The hardware database has been extended to support
10382 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
10385 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
10386 specify the click rate for mice which include a horizontal wheel with
10387 a click rate that is different than the one for the vertical wheel.
10389 * systemd-run gained a new --wait option that makes service execution
10390 synchronous. (Specifically, the command will not return until the
10391 specified service binary exited.)
10393 * systemctl gained a new --wait option that causes the start command to
10394 wait until the units being started have terminated again.
10396 * A new journal output mode "short-full" has been added which displays
10397 timestamps with abbreviated English day names and adds a timezone
10398 suffix. Those timestamps include more information than the default
10399 "short" output mode, and can be passed directly to journalctl's
10400 --since= and --until= options.
10402 * /etc/resolv.conf will be bind-mounted into containers started by
10403 systemd-nspawn, if possible, so any changes to resolv.conf contents
10404 are automatically propagated to the container.
10406 * The number of instances for socket-activated services originating
10407 from a single IP address can be limited with
10408 MaxConnectionsPerSource=, extending the existing setting of
10411 * systemd-networkd gained support for vcan ("Virtual CAN") interface
10414 * .netdev and .network configuration can now be extended through
10417 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
10418 Segmentation Offload, Generic Receive Offload, Large Receive Offload
10419 can be enabled and disabled using the new UDPSegmentationOffload=,
10420 TCPSegmentationOffload=, GenericSegmentationOffload=,
10421 GenericReceiveOffload=, LargeReceiveOffload= options in the
10422 [Link] section of .link files.
10424 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
10425 Port VLAN ID can be configured for bridge devices using the new STP=,
10426 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
10427 section of .netdev files.
10429 * The route table to which routes received over DHCP or RA should be
10430 added can be configured with the new RouteTable= option in the [DHCP]
10431 and [IPv6AcceptRA] sections of .network files.
10433 * The Address Resolution Protocol can be disabled on links managed by
10434 systemd-networkd using the ARP=no setting in the [Link] section of
10437 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
10438 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
10439 encode information about the result and exit codes of the current
10440 service runtime cycle.
10442 * systemd-sysctl will now configure kernel parameters in the order
10443 they occur in the configuration files. This matches what sysctl
10444 has been traditionally doing.
10446 * kernel-install "plugins" that are executed to perform various
10447 tasks after a new kernel is added and before an old one is removed
10448 can now return a special value to terminate the procedure and
10449 prevent any later plugins from running.
10451 * Journald's SplitMode=login setting has been deprecated. It has been
10452 removed from documentation, and its use is discouraged. In a future
10453 release it will be completely removed, and made equivalent to current
10454 default of SplitMode=uid.
10456 * Storage=both option setting in /etc/systemd/coredump.conf has been
10457 removed. With fast LZ4 compression storing the core dump twice is not
10460 * The --share-system systemd-nspawn option has been replaced with an
10461 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
10462 this functionality is discouraged. In addition the variables
10463 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
10464 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
10465 individual namespaces.
10467 * "machinectl list" now shows the IP address of running containers in
10468 the output, as well as OS release information.
10470 * "loginctl list" now shows the TTY of each session in the output.
10472 * sd-bus gained new API calls sd_bus_track_set_recursive(),
10473 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
10474 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
10475 tracking objects in a "recursive" mode, where a single client can be
10476 counted multiple times, if it takes multiple references.
10478 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
10479 sd_bus_get_exit_on_disconnect(). They may be used to make a
10480 process using sd-bus automatically exit if the bus connection is
10483 * Bus clients of the service manager may now "pin" loaded units into
10484 memory, by taking an explicit reference on them. This is useful to
10485 ensure the client can retrieve runtime data about the service even
10486 after the service completed execution. Taking such a reference is
10487 available only for privileged clients and should be helpful to watch
10488 running services in a race-free manner, and in particular collect
10489 information about exit statuses and results.
10491 * The nss-resolve module has been changed to strictly return UNAVAIL
10492 when communication via D-Bus with resolved failed, and NOTFOUND when
10493 a lookup completed but was negative. This means it is now possible to
10494 neatly configure fallbacks using nsswitch.conf result checking
10495 expressions. Taking benefit of this, the new recommended
10496 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
10498 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
10500 * A new setting CtrlAltDelBurstAction= has been added to
10501 /etc/systemd/system.conf which may be used to configure the precise
10502 behaviour if the user on the console presses Ctrl-Alt-Del more often
10503 than 7 times in 2s. Previously this would unconditionally result in
10504 an expedited, immediate reboot. With this new setting the precise
10505 operation may be configured in more detail, and also turned off
10508 * In .netdev files two new settings RemoteChecksumTx= and
10509 RemoteChecksumRx= are now understood that permit configuring the
10510 remote checksumming logic for VXLAN networks.
10512 * The service manager learnt a new "invocation ID" concept for invoked
10513 services. Each runtime cycle of a service will get a new invocation
10514 ID (a 128-bit random UUID) assigned that identifies the current
10515 run of the service uniquely and globally. A new invocation ID
10516 is generated each time a service starts up. The journal will store
10517 the invocation ID of a service along with any logged messages, thus
10518 making the invocation ID useful for matching the online runtime of a
10519 service with the offline log data it generated in a safe way without
10520 relying on synchronized timestamps. In many ways this new service
10521 invocation ID concept is similar to the kernel's boot ID concept that
10522 uniquely and globally identifies the runtime of each boot. The
10523 invocation ID of a service is passed to the service itself via an
10524 environment variable ($INVOCATION_ID). A new bus call
10525 GetUnitByInvocationID() has been added that is similar to GetUnit()
10526 but instead of retrieving the bus path for a unit by its name
10527 retrieves it by its invocation ID. The returned path is valid only as
10528 long as the passed invocation ID is current.
10530 * systemd-resolved gained a new "DNSStubListener" setting in
10531 resolved.conf. It either takes a boolean value or the special values
10532 "udp" and "tcp", and configures whether to enable the stub DNS
10533 listener on 127.0.0.53:53.
10535 * IP addresses configured via networkd may now carry additional
10536 configuration settings supported by the kernel. New options include:
10537 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
10538 PrefixRoute=, AutoJoin=.
10540 * The PAM configuration fragment file for "user@.service" shipped with
10541 systemd (i.e. the --user instance of systemd) has been stripped to
10542 the minimum necessary to make the system boot. Previously, it
10543 contained Fedora-specific stanzas that did not apply to other
10544 distributions. It is expected that downstream distributions add
10545 additional configuration lines, matching their needs to this file,
10546 using it only as rough template of what systemd itself needs. Note
10547 that this reduced fragment does not even include an invocation of
10548 pam_limits which most distributions probably want to add, even though
10549 systemd itself does not need it. (There's also the new build time
10550 option --with-pamconfdir=no to disable installation of the PAM
10551 fragment entirely.)
10553 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
10554 capability is now also dropped from its set (in addition to
10555 CAP_SYS_MKNOD as before).
10557 * In service unit files it is now possible to connect a specific named
10558 file descriptor with stdin/stdout/stdout of an executed service. The
10559 name may be specified in matching .socket units using the
10560 FileDescriptorName= setting.
10562 * A number of journal settings may now be configured on the kernel
10563 command line. Specifically, the following options are now understood:
10564 systemd.journald.max_level_console=,
10565 systemd.journald.max_level_store=,
10566 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
10567 systemd.journald.max_level_wall=.
10569 * "systemctl is-enabled --full" will now show by which symlinks a unit
10570 file is enabled in the unit dependency tree.
10572 * Support for VeraCrypt encrypted partitions has been added to the
10573 "cryptsetup" logic and /etc/crypttab.
10575 * systemd-detect-virt gained support for a new --private-users switch
10576 that checks whether the invoking processes are running inside a user
10577 namespace. Similar, a new special value "private-users" for the
10578 existing ConditionVirtualization= setting has been added, permitting
10579 skipping of specific units in user namespace environments.
10581 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
10582 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
10583 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
10584 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
10585 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
10586 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
10587 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
10588 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
10589 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
10590 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
10591 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
10592 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
10593 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
10594 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
10595 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
10596 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
10597 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
10598 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
10599 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
10600 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
10601 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
10602 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
10603 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
10604 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
10605 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
10606 Jędrzejewski-Szmek, Zeal Jagannatha
10608 — Santa Fe, 2016-11-03
10612 * In service units the various ExecXYZ= settings have been extended
10613 with an additional special character as first argument of the
10614 assigned value: if the character '+' is used the specified command
10615 line it will be run with full privileges, regardless of User=,
10616 Group=, CapabilityBoundingSet= and similar options. The effect is
10617 similar to the existing PermissionsStartOnly= option, but allows
10618 configuration of this concept for each executed command line
10621 * Services may now alter the service watchdog timeout at runtime by
10622 sending a WATCHDOG_USEC= message via sd_notify().
10624 * MemoryLimit= and related unit settings now optionally take percentage
10625 specifications. The percentage is taken relative to the amount of
10626 physical memory in the system (or in case of containers, the assigned
10627 amount of memory). This allows scaling service resources neatly with
10628 the amount of RAM available on the system. Similarly, systemd-logind's
10629 RuntimeDirectorySize= option now also optionally takes percentage
10632 * In similar fashion TasksMax= takes percentage values now, too. The
10633 value is taken relative to the configured maximum number of processes
10634 on the system. The per-service task maximum has been changed to 15%
10635 using this functionality. (Effectively this is an increase of 512 →
10636 4915 for service units, given the kernel's default pid_max setting.)
10638 * Calendar time specifications in .timer units now understand a ".."
10639 syntax for time ranges. Example: "4..7:10" may now be used for
10640 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
10643 * The InaccessableDirectories=, ReadOnlyDirectories= and
10644 ReadWriteDirectories= unit file settings have been renamed to
10645 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
10646 applied to all kinds of file nodes, and not just directories, with
10647 the exception of symlinks. Specifically these settings may now be
10648 used on block and character device nodes, UNIX sockets and FIFOS as
10649 well as regular files. The old names of these settings remain
10650 available for compatibility.
10652 * systemd will now log about all service processes it kills forcibly
10653 (using SIGKILL) because they remained after the clean shutdown phase
10654 of the service completed. This should help identifying services that
10655 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
10656 systemd-logind's configuration a similar log message is generated for
10657 processes killed at the end of each session due to this setting.
10659 * systemd will now set the $JOURNAL_STREAM environment variable for all
10660 services whose stdout/stderr are connected to the Journal (which
10661 effectively means by default: all services). The variable contains
10662 the device and inode number of the file descriptor used for
10663 stdout/stderr. This may be used by invoked programs to detect whether
10664 their stdout/stderr is connected to the Journal, in which case they
10665 can switch over to direct Journal communication, thus being able to
10666 pass extended, structured metadata along with their log messages. As
10667 one example, this is now used by glib's logging primitives.
10669 * When using systemd's default tmp.mount unit for /tmp, the mount point
10670 will now be established with the "nosuid" and "nodev" options. This
10671 avoids privilege escalation attacks that put traps and exploits into
10672 /tmp. However, this might cause problems if you e.g. put container
10673 images or overlays into /tmp; if you need this, override tmp.mount's
10674 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
10677 * systemd now supports the "memory" cgroup controller also on
10680 * The systemd-cgtop tool now optionally takes a control group path as
10681 command line argument. If specified, the control group list shown is
10682 limited to subgroups of that group.
10684 * The SystemCallFilter= unit file setting gained support for
10685 pre-defined, named system call filter sets. For example
10686 SystemCallFilter=@clock is now an effective way to make all clock
10687 changing-related system calls unavailable to a service. A number of
10688 similar pre-defined groups are defined. Writing system call filters
10689 for system services is simplified substantially with this new
10690 concept. Accordingly, all of systemd's own, long-running services now
10691 enable system call filtering based on this, by default.
10693 * A new service setting MemoryDenyWriteExecute= has been added, taking
10694 a boolean value. If turned on, a service may no longer create memory
10695 mappings that are writable and executable at the same time. This
10696 enhances security for services where this is enabled as it becomes
10697 harder to dynamically write and then execute memory in exploited
10698 service processes. This option has been enabled for all of systemd's
10699 own long-running services.
10701 * A new RestrictRealtime= service setting has been added, taking a
10702 boolean argument. If set the service's processes may no longer
10703 acquire realtime scheduling. This improves security as realtime
10704 scheduling may otherwise be used to easily freeze the system.
10706 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
10707 value. This may be used for requesting that the system manager inside
10708 of the container reports start-up completion to nspawn which then
10709 propagates this notification further to the service manager
10710 supervising nspawn itself. A related option NotifyReady= in .nspawn
10711 files has been added too. This functionality allows ordering of the
10712 start-up of multiple containers using the usual systemd ordering
10715 * machinectl gained a new command "stop" that is an alias for
10718 * systemd-resolved gained support for contacting DNS servers on
10719 link-local IPv6 addresses.
10721 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
10722 its caches. A method call for requesting the same operation has been
10723 added to the bus API too, and is made available via "systemd-resolve
10726 * systemd-resolve gained a new --status switch. If passed a brief
10727 summary of the used DNS configuration with per-interface information
10730 * resolved.conf gained a new Cache= boolean option, defaulting to
10731 on. If turned off local DNS caching is disabled. This comes with a
10732 performance penalty in particular when DNSSEC is enabled. Note that
10733 resolved disables its internal caching implicitly anyway, when the
10734 configured DNS server is on a host-local IP address such as ::1 or
10735 127.0.0.1, thus automatically avoiding double local caching.
10737 * systemd-resolved now listens on the local IP address 127.0.0.53:53
10738 for DNS requests. This improves compatibility with local programs
10739 that do not use the libc NSS or systemd-resolved's bus APIs for name
10740 resolution. This minimal DNS service is only available to local
10741 programs and does not implement the full DNS protocol, but enough to
10742 cover local DNS clients. A new, static resolv.conf file, listing just
10743 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
10744 now recommended to make /etc/resolv.conf a symlink to this file in
10745 order to route all DNS lookups to systemd-resolved, regardless if
10746 done via NSS, the bus API or raw DNS packets. Note that this local
10747 DNS service is not as fully featured as the libc NSS or
10748 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
10749 used to deliver link-local address information (as this implies
10750 sending a local interface index along), LLMNR/mDNS support via this
10751 interface is severely restricted. It is thus strongly recommended for
10752 all applications to use the libc NSS API or native systemd-resolved
10755 * systemd-networkd's bridge support learned a new setting
10756 VLANFiltering= for controlling VLAN filtering. Moreover a new section
10757 in .network files has been added for configuring VLAN bridging in
10758 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
10760 * systemd-networkd's IPv6 Router Advertisement code now makes use of
10761 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
10762 now be acquired without relying on DHCPv6. Two new options
10763 UseDomains= and UseDNS= have been added to configure this behaviour.
10765 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
10766 renamed IPv6AcceptRA=, without altering its behaviour. The old
10767 setting name remains available for compatibility reasons.
10769 * The systemd-networkd VTI/VTI6 tunneling support gained new options
10770 Key=, InputKey= and OutputKey=.
10772 * systemd-networkd gained support for VRF ("Virtual Routing Function")
10773 interface configuration.
10775 * "systemctl edit" may now be used to create new unit files by
10776 specifying the --force switch.
10778 * sd-event gained a new function sd_event_get_iteration() for
10779 requesting the current iteration counter of the event loop. It starts
10780 at zero and is increased by one with each event loop iteration.
10782 * A new rpm macro %systemd_ordering is provided by the macros.systemd
10783 file. It can be used in lieu of %systemd_requires in packages which
10784 don't use any systemd functionality and are intended to be installed
10785 in minimal containers without systemd present. This macro provides
10786 ordering dependencies to ensure that if the package is installed in
10787 the same rpm transaction as systemd, systemd will be installed before
10788 the scriptlets for the package are executed, allowing unit presets
10791 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
10792 been added to simplify packaging of generators.
10794 * The os-release file gained VERSION_CODENAME field for the
10795 distribution nickname (e.g. VERSION_CODENAME=woody).
10797 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
10798 can be set to disable parsing of metadata and the creation
10799 of persistent symlinks for that device.
10801 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
10802 to make them available to logged-in users has been reverted.
10804 * Much of the common code of the various systemd components is now
10805 built into an internal shared library libsystemd-shared-231.so
10806 (incorporating the systemd version number in the name, to be updated
10807 with future releases) that the components link to. This should
10808 decrease systemd footprint both in memory during runtime and on
10809 disk. Note that the shared library is not for public use, and is
10810 neither API nor ABI stable, but is likely to change with every new
10811 released update. Packagers need to make sure that binaries
10812 linking to libsystemd-shared.so are updated in step with the
10815 * Configuration for "mkosi" is now part of the systemd
10816 repository. mkosi is a tool to easily build legacy-free OS images,
10817 and is available on github: https://github.com/systemd/mkosi. If
10818 "mkosi" is invoked in the build tree a new raw OS image is generated
10819 incorporating the systemd sources currently being worked on and a
10820 clean, fresh distribution installation. The generated OS image may be
10821 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
10822 UEFI PC. This functionality is particularly useful to easily test
10823 local changes made to systemd in a pristine, defined environment. See
10824 doc/HACKING for details.
10826 * configure learned the --with-support-url= option to specify the
10827 distribution's bugtracker.
10829 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
10830 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
10831 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
10832 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
10833 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
10834 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
10835 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
10836 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
10837 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
10838 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
10839 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
10840 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
10841 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
10842 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
10843 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
10844 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
10845 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
10846 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
10847 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
10849 — Berlin, 2016-07-25
10853 * DNSSEC is now turned on by default in systemd-resolved (in
10854 "allow-downgrade" mode), but may be turned off during compile time by
10855 passing "--with-default-dnssec=no" to "configure" (and of course,
10856 during runtime with DNSSEC= in resolved.conf). We recommend
10857 downstreams to leave this on at least during development cycles and
10858 report any issues with the DNSSEC logic upstream. We are very
10859 interested in collecting feedback about the DNSSEC validator and its
10860 limitations in the wild. Note however, that DNSSEC support is
10861 probably nothing downstreams should turn on in stable distros just
10862 yet, as it might create incompatibilities with a few DNS servers and
10863 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
10864 automatically whenever we detect such incompatible setups, but there
10865 might be systems we do not cover yet. Hence: please help us testing
10866 the DNSSEC code, leave this on where you can, report back, but then
10867 again don't consider turning this on in your stable, LTS or
10868 production release just yet. (Note that you have to enable
10869 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
10870 and its DNSSEC mode for hostname resolution from local
10873 * systemd-resolve conveniently resolves DANE records with the --tlsa
10874 option and OPENPGPKEY records with the --openpgp option. It also
10875 supports dumping raw DNS record data via the new --raw= switch.
10877 * systemd-logind will now by default terminate user processes that are
10878 part of the user session scope unit (session-XX.scope) when the user
10879 logs out. This behavior is controlled by the KillUserProcesses=
10880 setting in logind.conf, and the previous default of "no" is now
10881 changed to "yes". This means that user sessions will be properly
10882 cleaned up after, but additional steps are necessary to allow
10883 intentionally long-running processes to survive logout.
10885 While the user is logged in at least once, user@.service is running,
10886 and any service that should survive the end of any individual login
10887 session can be started at a user service or scope using systemd-run.
10888 systemd-run(1) man page has been extended with an example which shows
10889 how to run screen in a scope unit underneath user@.service. The same
10890 command works for tmux.
10892 After the user logs out of all sessions, user@.service will be
10893 terminated too, by default, unless the user has "lingering" enabled.
10894 To effectively allow users to run long-term tasks even if they are
10895 logged out, lingering must be enabled for them. See loginctl(1) for
10896 details. The default polkit policy was modified to allow users to
10897 set lingering for themselves without authentication.
10899 Previous defaults can be restored at compile time by the
10900 --without-kill-user-processes option to "configure".
10902 * systemd-logind gained new configuration settings SessionsMax= and
10903 InhibitorsMax=, both with a default of 8192. It will not register new
10904 user sessions or inhibitors above this limit.
10906 * systemd-logind will now reload configuration on SIGHUP.
10908 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
10909 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
10910 enable. Also, support for the "io" cgroup controller in the unified
10911 hierarchy has been added, so that the "memory", "pids" and "io" are
10912 now the controllers that are supported on the unified hierarchy.
10914 WARNING: it is not possible to use previous systemd versions with
10915 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
10916 is necessary to also update systemd in the initramfs if using the
10917 unified hierarchy. An updated SELinux policy is also required.
10919 * LLDP support has been extended, and both passive (receive-only) and
10920 active (sender) modes are supported. Passive mode ("routers-only") is
10921 enabled by default in systemd-networkd. Active LLDP mode is enabled
10922 by default for containers on the internal network. The "networkctl
10923 lldp" command may be used to list information gathered. "networkctl
10924 status" will also show basic LLDP information on connected peers now.
10926 * The IAID and DUID unique identifier sent in DHCP requests may now be
10927 configured for the system and each .network file managed by
10928 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
10930 * systemd-networkd gained support for configuring proxy ARP support for
10931 each interface, via the ProxyArp= setting in .network files. It also
10932 gained support for configuring the multicast querier feature of
10933 bridge devices, via the new MulticastQuerier= setting in .netdev
10934 files. Similarly, snooping on the IGMP traffic can be controlled
10935 via the new setting MulticastSnooping=.
10937 A new setting PreferredLifetime= has been added for addresses
10938 configured in .network file to configure the lifetime intended for an
10941 The systemd-networkd DHCP server gained the option EmitRouter=, which
10942 defaults to yes, to configure whether the DHCP Option 3 (Router)
10945 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
10946 systemd-socket-activate and installed into /usr/bin. It is now fully
10949 * systemd-journald now uses separate threads to flush changes to disk
10950 when closing journal files, thus reducing impact of slow disk I/O on
10951 logging performance.
10953 * The sd-journal API gained two new calls
10954 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
10955 can be used to open journal files using file descriptors instead of
10956 file or directory paths. sd_journal_open_container() has been
10957 deprecated, sd_journal_open_directory_fd() should be used instead
10958 with the flag SD_JOURNAL_OS_ROOT.
10960 * journalctl learned a new output mode "-o short-unix" that outputs log
10961 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
10962 UTC). It also gained support for a new --no-hostname setting to
10963 suppress the hostname column in the family of "short" output modes.
10965 * systemd-ask-password now optionally skips printing of the password to
10966 stdout with --no-output which can be useful in scripts.
10968 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
10969 (devices tagged with ID_MAKER_TOOL) are now tagged with
10970 "uaccess" and are available to logged in users.
10972 * The DeviceAllow= unit setting now supports specifiers (with "%").
10974 * "systemctl show" gained a new --value switch, which allows print a
10975 only the contents of a specific unit property, without also printing
10976 the property's name. Similar support was added to "show*" verbs
10977 of loginctl and machinectl that output "key=value" lists.
10979 * A new unit type "generated" was added for files dynamically generated
10980 by generator tools. Similarly, a new unit type "transient" is used
10981 for unit files created using the runtime API. "systemctl enable" will
10982 refuse to operate on such files.
10984 * A new command "systemctl revert" has been added that may be used to
10985 revert to the vendor version of a unit file, in case local changes
10986 have been made by adding drop-ins or overriding the unit file.
10988 * "machinectl clean" gained a new verb to automatically remove all or
10989 just hidden container images.
10991 * systemd-tmpfiles gained support for a new line type "e" for emptying
10992 directories, if they exist, without creating them if they don't.
10994 * systemd-nspawn gained support for automatically patching the UID/GIDs
10995 of the owners and the ACLs of all files and directories in a
10996 container tree to match the UID/GID user namespacing range selected
10997 for the container invocation. This mode is enabled via the new
10998 --private-users-chown switch. It also gained support for
10999 automatically choosing a free, previously unused UID/GID range when
11000 starting a container, via the new --private-users=pick setting (which
11001 implies --private-users-chown). Together, these options for the first
11002 time make user namespacing for nspawn containers fully automatic and
11003 thus deployable. The systemd-nspawn@.service template unit file has
11004 been changed to use this functionality by default.
11006 * systemd-nspawn gained a new --network-zone= switch, that allows
11007 creating ad-hoc virtual Ethernet links between multiple containers,
11008 that only exist as long as at least one container referencing them is
11009 running. This allows easy connecting of multiple containers with a
11010 common link that implements an Ethernet broadcast domain. Each of
11011 these network "zones" may be named relatively freely by the user, and
11012 may be referenced by any number of containers, but each container may
11013 only reference one of these "zones". On the lower level, this is
11014 implemented by an automatically managed bridge network interface for
11015 each zone, that is created when the first container referencing its
11016 zone is created and removed when the last one referencing its zone
11019 * The default start timeout may now be configured on the kernel command
11020 line via systemd.default_timeout_start_sec=. It was already
11021 configurable via the DefaultTimeoutStartSec= option in
11022 /etc/systemd/system.conf.
11024 * Socket units gained a new TriggerLimitIntervalSec= and
11025 TriggerLimitBurst= setting to configure a limit on the activation
11026 rate of the socket unit.
11028 * The LimitNICE= setting now optionally takes normal UNIX nice values
11029 in addition to the raw integer limit value. If the specified
11030 parameter is prefixed with "+" or "-" and is in the range -20…19 the
11031 value is understood as UNIX nice value. If not prefixed like this it
11032 is understood as raw RLIMIT_NICE limit.
11034 * Note that the effect of the PrivateDevices= unit file setting changed
11035 slightly with this release: the per-device /dev file system will be
11036 mounted read-only from this version on, and will have "noexec"
11037 set. This (minor) change of behavior might cause some (exceptional)
11038 legacy software to break, when PrivateDevices=yes is set for its
11039 service. Please leave PrivateDevices= off if you run into problems
11042 * systemd-bootchart has been split out to a separate repository:
11043 https://github.com/systemd/systemd-bootchart
11045 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
11046 merged into the kernel in its current form.
11048 * The compatibility libraries libsystemd-daemon.so,
11049 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
11050 which have been deprecated since systemd-209 have been removed along
11051 with the corresponding pkg-config files. All symbols provided by
11052 those libraries are provided by libsystemd.so.
11054 * The Capabilities= unit file setting has been removed (it is ignored
11055 for backwards compatibility). AmbientCapabilities= and
11056 CapabilityBoundingSet= should be used instead.
11058 * A new special target has been added, initrd-root-device.target,
11059 which creates a synchronization point for dependencies of the root
11060 device in early userspace. Initramfs builders must ensure that this
11061 target is now included in early userspace.
11063 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
11064 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
11065 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
11066 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
11067 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
11068 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
11069 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
11070 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
11071 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
11072 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
11073 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
11074 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
11075 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
11076 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
11077 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
11078 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
11079 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
11080 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
11081 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
11082 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
11083 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
11084 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
11085 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
11086 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
11087 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
11090 — Fairfax, 2016-05-21
11094 * The systemd-resolved DNS resolver service has gained a substantial
11095 set of new features, most prominently it may now act as a DNSSEC
11096 validating stub resolver. DNSSEC mode is currently turned off by
11097 default, but is expected to be turned on by default in one of the
11098 next releases. For now, we invite everybody to test the DNSSEC logic
11099 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
11100 service also gained a full set of D-Bus interfaces, including calls
11101 to configure DNS and DNSSEC settings per link (for use by external
11102 network management software). systemd-resolved and systemd-networkd
11103 now distinguish between "search" and "routing" domains. The former
11104 are used to qualify single-label names, the latter are used purely
11105 for routing lookups within certain domains to specific links.
11106 resolved now also synthesizes RRs for all entries from /etc/hosts.
11108 * The systemd-resolve tool (which is a client utility for
11109 systemd-resolved) has been improved considerably and is now fully
11110 supported and documented. Hence it has moved from /usr/lib/systemd to
11113 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
11116 * The coredump collection logic has been reworked: when a coredump is
11117 collected it is now written to disk, compressed and processed
11118 (including stacktrace extraction) from a new instantiated service
11119 systemd-coredump@.service, instead of directly from the
11120 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
11121 processing large coredumps can take up a substantial amount of
11122 resources and time, and this previously happened entirely outside of
11123 systemd's service supervision. With the new logic the core_pattern
11124 hook only does minimal metadata collection before passing off control
11125 to the new instantiated service, which is configured with a time
11126 limit, a nice level and other settings to minimize negative impact on
11127 the rest of the system. Also note that the new logic will honour the
11128 RLIMIT_CORE setting of the crashed process, which now allows users
11129 and processes to turn off coredumping for their processes by setting
11132 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
11133 and all forked processes by default. Previously, PID 1 would leave
11134 the setting at "0" for all processes, as set by the kernel. Note that
11135 the resource limit traditionally has no effect on the generated
11136 coredumps on the system if the /proc/sys/kernel/core_pattern hook
11137 logic is used. Since the limit is now honoured (see above) its
11138 default has been changed so that the coredumping logic is enabled by
11139 default for all processes, while allowing specific opt-out.
11141 * When the stacktrace is extracted from processes of system users, this
11142 is now done as "systemd-coredump" user, in order to sandbox this
11143 potentially security sensitive parsing operation. (Note that when
11144 processing coredumps of normal users this is done under the user ID
11145 of process that crashed, as before.) Packagers should take notice
11146 that it is now necessary to create the "systemd-coredump" system user
11147 and group at package installation time.
11149 * The systemd-activate socket activation testing tool gained support
11150 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
11151 and --seqpacket switches. It also has been extended to support both
11152 new-style and inetd-style file descriptor passing. Use the new
11153 --inetd switch to request inetd-style file descriptor passing.
11155 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
11156 variable, which takes a boolean value. If set to false, ANSI color
11157 output is disabled in the tools even when run on a terminal that
11160 * The VXLAN support in networkd now supports two new settings
11161 DestinationPort= and PortRange=.
11163 * A new systemd.machine_id= kernel command line switch has been added,
11164 that may be used to set the machine ID in /etc/machine-id if it is
11165 not initialized yet. This command line option has no effect if the
11166 file is already initialized.
11168 * systemd-nspawn gained a new --as-pid2 switch that invokes any
11169 specified command line as PID 2 rather than PID 1 in the
11170 container. In this mode PID 1 is a minimal stub init process that
11171 implements the special POSIX and Linux semantics of PID 1 regarding
11172 signal and child process management. Note that this stub init process
11173 is implemented in nspawn itself and requires no support from the
11174 container image. This new logic is useful to support running
11175 arbitrary commands in the container, as normal processes are
11176 generally not prepared to run as PID 1.
11178 * systemd-nspawn gained a new --chdir= switch for setting the current
11179 working directory for the process started in the container.
11181 * "journalctl /dev/sda" will now output all kernel log messages for
11182 specified device from the current boot, in addition to all devices
11183 that are parents of it. This should make log output about devices
11184 pretty useful, as long as kernel drivers attach enough metadata to
11185 the log messages. (The usual SATA drivers do.)
11187 * The sd-journal API gained two new calls
11188 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
11189 that report whether log data from /run or /var has been found.
11191 * journalctl gained a new switch "--fields" that prints all journal
11192 record field names currently in use in the journal. This is backed
11193 by two new sd-journal API calls sd_journal_enumerate_fields() and
11194 sd_journal_restart_fields().
11196 * Most configurable timeouts in systemd now expect an argument of
11197 "infinity" to turn them off, instead of "0" as before. The semantics
11198 from now on is that a timeout of "0" means "now", and "infinity"
11199 means "never". To maintain backwards compatibility, "0" continues to
11200 turn off previously existing timeout settings.
11202 * "systemctl reload-or-try-restart" has been renamed to "systemctl
11203 try-reload-or-restart" to clarify what it actually does: the "try"
11204 logic applies to both reloading and restarting, not just restarting.
11205 The old name continues to be accepted for compatibility.
11207 * On boot-up, when PID 1 detects that the system clock is behind the
11208 release date of the systemd version in use, the clock is now set
11209 to the latter. Previously, this was already done in timesyncd, in order
11210 to avoid running with clocks set to the various clock epochs such as
11211 1902, 1938 or 1970. With this change the logic is now done in PID 1
11212 in addition to timesyncd during early boot-up, so that it is enforced
11213 before the first process is spawned by systemd. Note that the logic
11214 in timesyncd remains, as it is more comprehensive and ensures
11215 clock monotonicity by maintaining a persistent timestamp file in
11216 /var. Since /var is generally not available in earliest boot or the
11217 initrd, this part of the logic remains in timesyncd, and is not done
11220 * Support for tweaking details in net_cls.class_id through the
11221 NetClass= configuration directive has been removed, as the kernel
11222 people have decided to deprecate that controller in cgroup v2.
11223 Userspace tools such as nftables are moving over to setting rules
11224 that are specific to the full cgroup path of a task, which obsoletes
11225 these controllers anyway. The NetClass= directive is kept around for
11226 legacy compatibility reasons. For a more in-depth description of the
11227 kernel change, please refer to the respective upstream commit:
11229 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
11231 * A new service setting RuntimeMaxSec= has been added that may be used
11232 to specify a maximum runtime for a service. If the timeout is hit, the
11233 service is terminated and put into a failure state.
11235 * A new service setting AmbientCapabilities= has been added. It allows
11236 configuration of additional Linux process capabilities that are
11237 passed to the activated processes. This is only available on very
11240 * The process resource limit settings in service units may now be used
11241 to configure hard and soft limits individually.
11243 * The various libsystemd APIs such as sd-bus or sd-event now publicly
11244 expose support for gcc's __attribute__((cleanup())) C extension.
11245 Specifically, for many object destructor functions alternative
11246 versions have been added that have names suffixed with "p" and take a
11247 pointer to a pointer to the object to destroy, instead of just a
11248 pointer to the object itself. This is useful because these destructor
11249 functions may be used directly as parameters to the cleanup
11250 construct. Internally, systemd has been a heavy user of this GCC
11251 extension for a long time, and with this change similar support is
11252 now available to consumers of the library outside of systemd. Note
11253 that by using this extension in your sources compatibility with old
11254 and strictly ANSI compatible C compilers is lost. However, all gcc or
11255 LLVM versions of recent years support this extension.
11257 * Timer units gained support for a new setting RandomizedDelaySec= that
11258 allows configuring some additional randomized delay to the configured
11259 time. This is useful to spread out timer events to avoid load peaks in
11260 clusters or larger setups.
11262 * Calendar time specifications now support sub-second accuracy.
11264 * Socket units now support listening on SCTP and UDP-lite protocol
11267 * The sd-event API now comes with a full set of man pages.
11269 * Older versions of systemd contained experimental support for
11270 compressing journal files and coredumps with the LZ4 compressor that
11271 was not compatible with the lz4 binary (due to API limitations of the
11272 lz4 library). This support has been removed; only support for files
11273 compatible with the lz4 binary remains. This LZ4 logic is now
11274 officially supported and no longer considered experimental.
11276 * The dkr image import logic has been removed again from importd. dkr's
11277 micro-services focus doesn't fit into the machine image focus of
11278 importd, and quickly got out of date with the upstream dkr API.
11280 * Creation of the /run/lock/lockdev/ directory was dropped from
11281 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
11282 been available for many years. If you still need this, you need to
11283 create your own tmpfiles.d config file with:
11285 d /run/lock/lockdev 0775 root lock -
11287 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
11288 and RebootArgument= have been moved from the [Service] section of
11289 unit files to [Unit], and they are now supported on all unit types,
11290 not just service units. Of course, systemd will continue to
11291 understand these settings also at the old location, in order to
11292 maintain compatibility.
11294 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
11295 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
11296 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
11297 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
11298 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
11299 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
11300 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
11301 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
11302 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
11303 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
11304 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
11305 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
11306 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
11307 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
11308 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
11309 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
11310 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
11311 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
11312 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
11314 — Berlin, 2016-02-11
11318 * A number of properties previously only settable in unit
11319 files are now also available as properties to set when
11320 creating transient units programmatically via the bus, as it
11321 is exposed with systemd-run's --property=
11322 setting. Specifically, these are: SyslogIdentifier=,
11323 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
11324 EnvironmentFile=, ReadWriteDirectories=,
11325 ReadOnlyDirectories=, InaccessibleDirectories=,
11326 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
11328 * When creating transient services via the bus API it is now
11329 possible to pass in a set of file descriptors to use as
11330 STDIN/STDOUT/STDERR for the invoked process.
11332 * Slice units may now be created transiently via the bus APIs,
11333 similar to the way service and scope units may already be
11334 created transiently.
11336 * Wherever systemd expects a calendar timestamp specification
11337 (like in journalctl's --since= and --until= switches) UTC
11338 timestamps are now supported. Timestamps suffixed with "UTC"
11339 are now considered to be in Universal Time Coordinated
11340 instead of the local timezone. Also, timestamps may now
11341 optionally be specified with sub-second accuracy. Both of
11342 these additions also apply to recurring calendar event
11343 specification, such as OnCalendar= in timer units.
11345 * journalctl gained a new "--sync" switch that asks the
11346 journal daemon to write all so far unwritten log messages to
11347 disk and sync the files, before returning.
11349 * systemd-tmpfiles learned two new line types "q" and "Q" that
11350 operate like "v", but also set up a basic btrfs quota
11351 hierarchy when used on a btrfs file system with quota
11354 * tmpfiles' "v", "q" and "Q" will now create a plain directory
11355 instead of a subvolume (even on a btrfs file system) if the
11356 root directory is a plain directory, and not a
11357 subvolume. This should simplify things with certain chroot()
11358 environments which are not aware of the concept of btrfs
11361 * systemd-detect-virt gained a new --chroot switch to detect
11362 whether execution takes place in a chroot() environment.
11364 * CPUAffinity= now takes CPU index ranges in addition to
11365 individual indexes.
11367 * The various memory-related resource limit settings (such as
11368 LimitAS=) now understand the usual K, M, G, … suffixes to
11369 the base of 1024 (IEC). Similar, the time-related resource
11370 limit settings understand the usual min, h, day, … suffixes
11373 * There's a new system.conf setting DefaultTasksMax= to
11374 control the default TasksMax= setting for services and
11375 scopes running on the system. (TasksMax= is the primary
11376 setting that exposes the "pids" cgroup controller on systemd
11377 and was introduced in the previous systemd release.) The
11378 setting now defaults to 512, which means services that are
11379 not explicitly configured otherwise will only be able to
11380 create 512 processes or threads at maximum, from this
11381 version on. Note that this means that thread- or
11382 process-heavy services might need to be reconfigured to set
11383 TasksMax= to a higher value. It is sufficient to set
11384 TasksMax= in these specific unit files to a higher value, or
11385 even "infinity". Similar, there's now a logind.conf setting
11386 UserTasksMax= that defaults to 4096 and limits the total
11387 number of processes or tasks each user may own
11388 concurrently. nspawn containers also have the TasksMax=
11389 value set by default now, to 8192. Note that all of this
11390 only has an effect if the "pids" cgroup controller is
11391 enabled in the kernel. The general benefit of these changes
11392 should be a more robust and safer system, that provides a
11393 certain amount of per-service fork() bomb protection.
11395 * systemd-nspawn gained the new --network-veth-extra= switch
11396 to define additional and arbitrarily-named virtual Ethernet
11397 links between the host and the container.
11399 * A new service execution setting PassEnvironment= has been
11400 added that allows importing select environment variables
11401 from PID1's environment block into the environment block of
11404 * Timer units gained support for a new RemainAfterElapse=
11405 setting which takes a boolean argument. It defaults to on,
11406 exposing behaviour unchanged to previous releases. If set to
11407 off, timer units are unloaded after they elapsed if they
11408 cannot elapse again. This is particularly useful for
11409 transient timer units, which shall not stay around longer
11410 than until they first elapse.
11412 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
11413 default now (the kernel default is 16). This is beneficial
11414 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
11415 allows substantially larger numbers of queued
11416 datagrams. This should increase the capability of systemd to
11417 parallelize boot-up, as logging and sd_notify() are unlikely
11418 to stall execution anymore. If you need to change the value
11419 from the new defaults, use the usual sysctl.d/ snippets.
11421 * The compression framing format used by the journal or
11422 coredump processing has changed to be in line with what the
11423 official LZ4 tools generate. LZ4 compression support in
11424 systemd was considered unsupported previously, as the format
11425 was not compatible with the normal tools. With this release
11426 this has changed now, and it is hence safe for downstream
11427 distributions to turn it on. While not compressing as well
11428 as the XZ, LZ4 is substantially faster, which makes
11429 it a good default choice for the compression logic in the
11430 journal and in coredump handling.
11432 * Any reference to /etc/mtab has been dropped from
11433 systemd. The file has been obsolete since a while, but
11434 systemd refused to work on systems where it was incorrectly
11435 set up (it should be a symlink or non-existent). Please make
11436 sure to update to util-linux 2.27.1 or newer in conjunction
11437 with this systemd release, which also drops any reference to
11438 /etc/mtab. If you maintain a distribution make sure that no
11439 software you package still references it, as this is a
11440 likely source of bugs. There's also a glibc bug pending,
11441 asking for removal of any reference to this obsolete file:
11443 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
11445 Note that only util-linux versions built with
11446 --enable-libmount-force-mountinfo are supported.
11448 * Support for the ".snapshot" unit type has been removed. This
11449 feature turned out to be little useful and little used, and
11450 has now been removed from the core and from systemctl.
11452 * The dependency types RequiresOverridable= and
11453 RequisiteOverridable= have been removed from systemd. They
11454 have been used only very sparingly to our knowledge and
11455 other options that provide a similar effect (such as
11456 systemctl --mode=ignore-dependencies) are much more useful
11457 and commonly used. Moreover, they were only half-way
11458 implemented as the option to control behaviour regarding
11459 these dependencies was never added to systemctl. By removing
11460 these dependency types the execution engine becomes a bit
11461 simpler. Unit files that use these dependencies should be
11462 changed to use the non-Overridable dependency types
11463 instead. In fact, when parsing unit files with these
11464 options, that's what systemd will automatically convert them
11465 too, but it will also warn, asking users to fix the unit
11466 files accordingly. Removal of these dependency types should
11467 only affect a negligible number of unit files in the wild.
11469 * Behaviour of networkd's IPForward= option changed
11470 (again). It will no longer maintain a per-interface setting,
11471 but propagate one way from interfaces where this is enabled
11472 to the global kernel setting. The global setting will be
11473 enabled when requested by a network that is set up, but
11474 never be disabled again. This change was made to make sure
11475 IPv4 and IPv6 behaviour regarding packet forwarding is
11476 similar (as the Linux IPv6 stack does not support
11477 per-interface control of this setting) and to minimize
11480 * In unit files the behaviour of %u, %U, %h, %s has
11481 changed. These specifiers will now unconditionally resolve
11482 to the various user database fields of the user that the
11483 systemd instance is running as, instead of the user
11484 configured in the specific unit via User=. Note that this
11485 effectively doesn't change much, as resolving of these
11486 specifiers was already turned off in the --system instance
11487 of systemd, as we cannot do NSS lookups from PID 1. In the
11488 --user instance of systemd these specifiers where correctly
11489 resolved, but hardly made any sense, since the user instance
11490 lacks privileges to do user switches anyway, and User= is
11491 hence useless. Moreover, even in the --user instance of
11492 systemd behaviour was awkward as it would only take settings
11493 from User= assignment placed before the specifier into
11494 account. In order to unify and simplify the logic around
11495 this the specifiers will now always resolve to the
11496 credentials of the user invoking the manager (which in case
11497 of PID 1 is the root user).
11499 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
11500 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
11501 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
11502 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
11503 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
11504 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
11505 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
11506 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
11507 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
11508 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
11509 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
11510 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
11511 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
11512 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
11515 — Berlin, 2015-11-18
11519 * systemd now depends on util-linux v2.27. More specifically,
11520 the newly added mount monitor feature in libmount now
11521 replaces systemd's former own implementation.
11523 * libmount mandates /etc/mtab not to be regular file, and
11524 systemd now enforces this condition at early boot.
11525 /etc/mtab has been deprecated and warned about for a very
11526 long time, so systems running systemd should already have
11527 stopped having this file around as anything else than a
11528 symlink to /proc/self/mounts.
11530 * Support for the "pids" cgroup controller has been added. It
11531 allows accounting the number of tasks in a cgroup and
11532 enforcing limits on it. This adds two new setting
11533 TasksAccounting= and TasksMax= to each unit, as well as a
11534 global option DefaultTasksAccounting=.
11536 * Support for the "net_cls" cgroup controller has been added.
11537 It allows assigning a net class ID to each task in the
11538 cgroup, which can then be used in firewall rules and traffic
11539 shaping configurations. Note that the kernel netfilter net
11540 class code does not currently work reliably for ingress
11541 packets on unestablished sockets.
11543 This adds a new config directive called NetClass= to CGroup
11544 enabled units. Allowed values are positive numbers for fixed
11545 assignments and "auto" for picking a free value
11548 * 'systemctl is-system-running' now returns 'offline' if the
11549 system is not booted with systemd. This command can now be
11550 used as a substitute for 'systemd-notify --booted'.
11552 * Watchdog timeouts have been increased to 3 minutes for all
11553 in-tree service files. Apparently, disk IO issues are more
11554 frequent than we hoped, and user reported >1 minute waiting
11557 * 'machine-id-commit' functionality has been merged into
11558 'machine-id-setup --commit'. The separate binary has been
11561 * The WorkingDirectory= directive in unit files may now be set
11562 to the special value '~'. In this case, the working
11563 directory is set to the home directory of the user
11564 configured in User=.
11566 * "machinectl shell" will now open the shell in the home
11567 directory of the selected user by default.
11569 * The CrashChVT= configuration file setting is renamed to
11570 CrashChangeVT=, following our usual logic of not
11571 abbreviating unnecessarily. The old directive is still
11572 supported for compat reasons. Also, this directive now takes
11573 an integer value between 1 and 63, or a boolean value. The
11574 formerly supported '-1' value for disabling stays around for
11577 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
11578 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
11579 RootDirectory= properties can now be set for transient
11582 * The systemd-analyze tool gained a new "set-log-target" verb
11583 to change the logging target the system manager logs to
11584 dynamically during runtime. This is similar to how
11585 "systemd-analyze set-log-level" already changes the log
11588 * In nspawn /sys is now mounted as tmpfs, with only a selected
11589 set of subdirectories mounted in from the real sysfs. This
11590 enhances security slightly, and is useful for ensuring user
11591 namespaces work correctly.
11593 * Support for USB FunctionFS activation has been added. This
11594 allows implementation of USB gadget services that are
11595 activated as soon as they are requested, so that they don't
11596 have to run continuously, similar to classic socket
11599 * The "systemctl exit" command now optionally takes an
11600 additional parameter that sets the exit code to return from
11601 the systemd manager when exiting. This is only relevant when
11602 running the systemd user instance, or when running the
11603 system instance in a container.
11605 * sd-bus gained the new API calls sd_bus_path_encode_many()
11606 and sd_bus_path_decode_many() that allow easy encoding and
11607 decoding of multiple identifier strings inside a D-Bus
11608 object path. Another new call sd_bus_default_flush_close()
11609 has been added to flush and close per-thread default
11612 * systemd-cgtop gained support for a -M/--machine= switch to
11613 show the control groups within a certain container only.
11615 * "systemctl kill" gained support for an optional --fail
11616 switch. If specified the requested operation will fail of no
11617 processes have been killed, because the unit had no
11618 processes attached, or similar.
11620 * A new systemd.crash_reboot=1 kernel command line option has
11621 been added that triggers a reboot after crashing. This can
11622 also be set through CrashReboot= in systemd.conf.
11624 * The RuntimeDirectory= setting now understands unit
11625 specifiers like %i or %f.
11627 * A new (still internal) library API sd-ipv4acd has been added,
11628 that implements address conflict detection for IPv4. It's
11629 based on code from sd-ipv4ll, and will be useful for
11630 detecting DHCP address conflicts.
11632 * File descriptors passed during socket activation may now be
11633 named. A new API sd_listen_fds_with_names() is added to
11634 access the names. The default names may be overridden,
11635 either in the .socket file using the FileDescriptorName=
11636 parameter, or by passing FDNAME= when storing the file
11637 descriptors using sd_notify().
11639 * systemd-networkd gained support for:
11641 - Setting the IPv6 Router Advertisement settings via
11642 IPv6AcceptRouterAdvertisements= in .network files.
11644 - Configuring the HelloTimeSec=, MaxAgeSec= and
11645 ForwardDelaySec= bridge parameters in .netdev files.
11647 - Configuring PreferredSource= for static routes in
11650 * The "ask-password" framework used to query for LUKS harddisk
11651 passwords or SSL passwords during boot gained support for
11652 caching passwords in the kernel keyring, if it is
11653 available. This makes sure that the user only has to type in
11654 a passphrase once if there are multiple objects to unlock
11655 with the same one. Previously, such password caching was
11656 available only when Plymouth was used; this moves the
11657 caching logic into the systemd codebase itself. The
11658 "systemd-ask-password" utility gained a new --keyname=
11659 switch to control which kernel keyring key to use for
11660 caching a password in. This functionality is also useful for
11661 enabling display managers such as gdm to automatically
11662 unlock the user's GNOME keyring if its passphrase, the
11663 user's password and the harddisk password are the same, if
11664 gdm-autologin is used.
11666 * When downloading tar or raw images using "machinectl
11667 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
11668 file is now also downloaded, if it is available and stored
11669 next to the image file.
11671 * Units of type ".socket" gained a new boolean setting
11672 Writable= which is only useful in conjunction with
11673 ListenSpecial=. If true, enables opening the specified
11674 special file in O_RDWR mode rather than O_RDONLY mode.
11676 * systemd-rfkill has been reworked to become a singleton
11677 service that is activated through /dev/rfkill on each rfkill
11678 state change and saves the settings to disk. This way,
11679 systemd-rfkill is now compatible with devices that exist
11680 only intermittendly, and even restores state if the previous
11681 system shutdown was abrupt rather than clean.
11683 * The journal daemon gained support for vacuuming old journal
11684 files controlled by the number of files that shall remain,
11685 in addition to the already existing control by size and by
11686 date. This is useful as journal interleaving performance
11687 degrades with too many separate journal files, and allows
11688 putting an effective limit on them. The new setting defaults
11689 to 100, but this may be changed by setting SystemMaxFiles=
11690 and RuntimeMaxFiles= in journald.conf. Also, the
11691 "journalctl" tool gained the new --vacuum-files= switch to
11692 manually vacuum journal files to leave only the specified
11693 number of files in place.
11695 * udev will now create /dev/disk/by-path links for ATA devices
11696 on kernels where that is supported.
11698 * Galician, Serbian, Turkish and Korean translations were added.
11700 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
11701 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
11702 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
11703 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
11704 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
11705 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
11706 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
11707 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
11708 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
11709 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
11710 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
11711 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
11712 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
11713 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
11714 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
11715 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
11716 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
11717 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
11719 — Berlin, 2015-10-07
11723 * The DHCP implementation of systemd-networkd gained a set of
11726 - The DHCP server now supports emitting DNS and NTP
11727 information. It may be enabled and configured via
11728 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
11729 and NTP information is enabled, but no servers are
11730 configured, the corresponding uplink information (if there
11731 is any) is propagated.
11733 - Server and client now support transmission and reception
11734 of timezone information. It can be configured via the
11735 newly introduced network options UseTimezone=,
11736 EmitTimezone=, and Timezone=. Transmission of timezone
11737 information is enabled between host and containers by
11738 default now: the container will change its local timezone
11739 to what the host has set.
11741 - Lease timeouts can now be configured via
11742 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
11744 - The DHCP server improved on the stability of
11745 leases. Clients are more likely to get the same lease
11746 information back, even if the server loses state.
11748 - The DHCP server supports two new configuration options to
11749 control the lease address pool metrics, PoolOffset= and
11752 * The encapsulation limit of tunnels in systemd-networkd may
11753 now be configured via 'EncapsulationLimit='. It allows
11754 modifying the maximum additional levels of encapsulation
11755 that are permitted to be prepended to a packet.
11757 * systemd now supports the concept of user buses replacing
11758 session buses, if used with dbus-1.10 (and enabled via dbus
11759 --enable-user-session). It previously only supported this on
11760 kdbus-enabled systems, and this release expands this to
11761 'dbus-daemon' systems.
11763 * systemd-networkd now supports predictable interface names
11764 for virtio devices.
11766 * systemd now optionally supports the new Linux kernel
11767 "unified" control group hierarchy. If enabled via the kernel
11768 command-line option 'systemd.unified_cgroup_hierarchy=1',
11769 systemd will try to mount the unified cgroup hierarchy
11770 directly on /sys/fs/cgroup. If not enabled, or not
11771 available, systemd will fall back to the legacy cgroup
11772 hierarchy setup, as before. Host system and containers can
11773 mix and match legacy and unified hierarchies as they
11774 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
11775 environment variable to individually select the hierarchy to
11776 use for executed containers. By default, nspawn will use the
11777 unified hierarchy for the containers if the host uses the
11778 unified hierarchy, and the legacy hierarchy otherwise.
11779 Please note that at this point the unified hierarchy is an
11780 experimental kernel feature and is likely to change in one
11781 of the next kernel releases. Therefore, it should not be
11782 enabled by default in downstream distributions yet. The
11783 minimum required kernel version for the unified hierarchy to
11784 work is 4.2. Note that when the unified hierarchy is used
11785 for the first time delegated access to controllers is
11786 safe. Because of this systemd-nspawn containers will get
11787 access to controllers now, as will systemd user
11788 sessions. This means containers and user sessions may now
11789 manage their own resources, partitioning up what the system
11792 * A new special scope unit "init.scope" has been introduced
11793 that encapsulates PID 1 of the system. It may be used to
11794 determine resource usage and enforce resource limits on PID
11795 1 itself. PID 1 hence moved out of the root of the control
11798 * The cgtop tool gained support for filtering out kernel
11799 threads when counting tasks in a control group. Also, the
11800 count of processes is now recursively summed up by
11801 default. Two options -k and --recursive= have been added to
11802 revert to old behaviour. The tool has also been updated to
11803 work correctly in containers now.
11805 * systemd-nspawn's --bind= and --bind-ro= options have been
11806 extended to allow creation of non-recursive bind mounts.
11808 * libsystemd gained two new calls sd_pid_get_cgroup() and
11809 sd_peer_get_cgroup() which return the control group path of
11810 a process or peer of a connected AF_UNIX socket. This
11811 function call is particularly useful when implementing
11812 delegated subtrees support in the control group hierarchy.
11814 * The "sd-event" event loop API of libsystemd now supports
11815 correct dequeuing of real-time signals, without losing
11818 * When systemd requests a polkit decision when managing units it
11819 will now add additional fields to the request, including unit
11820 name and desired operation. This enables more powerful polkit
11821 policies, that make decisions depending on these parameters.
11823 * nspawn learnt support for .nspawn settings files, that may
11824 accompany the image files or directories of containers, and
11825 may contain additional settings for the container. This is
11826 an alternative to configuring container parameters via the
11827 nspawn command line.
11829 Contributions from: Cristian Rodríguez, Daniel Mack, David
11830 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
11831 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
11832 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
11833 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
11834 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
11835 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
11836 Andersen, Tom Gundersen, Torstein Husebø
11838 — Berlin, 2015-09-08
11842 * machinectl gained a new verb 'shell' which opens a fresh
11843 shell on the target container or the host. It is similar to
11844 the existing 'login' command of machinectl, but spawns the
11845 shell directly without prompting for username or
11846 password. The pseudo machine '.host' now refers to the local
11847 host and is used by default. Hence, 'machinectl shell' can
11848 be used as replacement for 'su -' which spawns a session as
11849 a fresh systemd unit in a way that is fully isolated from
11850 the originating session.
11852 * systemd-networkd learned to cope with private-zone DHCP
11853 options and allows other programs to query the values.
11855 * SELinux access control when enabling/disabling units is no
11856 longer enforced with this release. The previous implementation
11857 was incorrect, and a new corrected implementation is not yet
11858 available. As unit file operations are still protected via
11859 polkit and D-Bus policy this is not a security problem. Yet,
11860 distributions which care about optimal SELinux support should
11861 probably not stabilize on this release.
11863 * sd-bus gained support for matches of type "arg0has=", that
11864 test for membership of strings in string arrays sent in bus
11867 * systemd-resolved now dumps the contents of its DNS and LLMNR
11868 caches to the logs on reception of the SIGUSR1 signal. This
11869 is useful to debug DNS behaviour.
11871 * The coredumpctl tool gained a new --directory= option to
11872 operate on journal files in a specific directory.
11874 * "systemctl reboot" and related commands gained a new
11875 "--message=" option which may be used to set a free-text
11876 wall message when shutting down or rebooting the
11877 system. This message is also logged, which is useful for
11878 figuring out the reason for a reboot or shutdown a
11881 * The "systemd-resolve-host" tool's -i switch now takes
11882 network interface numbers as alternative to interface names.
11884 * A new unit file setting for services has been introduced:
11885 UtmpMode= allows configuration of how precisely systemd
11886 handles utmp and wtmp entries for the service if this is
11887 enabled. This allows writing services that appear similar to
11888 user sessions in the output of the "w", "who", "last" and
11891 * systemd-resolved will now locally synthesize DNS resource
11892 records for the "localhost" and "gateway" domains as well as
11893 the local hostname. This should ensure that clients querying
11894 RRs via resolved will get similar results as those going via
11895 NSS, if nss-myhostname is enabled.
11897 Contributions from: Alastair Hughes, Alex Crawford, Daniel
11898 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
11899 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
11900 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
11901 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
11902 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
11903 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
11904 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
11905 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
11906 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
11907 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
11908 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11910 — Berlin, 2015-08-27
11914 * The systemd-efi-boot-generator functionality was merged into
11915 systemd-gpt-auto-generator.
11917 * systemd-networkd now supports Group Policy for vxlan
11918 devices. It can be enabled via the new boolean configuration
11919 option called 'GroupPolicyExtension='.
11921 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
11922 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
11923 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
11925 — Berlin, 2015-07-31
11929 * The python-systemd code has been removed from the systemd repository.
11930 A new repository has been created which accommodates the code from
11931 now on, and we kindly ask distributions to create a separate package
11932 for this: https://github.com/systemd/python-systemd
11934 * The systemd daemon will now reload its main configuration
11935 (/etc/systemd/system.conf) on daemon-reload.
11937 * sd-dhcp now exposes vendor specific extensions via
11938 sd_dhcp_lease_get_vendor_specific().
11940 * systemd-networkd gained a number of new configuration options.
11942 - A new boolean configuration option for TAP devices called
11943 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
11944 device, thus allowing to send and receive GSO packets.
11946 - A new tunnel configuration option called 'CopyDSCP='.
11947 If enabled, the DSCP field of ip6 tunnels is copied into the
11948 decapsulated packet.
11950 - A set of boolean bridge configuration options were added.
11951 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
11952 and 'UnicastFlood=' are now parsed by networkd and applied to the
11953 respective bridge link device via the respective IFLA_BRPORT_*
11956 - A new string configuration option to override the hostname sent
11957 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
11958 is true, networkd will use the configured hostname instead of the
11959 system hostname when sending DHCP requests.
11961 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
11962 networkd will configure the IPv6 flow-label of the tunnel device
11963 according to RFC2460.
11965 - The 'macvtap' virtual network devices are now supported, similar to
11966 the already supported 'macvlan' devices.
11968 * systemd-resolved now implements RFC5452 to improve resilience against
11969 cache poisoning. Additionally, source port randomization is enabled
11970 by default to further protect against DNS spoofing attacks.
11972 * nss-mymachines now supports translating UIDs and GIDs of running
11973 containers with user-namespaces enabled. If a container 'foo'
11974 translates a host uid 'UID' to the container uid 'TUID', then
11975 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
11976 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
11977 mapped as 'vg-foo-TGID'.
11979 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
11980 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
11981 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
11982 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
11983 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
11984 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
11985 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
11986 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
11987 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
11988 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
11990 — Berlin, 2015-07-29
11994 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
11995 There are no known issues with current sysfs, and udev does not need
11996 or should be used to work around such bugs.
11998 * udev does no longer enable USB HID power management. Several reports
11999 indicate, that some devices cannot handle that setting.
12001 * The udev accelerometer helper was removed. The functionality
12002 is now fully included in iio-sensor-proxy. But this means,
12003 older iio-sensor-proxy versions will no longer provide
12004 accelerometer/orientation data with this systemd version.
12005 Please upgrade iio-sensor-proxy to version 1.0.
12007 * networkd gained a new configuration option IPv6PrivacyExtensions=
12008 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
12009 for Stateless Address") on selected networks.
12011 * For the sake of fewer build-time dependencies and less code in the
12012 main repository, the python bindings are about to be removed in the
12013 next release. A new repository has been created which accommodates
12014 the code from now on, and we kindly ask distributions to create a
12015 separate package for this. The removal will take place in v223.
12017 https://github.com/systemd/python-systemd
12019 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
12020 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
12021 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
12022 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
12023 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
12024 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
12025 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
12026 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
12027 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12028 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
12030 — Berlin, 2015-07-07
12034 * The sd-bus.h and sd-event.h APIs have now been declared
12035 stable and have been added to the official interface of
12036 libsystemd.so. sd-bus implements an alternative D-Bus client
12037 library, that is relatively easy to use, very efficient and
12038 supports both classic D-Bus as well as kdbus as transport
12039 backend. sd-event is a generic event loop abstraction that
12040 is built around Linux epoll, but adds features such as event
12041 prioritization or efficient timer handling. Both APIs are good
12042 choices for C programs looking for a bus and/or event loop
12043 implementation that is minimal and does not have to be
12044 portable to other kernels.
12046 * kdbus support is no longer compile-time optional. It is now
12047 always built-in. However, it can still be disabled at
12048 runtime using the kdbus=0 kernel command line setting, and
12049 that setting may be changed to default to off, by specifying
12050 --disable-kdbus at build-time. Note though that the kernel
12051 command line setting has no effect if the kdbus.ko kernel
12052 module is not installed, in which case kdbus is (obviously)
12053 also disabled. We encourage all downstream distributions to
12054 begin testing kdbus by adding it to the kernel images in the
12055 development distributions, and leaving kdbus support in
12058 * The minimal required util-linux version has been bumped to
12061 * Support for chkconfig (--enable-chkconfig) was removed in
12062 favor of calling an abstraction tool
12063 /lib/systemd/systemd-sysv-install. This needs to be
12064 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
12065 in README for details.
12067 * If there's a systemd unit and a SysV init script for the
12068 same service name, and the user executes "systemctl enable"
12069 for it (or a related call), then this will now enable both
12070 (or execute the related operation on both), not just the
12073 * The libudev API documentation has been converted from gtkdoc
12076 * gudev has been removed from the systemd tree, it is now an
12079 * The systemd-cgtop tool learnt a new --raw switch to generate
12080 "raw" (machine parsable) output.
12082 * networkd's IPForwarding= .network file setting learnt the
12083 new setting "kernel", which ensures that networkd does not
12084 change the IP forwarding sysctl from the default kernel
12087 * The systemd-logind bus API now exposes a new boolean
12088 property "Docked" that reports whether logind considers the
12089 system "docked", i.e. connected to a docking station or not.
12091 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
12092 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
12093 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
12094 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
12095 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
12096 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
12097 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
12098 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
12099 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
12100 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
12101 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
12102 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
12103 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
12104 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12105 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
12106 Fink, Zbigniew Jędrzejewski-Szmek
12108 — Berlin, 2015-06-19
12112 * The gudev library has been extracted into a separate repository
12113 available at: https://git.gnome.org/browse/libgudev/
12114 It is now managed as part of the Gnome project. Distributions
12115 are recommended to pass --disable-gudev to systemd and use
12116 gudev from the Gnome project instead. gudev is still included
12117 in systemd, for now. It will be removed soon, though. Please
12118 also see the announcement-thread on systemd-devel:
12119 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
12121 * systemd now exposes a CPUUsageNSec= property for each
12122 service unit on the bus, that contains the overall consumed
12123 CPU time of a service (the sum of what each process of the
12124 service consumed). This value is only available if
12125 CPUAccounting= is turned on for a service, and is then shown
12126 in the "systemctl status" output.
12128 * Support for configuring alternative mappings of the old SysV
12129 runlevels to systemd targets has been removed. They are now
12130 hardcoded in a way that runlevels 2, 3, 4 all map to
12131 multi-user.target and 5 to graphical.target (which
12132 previously was already the default behaviour).
12134 * The auto-mounter logic gained support for mount point
12135 expiry, using a new TimeoutIdleSec= setting in .automount
12136 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
12138 * The EFI System Partition (ESP) as mounted to /boot by
12139 systemd-efi-boot-generator will now be unmounted
12140 automatically after 2 minutes of not being used. This should
12141 minimize the risk of ESP corruptions.
12143 * New /etc/fstab options x-systemd.requires= and
12144 x-systemd.requires-mounts-for= are now supported to express
12145 additional dependencies for mounts. This is useful for
12146 journaling file systems that support external journal
12147 devices or overlay file systems that require underlying file
12148 systems to be mounted.
12150 * systemd does not support direct live-upgrades (via systemctl
12151 daemon-reexec) from versions older than v44 anymore. As no
12152 distribution we are aware of shipped such old versions in a
12153 stable release this should not be problematic.
12155 * When systemd forks off a new per-connection service instance
12156 it will now set the $REMOTE_ADDR environment variable to the
12157 remote IP address, and $REMOTE_PORT environment variable to
12158 the remote IP port. This behaviour is similar to the
12159 corresponding environment variables defined by CGI.
12161 * systemd-networkd gained support for uplink failure
12162 detection. The BindCarrier= option allows binding interface
12163 configuration dynamically to the link sense of other
12164 interfaces. This is useful to achieve behaviour like in
12167 * systemd-networkd gained support for configuring the DHCP
12168 client identifier to use when requesting leases.
12170 * systemd-networkd now has a per-network UseNTP= option to
12171 configure whether NTP server information acquired via DHCP
12172 is passed on to services like systemd-timesyncd.
12174 * systemd-networkd gained support for vti6 tunnels.
12176 * Note that systemd-networkd manages the sysctl variable
12177 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
12178 it is configured for since v219. The variable controls IP
12179 forwarding, and is a per-interface alternative to the global
12180 /proc/sys/net/ipv[46]/ip_forward. This setting is
12181 configurable in the IPForward= option, which defaults to
12182 "no". This means if networkd is used for an interface it is
12183 no longer sufficient to set the global sysctl option to turn
12184 on IP forwarding! Instead, the .network file option
12185 IPForward= needs to be turned on! Note that the
12186 implementation of this behaviour was broken in v219 and has
12187 been fixed in v220.
12189 * Many bonding and vxlan options are now configurable in
12192 * systemd-nspawn gained a new --property= setting to set unit
12193 properties for the container scope. This is useful for
12194 setting resource parameters (e.g. "CPUShares=500") on
12195 containers started from the command line.
12197 * systemd-nspawn gained a new --private-users= switch to make
12198 use of user namespacing available on recent Linux kernels.
12200 * systemd-nspawn may now be called as part of a shell pipeline
12201 in which case the pipes used for stdin and stdout are passed
12202 directly to the process invoked in the container, without
12203 indirection via a pseudo tty.
12205 * systemd-nspawn gained a new switch to control the UNIX
12206 signal to use when killing the init process of the container
12207 when shutting down.
12209 * systemd-nspawn gained a new --overlay= switch for mounting
12210 overlay file systems into the container using the new kernel
12213 * When a container image is imported via systemd-importd and
12214 the host file system is not btrfs, a loopback block device
12215 file is created in /var/lib/machines.raw with a btrfs file
12216 system inside. It is then mounted to /var/lib/machines to
12217 enable btrfs features for container management. The loopback
12218 file and btrfs file system is grown as needed when container
12219 images are imported via systemd-importd.
12221 * systemd-machined/systemd-importd gained support for btrfs
12222 quota, to enforce container disk space limits on disk. This
12223 is exposed in "machinectl set-limit".
12225 * systemd-importd now can import containers from local .tar,
12226 .raw and .qcow2 images, and export them to .tar and .raw. It
12227 can also import dkr v2 images now from the network (on top
12230 * systemd-importd gained support for verifying downloaded
12231 images with gpg2 (previously only gpg1 was supported).
12233 * systemd-machined, systemd-logind, systemd: most bus calls are
12234 now accessible to unprivileged processes via polkit. Also,
12235 systemd-logind will now allow users to kill their own sessions
12236 without further privileges or authorization.
12238 * systemd-shutdownd has been removed. This service was
12239 previously responsible for implementing scheduled shutdowns
12240 as exposed in /usr/bin/shutdown's time parameter. This
12241 functionality has now been moved into systemd-logind and is
12242 accessible via a bus interface.
12244 * "systemctl reboot" gained a new switch --firmware-setup that
12245 can be used to reboot into the EFI firmware setup, if that
12246 is available. systemd-logind now exposes an API on the bus
12247 to trigger such reboots, in case graphical desktop UIs want
12248 to cover this functionality.
12250 * "systemctl enable", "systemctl disable" and "systemctl mask"
12251 now support a new "--now" switch. If specified the units
12252 that are enabled will also be started, and the ones
12253 disabled/masked also stopped.
12255 * The Gummiboot EFI boot loader tool has been merged into
12256 systemd, and renamed to "systemd-boot". The bootctl tool has been
12257 updated to support systemd-boot.
12259 * An EFI kernel stub has been added that may be used to create
12260 kernel EFI binaries that contain not only the actual kernel,
12261 but also an initrd, boot splash, command line and OS release
12262 information. This combined binary can then be signed as a
12263 single image, so that the firmware can verify it all in one
12264 step. systemd-boot has special support for EFI binaries created
12265 like this and can extract OS release information from them
12266 and show them in the boot menu. This functionality is useful
12267 to implement cryptographically verified boot schemes.
12269 * Optional support has been added to systemd-fsck to pass
12270 fsck's progress report to an AF_UNIX socket in the file
12273 * udev will no longer create device symlinks for all block devices by
12274 default. A deny list for excluding special block devices from this
12275 logic has been turned into an allow list that requires picking block
12276 devices explicitly that require device symlinks.
12278 * A new (currently still internal) API sd-device.h has been
12279 added to libsystemd. This modernized API is supposed to
12280 replace libudev eventually. In fact, already much of libudev
12281 is now just a wrapper around sd-device.h.
12283 * A new hwdb database for storing metadata about pointing
12284 stick devices has been added.
12286 * systemd-tmpfiles gained support for setting file attributes
12287 similar to the "chattr" tool with new 'h' and 'H' lines.
12289 * systemd-journald will no longer unconditionally set the
12290 btrfs NOCOW flag on new journal files. This is instead done
12291 with tmpfiles snippet using the new 'h' line type. This
12292 allows easy disabling of this logic, by masking the
12293 journal-nocow.conf tmpfiles file.
12295 * systemd-journald will now translate audit message types to
12296 human readable identifiers when writing them to the
12297 journal. This should improve readability of audit messages.
12299 * The LUKS logic gained support for the offset= and skip=
12300 options in /etc/crypttab, as previously implemented by
12303 * /usr/lib/os-release gained a new optional field VARIANT= for
12304 distributions that support multiple variants (such as a
12305 desktop edition, a server edition, …)
12307 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
12308 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
12309 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
12310 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
12311 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
12312 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
12313 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
12314 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
12315 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
12316 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
12317 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
12318 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
12319 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
12320 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
12321 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
12322 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
12323 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
12324 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
12325 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
12326 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
12327 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
12328 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
12329 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
12330 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
12331 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
12332 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
12333 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
12335 — Berlin, 2015-05-22
12339 * Introduce a new API "sd-hwdb.h" for querying the hardware
12340 metadata database. With this minimal interface one can query
12341 and enumerate the udev hwdb, decoupled from the old libudev
12342 library. libudev's interface for this is now only a wrapper
12343 around sd-hwdb. A new tool systemd-hwdb has been added to
12344 interface with and update the database.
12346 * When any of systemd's tools copies files (for example due to
12347 tmpfiles' C lines) a btrfs reflink will attempted first,
12348 before bytewise copying is done.
12350 * systemd-nspawn gained a new --ephemeral switch. When
12351 specified a btrfs snapshot is taken of the container's root
12352 directory, and immediately removed when the container
12353 terminates again. Thus, a container can be started whose
12354 changes never alter the container's root directory, and are
12355 lost on container termination. This switch can also be used
12356 for starting a container off the root file system of the
12357 host without affecting the host OS. This switch is only
12358 available on btrfs file systems.
12360 * systemd-nspawn gained a new --template= switch. It takes the
12361 path to a container tree to use as template for the tree
12362 specified via --directory=, should that directory be
12363 missing. This allows instantiating containers dynamically,
12364 on first run. This switch is only available on btrfs file
12367 * When a .mount unit refers to a mount point on which multiple
12368 mounts are stacked, and the .mount unit is stopped all of
12369 the stacked mount points will now be unmounted until no
12370 mount point remains.
12372 * systemd now has an explicit notion of supported and
12373 unsupported unit types. Jobs enqueued for unsupported unit
12374 types will now fail with an "unsupported" error code. More
12375 specifically .swap, .automount and .device units are not
12376 supported in containers, .busname units are not supported on
12377 non-kdbus systems. .swap and .automount are also not
12378 supported if their respective kernel compile time options
12381 * machinectl gained support for two new "copy-from" and
12382 "copy-to" commands for copying files from a running
12383 container to the host or vice versa.
12385 * machinectl gained support for a new "bind" command to bind
12386 mount host directories into local containers. This is
12387 currently only supported for nspawn containers.
12389 * networkd gained support for configuring bridge forwarding
12390 database entries (fdb) from .network files.
12392 * A new tiny daemon "systemd-importd" has been added that can
12393 download container images in tar, raw, qcow2 or dkr formats,
12394 and make them available locally in /var/lib/machines, so
12395 that they can run as nspawn containers. The daemon can GPG
12396 verify the downloads (not supported for dkr, since it has no
12397 provisions for verifying downloads). It will transparently
12398 decompress bz2, xz, gzip compressed downloads if necessary,
12399 and restore sparse files on disk. The daemon uses privilege
12400 separation to ensure the actual download logic runs with
12401 fewer privileges than the daemon itself. machinectl has
12402 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
12403 make the functionality of importd available to the
12404 user. With this in place the Fedora and Ubuntu "Cloud"
12405 images can be downloaded and booted as containers unmodified
12406 (the Fedora images lack the appropriate GPG signature files
12407 currently, so they cannot be verified, but this will change
12408 soon, hopefully). Note that downloading images is currently
12409 only fully supported on btrfs.
12411 * machinectl is now able to list container images found in
12412 /var/lib/machines, along with some metadata about sizes of
12413 disk and similar. If the directory is located on btrfs and
12414 quota is enabled, this includes quota display. A new command
12415 "image-status" has been added that shows additional
12416 information about images.
12418 * machinectl is now able to clone container images
12419 efficiently, if the underlying file system (btrfs) supports
12420 it, with the new "machinectl clone" command. It also
12421 gained commands for renaming and removing images, as well as
12422 marking them read-only or read-write (supported also on
12423 legacy file systems).
12425 * networkd gained support for collecting LLDP network
12426 announcements, from hardware that supports this. This is
12427 shown in networkctl output.
12429 * systemd-run gained support for a new -t (--pty) switch for
12430 invoking a binary on a pty whose input and output is
12431 connected to the invoking terminal. This allows executing
12432 processes as system services while interactively
12433 communicating with them via the terminal. Most interestingly
12434 this is supported across container boundaries. Invoking
12435 "systemd-run -t /bin/bash" is an alternative to running a
12436 full login session, the difference being that the former
12437 will not register a session, nor go through the PAM session
12440 * tmpfiles gained support for a new "v" line type for creating
12441 btrfs subvolumes. If the underlying file system is a legacy
12442 file system, this automatically degrades to creating a
12443 normal directory. Among others /var/lib/machines is now
12444 created like this at boot, should it be missing.
12446 * The directory /var/lib/containers/ has been deprecated and
12447 been replaced by /var/lib/machines. The term "machines" has
12448 been used in the systemd context as generic term for both
12449 VMs and containers, and hence appears more appropriate for
12450 this, as the directory can also contain raw images bootable
12453 * systemd-nspawn when invoked with -M but without --directory=
12454 or --image= is now capable of searching for the container
12455 root directory, subvolume or disk image automatically, in
12456 /var/lib/machines. systemd-nspawn@.service has been updated
12457 to make use of this, thus allowing it to be used for raw
12460 * A new machines.target unit has been introduced that is
12461 supposed to group all containers/VMs invoked as services on
12462 the system. systemd-nspawn@.service has been updated to
12463 integrate with that.
12465 * machinectl gained a new "start" command, for invoking a
12466 container as a service. "machinectl start foo" is mostly
12467 equivalent to "systemctl start systemd-nspawn@foo.service",
12468 but handles escaping in a nicer way.
12470 * systemd-nspawn will now mount most of the cgroupfs tree
12471 read-only into each container, with the exception of the
12472 container's own subtree in the name=systemd hierarchy.
12474 * journald now sets the special FS_NOCOW file flag for its
12475 journal files. This should improve performance on btrfs, by
12476 avoiding heavy fragmentation when journald's write-pattern
12477 is used on COW file systems. It degrades btrfs' data
12478 integrity guarantees for the files to the same levels as for
12479 ext3/ext4 however. This should be OK though as journald does
12480 its own data integrity checks and all its objects are
12481 checksummed on disk. Also, journald should handle btrfs disk
12482 full events a lot more gracefully now, by processing SIGBUS
12483 errors, and not relying on fallocate() anymore.
12485 * When journald detects that journal files it is writing to
12486 have been deleted it will immediately start new journal
12489 * systemd now provides a way to store file descriptors
12490 per-service in PID 1. This is useful for daemons to ensure
12491 that fds they require are not lost during a daemon
12492 restart. The fds are passed to the daemon on the next
12493 invocation in the same way socket activation fds are
12494 passed. This is now used by journald to ensure that the
12495 various sockets connected to all the system's stdout/stderr
12496 are not lost when journald is restarted. File descriptors
12497 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
12498 an extension to sd_notify(). Note that a limit is enforced
12499 on the number of fds a service can store in PID 1, and it
12500 defaults to 0, so that no fds may be stored, unless this is
12501 explicitly turned on.
12503 * The default TERM variable to use for units connected to a
12504 terminal, when no other value is explicitly is set is now
12505 vt220 rather than vt102. This should be fairly safe still,
12506 but allows PgUp/PgDn work.
12508 * The /etc/crypttab option header= as known from Debian is now
12511 * "loginctl user-status" and "loginctl session-status" will
12512 now show the last 10 lines of log messages of the
12513 user/session following the status output. Similar,
12514 "machinectl status" will show the last 10 log lines
12515 associated with a virtual machine or container
12516 service. (Note that this is usually not the log messages
12517 done in the VM/container itself, but simply what the
12518 container manager logs. For nspawn this includes all console
12521 * "loginctl session-status" without further argument will now
12522 show the status of the session of the caller. Similar,
12523 "lock-session", "unlock-session", "activate",
12524 "enable-linger", "disable-linger" may now be called without
12525 session/user parameter in which case they apply to the
12526 caller's session/user.
12528 * An X11 session scriptlet is now shipped that uploads
12529 $DISPLAY and $XAUTHORITY into the environment of the systemd
12530 --user daemon if a session begins. This should improve
12531 compatibility with X11 enabled applications run as systemd
12534 * Generators are now subject to masking via /etc and /run, the
12535 same way as unit files.
12537 * networkd .network files gained support for configuring
12538 per-link IPv4/IPv6 packet forwarding as well as IPv4
12539 masquerading. This is by default turned on for veth links to
12540 containers, as registered by systemd-nspawn. This means that
12541 nspawn containers run with --network-veth will now get
12542 automatic routed access to the host's networks without any
12543 further configuration or setup, as long as networkd runs on
12546 * systemd-nspawn gained the --port= (-p) switch to expose TCP
12547 or UDP posts of a container on the host. With this in place
12548 it is possible to run containers with private veth links
12549 (--network-veth), and have their functionality exposed on
12550 the host as if their services were running directly on the
12553 * systemd-nspawn's --network-veth switch now gained a short
12554 version "-n", since with the changes above it is now truly
12555 useful out-of-the-box. The systemd-nspawn@.service has been
12556 updated to make use of it too by default.
12558 * systemd-nspawn will now maintain a per-image R/W lock, to
12559 ensure that the same image is not started more than once
12560 writable. (It's OK to run an image multiple times
12561 simultaneously in read-only mode.)
12563 * systemd-nspawn's --image= option is now capable of
12564 dissecting and booting MBR and GPT disk images that contain
12565 only a single active Linux partition. Previously it
12566 supported only GPT disk images with proper GPT type
12567 IDs. This allows running cloud images from major
12568 distributions directly with systemd-nspawn, without
12571 * In addition to collecting mouse dpi data in the udev
12572 hardware database, there's now support for collecting angle
12573 information for mouse scroll wheels. The database is
12574 supposed to guarantee similar scrolling behavior on mice
12575 that it knows about. There's also support for collecting
12576 information about Touchpad types.
12578 * udev's input_id built-in will now also collect touch screen
12579 dimension data and attach it to probed devices.
12581 * /etc/os-release gained support for a Distribution Privacy
12584 * networkd gained support for creating "ipvlan", "gretap",
12585 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
12587 * systemd-tmpfiles gained support for "a" lines for setting
12590 * systemd-nspawn will now mount /tmp in the container to
12591 tmpfs, automatically.
12593 * systemd now exposes the memory.usage_in_bytes cgroup
12594 attribute and shows it for each service in the "systemctl
12595 status" output, if available.
12597 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
12598 immediate reboot is triggered. This useful if shutdown is
12599 hung and is unable to complete, to expedite the
12600 operation. Note that this kind of reboot will still unmount
12601 all file systems, and hence should not result in fsck being
12602 run on next reboot.
12604 * A .device unit for an optical block device will now be
12605 considered active only when a medium is in the drive. Also,
12606 mount units are now bound to their backing devices thus
12607 triggering automatic unmounting when devices become
12608 unavailable. With this in place systemd will now
12609 automatically unmount left-over mounts when a CD-ROM is
12610 ejected or a USB stick is yanked from the system.
12612 * networkd-wait-online now has support for waiting for
12613 specific interfaces only (with globbing), and for giving up
12614 after a configurable timeout.
12616 * networkd now exits when idle. It will be automatically
12617 restarted as soon as interfaces show up, are removed or
12618 change state. networkd will stay around as long as there is
12619 at least one DHCP state machine or similar around, that keep
12622 * networkd may now configure IPv6 link-local addressing in
12623 addition to IPv4 link-local addressing.
12625 * The IPv6 "token" for use in SLAAC may now be configured for
12626 each .network interface in networkd.
12628 * Routes configured with networkd may now be assigned a scope
12631 * networkd's [Match] sections now support globbing and lists
12632 of multiple space-separated matches per item.
12634 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
12635 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
12636 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
12637 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
12638 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
12639 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
12640 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
12641 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
12642 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
12643 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
12644 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
12645 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
12646 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
12647 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
12648 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
12649 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
12650 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
12651 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
12652 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
12653 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
12654 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
12655 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
12656 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
12657 Hoffmann, Zbigniew Jędrzejewski-Szmek
12659 — Berlin, 2015-02-16
12663 * When querying unit file enablement status (for example via
12664 "systemctl is-enabled"), a new state "indirect" is now known
12665 which indicates that a unit might not be enabled itself, but
12666 another unit listed in its Also= setting might be.
12668 * Similar to the various existing ConditionXYZ= settings for
12669 units, there are now matching AssertXYZ= settings. While
12670 failing conditions cause a unit to be skipped, but its job
12671 to succeed, failing assertions declared like this will cause
12672 a unit start operation and its job to fail.
12674 * hostnamed now knows a new chassis type "embedded".
12676 * systemctl gained a new "edit" command. When used on a unit
12677 file, this allows extending unit files with .d/ drop-in
12678 configuration snippets or editing the full file (after
12679 copying it from /usr/lib to /etc). This will invoke the
12680 user's editor (as configured with $EDITOR), and reload the
12681 modified configuration after editing.
12683 * "systemctl status" now shows the suggested enablement state
12684 for a unit, as declared in the (usually vendor-supplied)
12685 system preset files.
12687 * nss-myhostname will now resolve the single-label hostname
12688 "gateway" to the locally configured default IP routing
12689 gateways, ordered by their metrics. This assigns a stable
12690 name to the used gateways, regardless which ones are
12691 currently configured. Note that the name will only be
12692 resolved after all other name sources (if nss-myhostname is
12693 configured properly) and should hence not negatively impact
12694 systems that use the single-label hostname "gateway" in
12697 * systemd-inhibit now allows filtering by mode when listing
12700 * Scope and service units gained a new "Delegate" boolean
12701 property, which, when set, allows processes running inside the
12702 unit to further partition resources. This is primarily
12703 useful for systemd user instances as well as container
12706 * journald will now pick up audit messages directly from
12707 the kernel, and log them like any other log message. The
12708 audit fields are split up and fully indexed. This means that
12709 journalctl in many ways is now a (nicer!) alternative to
12710 ausearch, the traditional audit client. Note that this
12711 implements only a minimal audit client. If you want the
12712 special audit modes like reboot-on-log-overflow, please use
12713 the traditional auditd instead, which can be used in
12714 parallel to journald.
12716 * The ConditionSecurity= unit file option now understands the
12717 special string "audit" to check whether auditing is
12720 * journalctl gained two new commands --vacuum-size= and
12721 --vacuum-time= to delete old journal files until the
12722 remaining ones take up no more than the specified size on disk,
12723 or are not older than the specified time.
12725 * A new, native PPPoE library has been added to sd-network,
12726 systemd's library of light-weight networking protocols. This
12727 library will be used in a future version of networkd to
12728 enable PPPoE communication without an external pppd daemon.
12730 * The busctl tool now understands a new "capture" verb that
12731 works similar to "monitor", but writes a packet capture
12732 trace to STDOUT that can be redirected to a file which is
12733 compatible with libcap's capture file format. This can then
12734 be loaded in Wireshark and similar tools to inspect bus
12737 * The busctl tool now understands a new "tree" verb that shows
12738 the object trees of a specific service on the bus, or of all
12741 * The busctl tool now understands a new "introspect" verb that
12742 shows all interfaces and members of objects on the bus,
12743 including their signature and values. This is particularly
12744 useful to get more information about bus objects shown by
12745 the new "busctl tree" command.
12747 * The busctl tool now understands new verbs "call",
12748 "set-property" and "get-property" for invoking bus method
12749 calls, setting and getting bus object properties in a
12752 * busctl gained a new --augment-creds= argument that controls
12753 whether the tool shall augment credential information it
12754 gets from the bus with data from /proc, in a possibly
12757 * nspawn's --link-journal= switch gained two new values
12758 "try-guest" and "try-host" that work like "guest" and
12759 "host", but do not fail if the host has no persistent
12760 journaling enabled. -j is now equivalent to
12761 --link-journal=try-guest.
12763 * macvlan network devices created by nspawn will now have
12764 stable MAC addresses.
12766 * A new SmackProcessLabel= unit setting has been added, which
12767 controls the SMACK security label processes forked off by
12768 the respective unit shall use.
12770 * If compiled with --enable-xkbcommon, systemd-localed will
12771 verify x11 keymap settings by compiling the given keymap. It
12772 will spew out warnings if the compilation fails. This
12773 requires libxkbcommon to be installed.
12775 * When a coredump is collected, a larger number of metadata
12776 fields is now collected and included in the journal records
12777 created for it. More specifically, control group membership,
12778 environment variables, memory maps, working directory,
12779 chroot directory, /proc/$PID/status, and a list of open file
12780 descriptors is now stored in the log entry.
12782 * The udev hwdb now contains DPI information for mice. For
12785 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
12787 * All systemd programs that read standalone configuration
12788 files in /etc now also support a corresponding series of
12789 .conf.d configuration directories in /etc/, /run/,
12790 /usr/local/lib/, /usr/lib/, and (if configured with
12791 --enable-split-usr) /lib/. In particular, the following
12792 configuration files now have corresponding configuration
12793 directories: system.conf user.conf, logind.conf,
12794 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
12795 resolved.conf, timesyncd.conf, journal-remote.conf, and
12796 journal-upload.conf. Note that distributions should use the
12797 configuration directories in /usr/lib/; the directories in
12798 /etc/ are reserved for the system administrator.
12800 * systemd-rfkill will no longer take the rfkill device name
12801 into account when storing rfkill state on disk, as the name
12802 might be dynamically assigned and not stable. Instead, the
12803 ID_PATH udev variable combined with the rfkill type (wlan,
12804 bluetooth, …) is used.
12806 * A new service systemd-machine-id-commit.service has been
12807 added. When used on systems where /etc is read-only during
12808 boot, and /etc/machine-id is not initialized (but an empty
12809 file), this service will copy the temporary machine ID
12810 created as replacement into /etc after the system is fully
12811 booted up. This is useful for systems that are freshly
12812 installed with a non-initialized machine ID, but should get
12813 a fixed machine ID for subsequent boots.
12815 * networkd's .netdev files now provide a large set of
12816 configuration parameters for VXLAN devices. Similarly, the
12817 bridge port cost parameter is now configurable in .network
12818 files. There's also new support for configuring IP source
12819 routing. networkd .link files gained support for a new
12820 OriginalName= match that is useful to match against the
12821 original interface name the kernel assigned. .network files
12822 may include MTU= and MACAddress= fields for altering the MTU
12823 and MAC address while being connected to a specific network
12826 * The LUKS logic gained supported for configuring
12827 UUID-specific key files. There's also new support for naming
12828 LUKS device from the kernel command line, using the new
12829 luks.name= argument.
12831 * Timer units may now be transiently created via the bus API
12832 (this was previously already available for scope and service
12833 units). In addition it is now possible to create multiple
12834 transient units at the same time with a single bus call. The
12835 "systemd-run" tool has been updated to make use of this for
12836 running commands on a specified time, in at(1)-style.
12838 * tmpfiles gained support for "t" lines, for assigning
12839 extended attributes to files. Among other uses this may be
12840 used to assign SMACK labels to files.
12842 Contributions from: Alin Rauta, Alison Chaiken, Andrej
12843 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
12844 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
12845 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
12846 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
12847 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
12848 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
12849 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
12850 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
12851 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
12852 Michael Biebl, Michael Chapman, Michael Marineau, Michal
12853 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
12854 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
12855 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
12856 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
12857 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
12858 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
12859 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
12861 — Berlin, 2014-12-10
12865 * journalctl gained the new options -t/--identifier= to match
12866 on the syslog identifier (aka "tag"), as well as --utc to
12867 show log timestamps in the UTC timezone. journalctl now also
12868 accepts -n/--lines=all to disable line capping in a pager.
12870 * journalctl gained a new switch, --flush, that synchronously
12871 flushes logs from /run/log/journal to /var/log/journal if
12872 persistent storage is enabled. systemd-journal-flush.service
12873 now waits until the operation is complete.
12875 * Services can notify the manager before they start a reload
12876 (by sending RELOADING=1) or shutdown (by sending
12877 STOPPING=1). This allows the manager to track and show the
12878 internal state of daemons and closes a race condition when
12879 the process is still running but has closed its D-Bus
12882 * Services with Type=oneshot do not have to have any ExecStart
12885 * User units are now loaded also from
12886 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
12887 /run/systemd/user directory that was already previously
12888 supported, but is under the control of the user.
12890 * Job timeouts (i.e. timeouts on the time a job that is
12891 queued stays in the run queue) can now optionally result in
12892 immediate reboot or power-off actions (JobTimeoutAction= and
12893 JobTimeoutRebootArgument=). This is useful on ".target"
12894 units, to limit the maximum time a target remains
12895 undispatched in the run queue, and to trigger an emergency
12896 operation in such a case. This is now used by default to
12897 turn off the system if boot-up (as defined by everything in
12898 basic.target) hangs and does not complete for at least
12899 15min. Also, if power-off or reboot hang for at least 30min
12900 an immediate power-off/reboot operation is triggered. This
12901 functionality is particularly useful to increase reliability
12902 on embedded devices, but also on laptops which might
12903 accidentally get powered on when carried in a backpack and
12904 whose boot stays stuck in a hard disk encryption passphrase
12907 * systemd-logind can be configured to also handle lid switch
12908 events even when the machine is docked or multiple displays
12909 are attached (HandleLidSwitchDocked= option).
12911 * A helper binary and a service have been added which can be
12912 used to resume from hibernation in the initramfs. A
12913 generator will parse the resume= option on the kernel
12914 command line to trigger resume.
12916 * A user console daemon systemd-consoled has been
12917 added. Currently, it is a preview, and will so far open a
12918 single terminal on each session of the user marked as
12919 Desktop=systemd-console.
12921 * Route metrics can be specified for DHCP routes added by
12924 * The SELinux context of socket-activated services can be set
12925 from the information provided by the networking stack
12926 (SELinuxContextFromNet= option).
12928 * Userspace firmware loading support has been removed and
12929 the minimum supported kernel version is thus bumped to 3.7.
12931 * Timeout for udev workers has been increased from 1 to 3
12932 minutes, but a warning will be printed after 1 minute to
12933 help diagnose kernel modules that take a long time to load.
12935 * Udev rules can now remove tags on devices with TAG-="foobar".
12937 * systemd's readahead implementation has been removed. In many
12938 circumstances it didn't give expected benefits even for
12939 rotational disk drives and was becoming less relevant in the
12940 age of SSDs. As none of the developers has been using
12941 rotating media anymore, and nobody stepped up to actively
12942 maintain this component of systemd it has now been removed.
12944 * Swap units can use Options= to specify discard options.
12945 Discard options specified for swaps in /etc/fstab are now
12948 * Docker containers are now detected as a separate type of
12951 * The Password Agent protocol gained support for queries where
12952 the user input is shown, useful e.g. for user names.
12953 systemd-ask-password gained a new --echo option to turn that
12956 * The default sysctl.d/ snippets will now set:
12958 net.core.default_qdisc = fq_codel
12960 This selects Fair Queuing Controlled Delay as the default
12961 queuing discipline for network interfaces. fq_codel helps
12962 fight the network bufferbloat problem. It is believed to be
12963 a good default with no tuning required for most workloads.
12964 Downstream distributions may override this choice. On 10Gbit
12965 servers that do not do forwarding, "fq" may perform better.
12966 Systems without a good clocksource should use "pfifo_fast".
12968 * If kdbus is enabled during build a new option BusPolicy= is
12969 available for service units, that allows locking all service
12970 processes into a stricter bus policy, in order to limit
12971 access to various bus services, or even hide most of them
12972 from the service's view entirely.
12974 * networkctl will now show the .network and .link file
12975 networkd has applied to a specific interface.
12977 * sd-login gained a new API call sd_session_get_desktop() to
12978 query which desktop environment has been selected for a
12981 * UNIX utmp support is now compile-time optional to support
12982 legacy-free systems.
12984 * systemctl gained two new commands "add-wants" and
12985 "add-requires" for pulling in units from specific targets
12988 * If the word "rescue" is specified on the kernel command line
12989 the system will now boot into rescue mode (aka
12990 rescue.target), which was previously available only by
12991 specifying "1" or "systemd.unit=rescue.target" on the kernel
12992 command line. This new kernel command line option nicely
12993 mirrors the already existing "emergency" kernel command line
12996 * New kernel command line options mount.usr=, mount.usrflags=,
12997 mount.usrfstype= have been added that match root=, rootflags=,
12998 rootfstype= but allow mounting a specific file system to
13001 * The $NOTIFY_SOCKET is now also passed to control processes of
13002 services, not only the main process.
13004 * This version reenables support for fsck's -l switch. This
13005 means at least version v2.25 of util-linux is required for
13006 operation, otherwise dead-locks on device nodes may
13007 occur. Again: you need to update util-linux to at least
13008 v2.25 when updating systemd to v217.
13010 * The "multi-seat-x" tool has been removed from systemd, as
13011 its functionality has been integrated into X servers 1.16,
13012 and the tool is hence redundant. It is recommended to update
13013 display managers invoking this tool to simply invoke X
13014 directly from now on, again.
13016 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
13017 message flag has been added for all of systemd's polkit
13018 authenticated method calls has been added. In particular this
13019 now allows optional interactive authorization via polkit for
13020 many of PID1's privileged operations such as unit file
13021 enabling and disabling.
13023 * "udevadm hwdb --update" learnt a new switch "--usr" for
13024 placing the rebuilt hardware database in /usr instead of
13025 /etc. When used only hardware database entries stored in
13026 /usr will be used, and any user database entries in /etc are
13027 ignored. This functionality is useful for vendors to ship a
13028 pre-built database on systems where local configuration is
13029 unnecessary or unlikely.
13031 * Calendar time specifications in .timer units now also
13032 understand the strings "semi-annually", "quarterly" and
13033 "minutely" as shortcuts (in addition to the preexisting
13034 "annually", "hourly", …).
13036 * systemd-tmpfiles will now correctly create files in /dev
13037 at boot which are marked for creation only at boot. It is
13038 recommended to always create static device nodes with 'c!'
13039 and 'b!', so that they are created only at boot and not
13040 overwritten at runtime.
13042 * When the watchdog logic is used for a service (WatchdogSec=)
13043 and the watchdog timeout is hit the service will now be
13044 terminated with SIGABRT (instead of just SIGTERM), in order
13045 to make sure a proper coredump and backtrace is
13046 generated. This ensures that hanging services will result in
13047 similar coredump/backtrace behaviour as services that hit a
13048 segmentation fault.
13050 Contributions from: Andreas Henriksson, Andrei Borzenkov,
13051 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
13052 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
13053 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
13054 Herrmann, David Sommerseth, David Strauss, Emil Renner
13055 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
13056 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
13057 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
13058 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
13059 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
13060 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
13061 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
13062 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
13063 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
13064 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
13065 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
13066 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
13067 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
13068 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
13069 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
13070 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
13073 — Berlin, 2014-10-28
13077 * timedated no longer reads NTP implementation unit names from
13078 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
13079 implementations should add a
13081 Conflicts=systemd-timesyncd.service
13083 to their unit files to take over and replace systemd's NTP
13084 default functionality.
13086 * systemd-sysusers gained a new line type "r" for configuring
13087 which UID/GID ranges to allocate system users/groups
13088 from. Lines of type "u" may now add an additional column
13089 that specifies the home directory for the system user to be
13090 created. Also, systemd-sysusers may now optionally read user
13091 information from STDIN instead of a file. This is useful for
13092 invoking it from RPM preinst scriptlets that need to create
13093 users before the first RPM file is installed since these
13094 files might need to be owned by them. A new
13095 %sysusers_create_inline RPM macro has been introduced to do
13096 just that. systemd-sysusers now updates the shadow files as
13097 well as the user/group databases, which should enhance
13098 compatibility with certain tools like grpck.
13100 * A number of bus APIs of PID 1 now optionally consult polkit to
13101 permit access for otherwise unprivileged clients under certain
13102 conditions. Note that this currently doesn't support
13103 interactive authentication yet, but this is expected to be
13104 added eventually, too.
13106 * /etc/machine-info now has new fields for configuring the
13107 deployment environment of the machine, as well as the
13108 location of the machine. hostnamectl has been updated with
13109 new command to update these fields.
13111 * systemd-timesyncd has been updated to automatically acquire
13112 NTP server information from systemd-networkd, which might
13113 have been discovered via DHCP.
13115 * systemd-resolved now includes a caching DNS stub resolver
13116 and a complete LLMNR name resolution implementation. A new
13117 NSS module "nss-resolve" has been added which can be used
13118 instead of glibc's own "nss-dns" to resolve hostnames via
13119 systemd-resolved. Hostnames, addresses and arbitrary RRs may
13120 be resolved via systemd-resolved D-Bus APIs. In contrast to
13121 the glibc internal resolver systemd-resolved is aware of
13122 multi-homed system, and keeps DNS server and caches separate
13123 and per-interface. Queries are sent simultaneously on all
13124 interfaces that have DNS servers configured, in order to
13125 properly handle VPNs and local LANs which might resolve
13126 separate sets of domain names. systemd-resolved may acquire
13127 DNS server information from systemd-networkd automatically,
13128 which in turn might have discovered them via DHCP. A tool
13129 "systemd-resolve-host" has been added that may be used to
13130 query the DNS logic in resolved. systemd-resolved implements
13131 IDNA and automatically uses IDNA or UTF-8 encoding depending
13132 on whether classic DNS or LLMNR is used as transport. In the
13133 next releases we intend to add a DNSSEC and mDNS/DNS-SD
13134 implementation to systemd-resolved.
13136 * A new NSS module nss-mymachines has been added, that
13137 automatically resolves the names of all local registered
13138 containers to their respective IP addresses.
13140 * A new client tool "networkctl" for systemd-networkd has been
13141 added. It currently is entirely passive and will query
13142 networking configuration from udev, rtnetlink and networkd,
13143 and present it to the user in a very friendly
13144 way. Eventually, we hope to extend it to become a full
13145 control utility for networkd.
13147 * .socket units gained a new DeferAcceptSec= setting that
13148 controls the kernels' TCP_DEFER_ACCEPT sockopt for
13149 TCP. Similarly, support for controlling TCP keep-alive
13150 settings has been added (KeepAliveTimeSec=,
13151 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
13152 turning off Nagle's algorithm on TCP has been added
13155 * logind learned a new session type "web", for use in projects
13156 like Cockpit which register web clients as PAM sessions.
13158 * timer units with at least one OnCalendar= setting will now
13159 be started only after time-sync.target has been
13160 reached. This way they will not elapse before the system
13161 clock has been corrected by a local NTP client or
13162 similar. This is particular useful on RTC-less embedded
13163 machines, that come up with an invalid system clock.
13165 * systemd-nspawn's --network-veth= switch should now result in
13166 stable MAC addresses for both the outer and the inner side
13169 * systemd-nspawn gained a new --volatile= switch for running
13170 container instances with /etc or /var unpopulated.
13172 * The kdbus client code has been updated to use the new Linux
13173 3.17 memfd subsystem instead of the old kdbus-specific one.
13175 * systemd-networkd's DHCP client and server now support
13176 FORCERENEW. There are also new configuration options to
13177 configure the vendor client identifier and broadcast mode
13180 * systemd will no longer inform the kernel about the current
13181 timezone, as this is necessarily incorrect and racy as the
13182 kernel has no understanding of DST and similar
13183 concepts. This hence means FAT timestamps will be always
13184 considered UTC, similar to what Android is already
13185 doing. Also, when the RTC is configured to the local time
13186 (rather than UTC) systemd will never synchronize back to it,
13187 as this might confuse Windows at a later boot.
13189 * systemd-analyze gained a new command "verify" for offline
13190 validation of unit files.
13192 * systemd-networkd gained support for a couple of additional
13193 settings for bonding networking setups. Also, the metric for
13194 statically configured routes may now be configured. For
13195 network interfaces where this is appropriate the peer IP
13196 address may now be configured.
13198 * systemd-networkd's DHCP client will no longer request
13199 broadcasting by default, as this tripped up some networks.
13200 For hardware where broadcast is required the feature should
13201 be switched back on using RequestBroadcast=yes.
13203 * systemd-networkd will now set up IPv4LL addresses (when
13204 enabled) even if DHCP is configured successfully.
13206 * udev will now default to respect network device names given
13207 by the kernel when the kernel indicates that these are
13208 predictable. This behavior can be tweaked by changing
13209 NamePolicy= in the relevant .link file.
13211 * A new library systemd-terminal has been added that
13212 implements full TTY stream parsing and rendering. This
13213 library is supposed to be used later on for implementing a
13214 full userspace VT subsystem, replacing the current kernel
13217 * A new tool systemd-journal-upload has been added to push
13218 journal data to a remote system running
13219 systemd-journal-remote.
13221 * journald will no longer forward all local data to another
13222 running syslog daemon. This change has been made because
13223 rsyslog (which appears to be the most commonly used syslog
13224 implementation these days) no longer makes use of this, and
13225 instead pulls the data out of the journal on its own. Since
13226 forwarding the messages to a non-existent syslog server is
13227 more expensive than we assumed we have now turned this
13228 off. If you run a syslog server that is not a recent rsyslog
13229 version, you have to turn this option on again
13230 (ForwardToSyslog= in journald.conf).
13232 * journald now optionally supports the LZ4 compressor for
13233 larger journal fields. This compressor should perform much
13234 better than XZ which was the previous default.
13236 * machinectl now shows the IP addresses of local containers,
13237 if it knows them, plus the interface name of the container.
13239 * A new tool "systemd-escape" has been added that makes it
13240 easy to escape strings to build unit names and similar.
13242 * sd_notify() messages may now include a new ERRNO= field
13243 which is parsed and collected by systemd and shown among the
13244 "systemctl status" output for a service.
13246 * A new component "systemd-firstboot" has been added that
13247 queries the most basic systemd information (timezone,
13248 hostname, root password) interactively on first
13249 boot. Alternatively it may also be used to provision these
13250 things offline on OS images installed into directories.
13252 * The default sysctl.d/ snippets will now set
13254 net.ipv4.conf.default.promote_secondaries=1
13256 This has the benefit of no flushing secondary IP addresses
13257 when primary addresses are removed.
13259 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
13260 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
13261 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
13262 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
13263 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
13264 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
13265 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
13266 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
13267 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
13268 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
13269 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
13270 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
13271 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
13272 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
13273 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
13275 — Berlin, 2014-08-19
13279 * A new tool systemd-sysusers has been added. This tool
13280 creates system users and groups in /etc/passwd and
13281 /etc/group, based on static declarative system user/group
13282 definitions in /usr/lib/sysusers.d/. This is useful to
13283 enable factory resets and volatile systems that boot up with
13284 an empty /etc directory, and thus need system users and
13285 groups created during early boot. systemd now also ships
13286 with two default sysusers.d/ files for the most basic
13287 users and groups systemd and the core operating system
13290 * A new tmpfiles snippet has been added that rebuilds the
13291 essential files in /etc on boot, should they be missing.
13293 * A directive for ensuring automatic clean-up of
13294 /var/cache/man/ has been removed from the default
13295 configuration. This line should now be shipped by the man
13296 implementation. The necessary change has been made to the
13297 man-db implementation. Note that you need to update your man
13298 implementation to one that ships this line, otherwise no
13299 automatic clean-up of /var/cache/man will take place.
13301 * A new condition ConditionNeedsUpdate= has been added that
13302 may conditionalize services to only run when /etc or /var
13303 are "older" than the vendor operating system resources in
13304 /usr. This is useful for reconstructing or updating /etc
13305 after an offline update of /usr or a factory reset, on the
13306 next reboot. Services that want to run once after such an
13307 update or reset should use this condition and order
13308 themselves before the new systemd-update-done.service, which
13309 will mark the two directories as fully updated. A number of
13310 service files have been added making use of this, to rebuild
13311 the udev hardware database, the journald message catalog and
13312 dynamic loader cache (ldconfig). The systemd-sysusers tool
13313 described above also makes use of this now. With this in
13314 place it is now possible to start up a minimal operating
13315 system with /etc empty cleanly. For more information on the
13316 concepts involved see this recent blog story:
13318 https://0pointer.de/blog/projects/stateless.html
13320 * A new system group "input" has been introduced, and all
13321 input device nodes get this group assigned. This is useful
13322 for system-level software to get access to input devices. It
13323 complements what is already done for "audio" and "video".
13325 * systemd-networkd learnt minimal DHCPv4 server support in
13326 addition to the existing DHCPv4 client support. It also
13327 learnt DHCPv6 client and IPv6 Router Solicitation client
13328 support. The DHCPv4 client gained support for static routes
13329 passed in from the server. Note that the [DHCPv4] section
13330 known in older systemd-networkd versions has been renamed to
13331 [DHCP] and is now also used by the DHCPv6 client. Existing
13332 .network files using settings of this section should be
13333 updated, though compatibility is maintained. Optionally, the
13334 client hostname may now be sent to the DHCP server.
13336 * networkd gained support for vxlan virtual networks as well
13337 as tun/tap and dummy devices.
13339 * networkd gained support for automatic allocation of address
13340 ranges for interfaces from a system-wide pool of
13341 addresses. This is useful for dynamically managing a large
13342 number of interfaces with a single network configuration
13343 file. In particular this is useful to easily assign
13344 appropriate IP addresses to the veth links of a large number
13345 of nspawn instances.
13347 * RPM macros for processing sysusers, sysctl and binfmt
13348 drop-in snippets at package installation time have been
13351 * The /etc/os-release file should now be placed in
13352 /usr/lib/os-release. The old location is automatically
13353 created as symlink. /usr/lib is the more appropriate
13354 location of this file, since it shall actually describe the
13355 vendor operating system shipped in /usr, and not the
13356 configuration stored in /etc.
13358 * .mount units gained a new boolean SloppyOptions= setting
13359 that maps to mount(8)'s -s option which enables permissive
13360 parsing of unknown mount options.
13362 * tmpfiles learnt a new "L+" directive which creates a symlink
13363 but (unlike "L") deletes a pre-existing file first, should
13364 it already exist and not already be the correct
13365 symlink. Similarly, "b+", "c+" and "p+" directives have been
13366 added as well, which create block and character devices, as
13367 well as fifos in the filesystem, possibly removing any
13368 pre-existing files of different types.
13370 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
13371 'argument' field (which so far specified the source to
13372 symlink/copy the files from) is now optional. If omitted the
13373 same file os copied from /usr/share/factory/ suffixed by the
13374 full destination path. This is useful for populating /etc
13375 with essential files, by copying them from vendor defaults
13376 shipped in /usr/share/factory/etc.
13378 * A new command "systemctl preset-all" has been added that
13379 applies the service preset settings to all installed unit
13380 files. A new switch --preset-mode= has been added that
13381 controls whether only enable or only disable operations
13384 * A new command "systemctl is-system-running" has been added
13385 that allows checking the overall state of the system, for
13386 example whether it is fully up and running.
13388 * When the system boots up with an empty /etc, the equivalent
13389 to "systemctl preset-all" is executed during early boot, to
13390 make sure all default services are enabled after a factory
13393 * systemd now contains a minimal preset file that enables the
13394 most basic services systemd ships by default.
13396 * Unit files' [Install] section gained a new DefaultInstance=
13397 field for defining the default instance to create if a
13398 template unit is enabled with no instance specified.
13400 * A new passive target cryptsetup-pre.target has been added
13401 that may be used by services that need to make they run and
13402 finish before the first LUKS cryptographic device is set up.
13404 * The /dev/loop-control and /dev/btrfs-control device nodes
13405 are now owned by the "disk" group by default, opening up
13406 access to this group.
13408 * systemd-coredump will now automatically generate a
13409 stack trace of all core dumps taking place on the system,
13410 based on elfutils' libdw library. This stack trace is logged
13413 * systemd-coredump may now optionally store coredumps directly
13414 on disk (in /var/lib/systemd/coredump, possibly compressed),
13415 instead of storing them unconditionally in the journal. This
13416 mode is the new default. A new configuration file
13417 /etc/systemd/coredump.conf has been added to configure this
13418 and other parameters of systemd-coredump.
13420 * coredumpctl gained a new "info" verb to show details about a
13421 specific coredump. A new switch "-1" has also been added
13422 that makes sure to only show information about the most
13423 recent entry instead of all entries. Also, as the tool is
13424 generally useful now the "systemd-" prefix of the binary
13425 name has been removed. Distributions that want to maintain
13426 compatibility with the old name should add a symlink from
13427 the old name to the new name.
13429 * journald's SplitMode= now defaults to "uid". This makes sure
13430 that unprivileged users can access their own coredumps with
13431 coredumpctl without restrictions.
13433 * New kernel command line options "systemd.wants=" (for
13434 pulling an additional unit during boot), "systemd.mask="
13435 (for masking a specific unit for the boot), and
13436 "systemd.debug-shell" (for enabling the debug shell on tty9)
13437 have been added. This is implemented in the new generator
13438 "systemd-debug-generator".
13440 * systemd-nspawn will now by default filter a couple of
13441 syscalls for containers, among them those required for
13442 kernel module loading, direct x86 IO port access, swap
13443 management, and kexec. Most importantly though
13444 open_by_handle_at() is now prohibited for containers,
13445 closing a hole similar to a recently discussed vulnerability
13446 in docker regarding access to files on file hierarchies the
13447 container should normally not have access to. Note that, for
13448 nspawn, we generally make no security claims anyway (and
13449 this is explicitly documented in the man page), so this is
13450 just a fix for one of the most obvious problems.
13452 * A new man page file-hierarchy(7) has been added that
13453 contains a minimized, modernized version of the file system
13454 layout systemd expects, similar in style to the FHS
13455 specification or hier(5). A new tool systemd-path(1) has
13456 been added to query many of these paths for the local
13459 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
13460 longer done. Since the directory now has a per-user size
13461 limit, and is cleaned on logout this appears unnecessary,
13462 in particular since this now brings the lifecycle of this
13463 directory closer in line with how IPC objects are handled.
13465 * systemd.pc now exports a number of additional directories,
13466 including $libdir (which is useful to identify the library
13467 path for the primary architecture of the system), and a
13468 couple of drop-in directories.
13470 * udev's predictable network interface names now use the dev_port
13471 sysfs attribute, introduced in linux 3.15 instead of dev_id to
13472 distinguish between ports of the same PCI function. dev_id should
13473 only be used for ports using the same HW address, hence the need
13476 * machined has been updated to export the OS version of a
13477 container (read from /etc/os-release and
13478 /usr/lib/os-release) on the bus. This is now shown in
13479 "machinectl status" for a machine.
13481 * A new service setting RestartForceExitStatus= has been
13482 added. If configured to a set of exit signals or process
13483 return values, the service will be restarted when the main
13484 daemon process exits with any of them, regardless of the
13487 * systemctl's -H switch for connecting to remote systemd
13488 machines has been extended so that it may be used to
13489 directly connect to a specific container on the
13490 host. "systemctl -H root@foobar:waldi" will now connect as
13491 user "root" to host "foobar", and then proceed directly to
13492 the container named "waldi". Note that currently you have to
13493 authenticate as user "root" for this to work, as entering
13494 containers is a privileged operation.
13496 Contributions from: Andreas Henriksson, Benjamin Steinwender,
13497 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
13498 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
13499 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
13500 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
13501 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
13502 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
13503 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
13504 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
13505 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
13506 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
13507 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
13509 — Berlin, 2014-07-03
13513 * As an experimental feature, udev now tries to lock the
13514 disk device node (flock(LOCK_SH|LOCK_NB)) while it
13515 executes events for the disk or any of its partitions.
13516 Applications like partitioning programs can lock the
13517 disk device node (flock(LOCK_EX)) and claim temporary
13518 device ownership that way; udev will entirely skip all event
13519 handling for this disk and its partitions. If the disk
13520 was opened for writing, the close will trigger a partition
13521 table rescan in udev's "watch" facility, and if needed
13522 synthesize "change" events for the disk and all its partitions.
13523 This is now unconditionally enabled, and if it turns out to
13524 cause major problems, we might turn it on only for specific
13525 devices, or might need to disable it entirely. Device Mapper
13526 devices are excluded from this logic.
13528 * We temporarily dropped the "-l" switch for fsck invocations,
13529 since they collide with the flock() logic above. util-linux
13530 upstream has been changed already to avoid this conflict,
13531 and we will re-add "-l" as soon as util-linux with this
13532 change has been released.
13534 * The dependency on libattr has been removed. Since a long
13535 time, the extended attribute calls have moved to glibc, and
13536 libattr is thus unnecessary.
13538 * Virtualization detection works without privileges now. This
13539 means the systemd-detect-virt binary no longer requires
13540 CAP_SYS_PTRACE file capabilities, and our daemons can run
13541 with fewer privileges.
13543 * systemd-networkd now runs under its own "systemd-network"
13544 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
13545 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
13546 loses the ability to write to files owned by root this way.
13548 * Similarly, systemd-resolved now runs under its own
13549 "systemd-resolve" user with no capabilities remaining.
13551 * Similarly, systemd-bus-proxyd now runs under its own
13552 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
13554 * systemd-networkd gained support for setting up "veth"
13555 virtual Ethernet devices for container connectivity, as well
13556 as GRE and VTI tunnels.
13558 * systemd-networkd will no longer automatically attempt to
13559 manually load kernel modules necessary for certain tunnel
13560 transports. Instead, it is assumed the kernel loads them
13561 automatically when required. This only works correctly on
13562 very new kernels. On older kernels, please consider adding
13563 the kernel modules to /etc/modules-load.d/ as a work-around.
13565 * The resolv.conf file systemd-resolved generates has been
13566 moved to /run/systemd/resolve/. If you have a symlink from
13567 /etc/resolv.conf, it might be necessary to correct it.
13569 * Two new service settings, ProtectHome= and ProtectSystem=,
13570 have been added. When enabled, they will make the user data
13571 (such as /home) inaccessible or read-only and the system
13572 (such as /usr) read-only, for specific services. This allows
13573 very light-weight per-service sandboxing to avoid
13574 modifications of user data or system files from
13575 services. These two new switches have been enabled for all
13576 of systemd's long-running services, where appropriate.
13578 * Socket units gained new SocketUser= and SocketGroup=
13579 settings to set the owner user and group of AF_UNIX sockets
13580 and FIFOs in the file system.
13582 * Socket units gained a new RemoveOnStop= setting. If enabled,
13583 all FIFOS and sockets in the file system will be removed
13584 when the specific socket unit is stopped.
13586 * Socket units gained a new Symlinks= setting. It takes a list
13587 of symlinks to create to file system sockets or FIFOs
13588 created by the specific Unix sockets. This is useful to
13589 manage symlinks to socket nodes with the same lifecycle as
13592 * The /dev/log socket and /dev/initctl FIFO have been moved to
13593 /run, and have been replaced by symlinks. This allows
13594 connecting to these facilities even if PrivateDevices=yes is
13595 used for a service (which makes /dev/log itself unavailable,
13596 but /run is left). This also has the benefit of ensuring
13597 that /dev only contains device nodes, directories and
13598 symlinks, and nothing else.
13600 * sd-daemon gained two new calls sd_pid_notify() and
13601 sd_pid_notifyf(). They are similar to sd_notify() and
13602 sd_notifyf(), but allow overriding of the source PID of
13603 notification messages if permissions permit this. This is
13604 useful to send notify messages on behalf of a different
13605 process (for example, the parent process). The
13606 systemd-notify tool has been updated to make use of this
13607 when sending messages (so that notification messages now
13608 originate from the shell script invoking systemd-notify and
13609 not the systemd-notify process itself. This should minimize
13610 a race where systemd fails to associate notification
13611 messages to services when the originating process already
13614 * A new "on-abnormal" setting for Restart= has been added. If
13615 set, it will result in automatic restarts on all "abnormal"
13616 reasons for a process to exit, which includes unclean
13617 signals, core dumps, timeouts and watchdog timeouts, but
13618 does not include clean and unclean exit codes or clean
13619 signals. Restart=on-abnormal is an alternative for
13620 Restart=on-failure for services that shall be able to
13621 terminate and avoid restarts on certain errors, by
13622 indicating so with an unclean exit code. Restart=on-failure
13623 or Restart=on-abnormal is now the recommended setting for
13624 all long-running services.
13626 * If the InaccessibleDirectories= service setting points to a
13627 mount point (or if there are any submounts contained within
13628 it), it is now attempted to completely unmount it, to make
13629 the file systems truly unavailable for the respective
13632 * The ReadOnlyDirectories= service setting and
13633 systemd-nspawn's --read-only parameter are now recursively
13634 applied to all submounts, too.
13636 * Mount units may now be created transiently via the bus APIs.
13638 * The support for SysV and LSB init scripts has been removed
13639 from the systemd daemon itself. Instead, it is now
13640 implemented as a generator that creates native systemd units
13641 from these scripts when needed. This enables us to remove a
13642 substantial amount of legacy code from PID 1, following the
13643 fact that many distributions only ship a very small number
13644 of LSB/SysV init scripts nowadays.
13646 * Privileged Xen (dom0) domains are not considered
13647 virtualization anymore by the virtualization detection
13648 logic. After all, they generally have unrestricted access to
13649 the hardware and usually are used to manage the unprivileged
13652 * systemd-tmpfiles gained a new "C" line type, for copying
13653 files or entire directories.
13655 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
13656 lines. So far, they have been non-globbing versions of the
13657 latter, and have thus been redundant. In future, it is
13658 recommended to only use "z". "m" has hence been removed
13659 from the documentation, even though it stays supported.
13661 * A tmpfiles snippet to recreate the most basic structure in
13662 /var has been added. This is enough to create the /var/run →
13663 /run symlink and create a couple of structural
13664 directories. This allows systems to boot up with an empty or
13665 volatile /var. Of course, while with this change, the core OS
13666 now is capable with dealing with a volatile /var, not all
13667 user services are ready for it. However, we hope that sooner
13668 or later, many service daemons will be changed upstream so
13669 that they are able to automatically create their necessary
13670 directories in /var at boot, should they be missing. This is
13671 the first step to allow state-less systems that only require
13672 the vendor image for /usr to boot.
13674 * systemd-nspawn has gained a new --tmpfs= switch to mount an
13675 empty tmpfs instance to a specific directory. This is
13676 particularly useful for making use of the automatic
13677 reconstruction of /var (see above), by passing --tmpfs=/var.
13679 * Access modes specified in tmpfiles snippets may now be
13680 prefixed with "~", which indicates that they shall be masked
13681 by whether the existing file or directory is currently
13682 writable, readable or executable at all. Also, if specified,
13683 the sgid/suid/sticky bits will be masked for all
13686 * A new passive target unit "network-pre.target" has been
13687 added which is useful for services that shall run before any
13688 network is configured, for example firewall scripts.
13690 * The "floppy" group that previously owned the /dev/fd*
13691 devices is no longer used. The "disk" group is now used
13692 instead. Distributions should probably deprecate usage of
13695 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
13696 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
13697 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
13698 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
13699 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
13700 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
13703 — Berlin, 2014-06-11
13707 * A new "systemd-timesyncd" daemon has been added for
13708 synchronizing the system clock across the network. It
13709 implements an SNTP client. In contrast to NTP
13710 implementations such as chrony or the NTP reference server,
13711 this only implements a client side, and does not bother with
13712 the full NTP complexity, focusing only on querying time from
13713 one remote server and synchronizing the local clock to
13714 it. Unless you intend to serve NTP to networked clients or
13715 want to connect to local hardware clocks, this simple NTP
13716 client should be more than appropriate for most
13717 installations. The daemon runs with minimal privileges, and
13718 has been hooked up with networkd to only operate when
13719 network connectivity is available. The daemon saves the
13720 current clock to disk every time a new NTP sync has been
13721 acquired, and uses this to possibly correct the system clock
13722 early at bootup, in order to accommodate for systems that
13723 lack an RTC such as the Raspberry Pi and embedded devices,
13724 and to make sure that time monotonically progresses on these
13725 systems, even if it is not always correct. To make use of
13726 this daemon, a new system user and group "systemd-timesync"
13727 needs to be created on installation of systemd.
13729 * The queue "seqnum" interface of libudev has been disabled, as
13730 it was generally incompatible with device namespacing as
13731 sequence numbers of devices go "missing" if the devices are
13732 part of a different namespace.
13734 * "systemctl list-timers" and "systemctl list-sockets" gained
13735 a --recursive switch for showing units of these types also
13736 for all local containers, similar in style to the already
13737 supported --recursive switch for "systemctl list-units".
13739 * A new RebootArgument= setting has been added for service
13740 units, which may be used to specify a kernel reboot argument
13741 to use when triggering reboots with StartLimitAction=.
13743 * A new FailureAction= setting has been added for service
13744 units which may be used to specify an operation to trigger
13745 when a service fails. This works similarly to
13746 StartLimitAction=, but unlike it, controls what is done
13747 immediately rather than only after several attempts to
13748 restart the service in question.
13750 * hostnamed got updated to also expose the kernel name,
13751 release, and version on the bus. This is useful for
13752 executing commands like hostnamectl with the -H switch.
13753 systemd-analyze makes use of this to properly display
13754 details when running non-locally.
13756 * The bootchart tool can now show cgroup information in the
13757 graphs it generates.
13759 * The CFS CPU quota cgroup attribute is now exposed for
13760 services. The new CPUQuota= switch has been added for this
13761 which takes a percentage value. Setting this will have the
13762 result that a service may never get more CPU time than the
13763 specified percentage, even if the machine is otherwise idle.
13765 * systemd-networkd learned IPIP and SIT tunnel support.
13767 * LSB init scripts exposing a dependency on $network will now
13768 get a dependency on network-online.target rather than simply
13769 network.target. This should bring LSB handling closer to
13770 what it was on SysV systems.
13772 * A new fsck.repair= kernel option has been added to control
13773 how fsck shall deal with unclean file systems at boot.
13775 * The (.ini) configuration file parser will now silently ignore
13776 sections whose names begin with "X-". This may be used to maintain
13777 application-specific extension sections in unit files.
13779 * machined gained a new API to query the IP addresses of
13780 registered containers. "machinectl status" has been updated
13781 to show these addresses in its output.
13783 * A new call sd_uid_get_display() has been added to the
13784 sd-login APIs for querying the "primary" session of a
13785 user. The "primary" session of the user is elected from the
13786 user's sessions and generally a graphical session is
13787 preferred over a text one.
13789 * A minimal systemd-resolved daemon has been added. It
13790 currently simply acts as a companion to systemd-networkd and
13791 manages resolv.conf based on per-interface DNS
13792 configuration, possibly supplied via DHCP. In the long run
13793 we hope to extend this into a local DNSSEC enabled DNS and
13796 * The systemd-networkd-wait-online tool is now enabled by
13797 default. It will delay network-online.target until a network
13798 connection has been configured. The tool primarily integrates
13799 with networkd, but will also make a best effort to make sense
13800 of network configuration performed in some other way.
13802 * Two new service options StartupCPUShares= and
13803 StartupBlockIOWeight= have been added that work similarly to
13804 CPUShares= and BlockIOWeight= however only apply during
13805 system startup. This is useful to prioritize certain services
13806 differently during bootup than during normal runtime.
13808 * hostnamed has been changed to prefer the statically
13809 configured hostname in /etc/hostname (unless set to
13810 'localhost' or empty) over any dynamic one supplied by
13811 dhcp. With this change, the rules for picking the hostname
13812 match more closely the rules of other configuration settings
13813 where the local administrator's configuration in /etc always
13814 overrides any other settings.
13816 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
13817 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
13818 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
13819 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
13820 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
13821 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
13822 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
13823 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
13824 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
13825 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
13826 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
13827 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
13828 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
13829 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
13830 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
13831 Lindskog, WaLyong Cho, Will Woods, Zbigniew
13834 — Beijing, 2014-05-28
13838 * When restoring the screen brightness at boot, stay away from
13839 the darkest setting or from the lowest 5% of the available
13840 range, depending on which is the larger value of both. This
13841 should effectively protect the user from rebooting into a
13842 black screen, should the brightness have been set to minimum
13845 * sd-login gained a new sd_machine_get_class() call to
13846 determine the class ("vm" or "container") of a machine
13847 registered with machined.
13849 * sd-login gained new calls
13850 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
13851 to query the identity of the peer of a local AF_UNIX
13852 connection. They operate similarly to their sd_pid_get_xyz()
13855 * PID 1 will now maintain a system-wide system state engine
13856 with the states "starting", "running", "degraded",
13857 "maintenance", "stopping". These states are bound to system
13858 startup, normal runtime, runtime with at least one failed
13859 service, rescue/emergency mode and system shutdown. This
13860 state is shown in the "systemctl status" output when no unit
13861 name is passed. It is useful to determine system state, in
13862 particularly when doing so for many systems or containers at
13865 * A new command "list-machines" has been added to "systemctl"
13866 that lists all local OS containers and shows their system
13867 state (see above), if systemd runs inside of them.
13869 * systemctl gained a new "-r" switch to recursively enumerate
13870 units on all local containers, when used with the
13871 "list-unit" command (which is the default one that is
13872 executed when no parameters are specified).
13874 * The GPT automatic partition discovery logic will now honour
13875 two GPT partition flags: one may be set on a partition to
13876 cause it to be mounted read-only, and the other may be set
13877 on a partition to ignore it during automatic discovery.
13879 * Two new GPT type UUIDs have been added for automatic root
13880 partition discovery, for 32-bit and 64-bit ARM. This is not
13881 particularly useful for discovering the root directory on
13882 these architectures during bare-metal boots (since UEFI is
13883 not common there), but still very useful to allow booting of
13884 ARM disk images in nspawn with the -i option.
13886 * MAC addresses of interfaces created with nspawn's
13887 --network-interface= switch will now be generated from the
13888 machine name, and thus be stable between multiple invocations
13891 * logind will now automatically remove all IPC objects owned
13892 by a user if she or he fully logs out. This makes sure that
13893 users who are logged out cannot continue to consume IPC
13894 resources. This covers SysV memory, semaphores and message
13895 queues as well as POSIX shared memory and message
13896 queues. Traditionally, SysV and POSIX IPC had no lifecycle
13897 limits. With this functionality, that is corrected. This may
13898 be turned off by using the RemoveIPC= switch of logind.conf.
13900 * The systemd-machine-id-setup and tmpfiles tools gained a
13901 --root= switch to operate on a specific root directory,
13904 * journald can now forward logged messages to the TTYs of all
13905 logged in users ("wall"). This is the default for all
13906 emergency messages now.
13908 * A new tool systemd-journal-remote has been added to stream
13909 journal log messages across the network.
13911 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
13912 controller trees are mounted into it. Note that the
13913 directories mounted beneath it are not read-only. This is a
13914 security measure and is particularly useful because glibc
13915 actually includes a search logic to pick any tmpfs it can
13916 find to implement shm_open() if /dev/shm is not available
13917 (which it might very well be in namespaced setups).
13919 * machinectl gained a new "poweroff" command to cleanly power
13920 down a local OS container.
13922 * The PrivateDevices= unit file setting will now also drop the
13923 CAP_MKNOD capability from the capability bound set, and
13924 imply DevicePolicy=closed.
13926 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
13927 comprehensively on all long-running systemd services where
13928 this is appropriate.
13930 * systemd-udevd will now run in a disassociated mount
13931 namespace. To mount directories from udev rules, make sure to
13932 pull in mount units via SYSTEMD_WANTS properties.
13934 * The kdbus support gained support for uploading policy into
13935 the kernel. sd-bus gained support for creating "monitoring"
13936 connections that can eavesdrop into all bus communication
13937 for debugging purposes.
13939 * Timestamps may now be specified in seconds since the UNIX
13940 epoch Jan 1st, 1970 by specifying "@" followed by the value
13943 * Native tcpwrap support in systemd has been removed. tcpwrap
13944 is old code, not really maintained anymore and has serious
13945 shortcomings, and better options such as firewalls
13946 exist. For setups that require tcpwrap usage, please
13947 consider invoking your socket-activated service via tcpd,
13948 like on traditional inetd.
13950 * A new system.conf configuration option
13951 DefaultTimerAccuracySec= has been added that controls the
13952 default AccuracySec= setting of .timer units.
13954 * Timer units gained a new WakeSystem= switch. If enabled,
13955 timers configured this way will cause the system to resume
13956 from system suspend (if the system supports that, which most
13959 * Timer units gained a new Persistent= switch. If enabled,
13960 timers configured this way will save to disk when they have
13961 been last triggered. This information is then used on next
13962 reboot to possible execute overdue timer events, that
13963 could not take place because the system was powered off.
13964 This enables simple anacron-like behaviour for timer units.
13966 * systemctl's "list-timers" will now also list the time a
13967 timer unit was last triggered in addition to the next time
13968 it will be triggered.
13970 * systemd-networkd will now assign predictable IPv4LL
13971 addresses to its local interfaces.
13973 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
13974 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
13975 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
13976 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
13977 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
13978 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
13979 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
13980 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
13983 — Berlin, 2014-03-25
13987 * A new unit file setting RestrictAddressFamilies= has been
13988 added to restrict which socket address families unit
13989 processes gain access to. This takes address family names
13990 like "AF_INET" or "AF_UNIX", and is useful to minimize the
13991 attack surface of services via exotic protocol stacks. This
13992 is built on seccomp system call filters.
13994 * Two new unit file settings RuntimeDirectory= and
13995 RuntimeDirectoryMode= have been added that may be used to
13996 manage a per-daemon runtime directories below /run. This is
13997 an alternative for setting up directory permissions with
13998 tmpfiles snippets, and has the advantage that the runtime
13999 directory's lifetime is bound to the daemon runtime and that
14000 the daemon starts up with an empty directory each time. This
14001 is particularly useful when writing services that drop
14002 privileges using the User= or Group= setting.
14004 * The DeviceAllow= unit setting now supports globbing for
14005 matching against device group names.
14007 * The systemd configuration file system.conf gained new
14008 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
14009 DefaultMemoryAccounting= to globally turn on/off accounting
14010 for specific resources (cgroups) for all units. These
14011 settings may still be overridden individually in each unit
14014 * systemd-gpt-auto-generator is now able to discover /srv and
14015 root partitions in addition to /home and swap partitions. It
14016 also supports LUKS-encrypted partitions now. With this in
14017 place, automatic discovery of partitions to mount following
14018 the Discoverable Partitions Specification
14019 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
14020 is now a lot more complete. This allows booting without
14021 /etc/fstab and without root= on the kernel command line on
14022 systems prepared appropriately.
14024 * systemd-nspawn gained a new --image= switch which allows
14025 booting up disk images and Linux installations on any block
14026 device that follow the Discoverable Partitions Specification
14027 (see above). This means that installations made with
14028 appropriately updated installers may now be started and
14029 deployed using container managers, completely
14030 unmodified. (We hope that libvirt-lxc will add support for
14031 this feature soon, too.)
14033 * systemd-nspawn gained a new --network-macvlan= setting to
14034 set up a private macvlan interface for the
14035 container. Similarly, systemd-networkd gained a new
14036 Kind=macvlan setting in .netdev files.
14038 * systemd-networkd now supports configuring local addresses
14041 * A new tool systemd-network-wait-online has been added to
14042 synchronously wait for network connectivity using
14045 * The sd-bus.h bus API gained a new sd_bus_track object for
14046 tracking the lifecycle of bus peers. Note that sd-bus.h is
14047 still not a public API though (unless you specify
14048 --enable-kdbus on the configure command line, which however
14049 voids your warranty and you get no API stability guarantee).
14051 * The $XDG_RUNTIME_DIR runtime directories for each user are
14052 now individual tmpfs instances, which has the benefit of
14053 introducing separate pools for each user, with individual
14054 size limits, and thus making sure that unprivileged clients
14055 can no longer negatively impact the system or other users by
14056 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
14057 RuntimeDirectorySize= has been introduced that allows
14058 controlling the default size limit for all users. It
14059 defaults to 10% of the available physical memory. This is no
14060 replacement for quotas on tmpfs though (which the kernel
14061 still does not support), as /dev/shm and /tmp are still
14062 shared resources used by both the system and unprivileged
14065 * logind will now automatically turn off automatic suspending
14066 on laptop lid close when more than one display is
14067 connected. This was previously expected to be implemented
14068 individually in desktop environments (such as GNOME),
14069 however has been added to logind now, in order to fix a
14070 boot-time race where a desktop environment might not have
14071 been started yet and thus not been able to take an inhibitor
14072 lock at the time where logind already suspends the system
14073 due to a closed lid.
14075 * logind will now wait at least 30s after each system
14076 suspend/resume cycle, and 3min after system boot before
14077 suspending the system due to a closed laptop lid. This
14078 should give USB docking stations and similar enough time to
14079 be probed and configured after system resume and boot in
14080 order to then act as suspend blocker.
14082 * systemd-run gained a new --property= setting which allows
14083 initialization of resource control properties (and others)
14084 for the created scope or service unit. Example: "systemd-run
14085 --property=BlockIOWeight=10 updatedb" may be used to run
14086 updatedb at a low block IO scheduling weight.
14088 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
14089 now also work in --scope mode.
14091 * When systemd is compiled with kdbus support, basic support
14092 for enforced policies is now in place. (Note that enabling
14093 kdbus still voids your warranty and no API compatibility
14094 promises are made.)
14096 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
14097 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14098 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
14099 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
14100 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
14101 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
14102 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
14103 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
14104 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
14105 Zbigniew Jędrzejewski-Szmek
14107 — Berlin, 2014-03-12
14111 * systemd will now relabel /dev after loading the SMACK policy
14112 according to SMACK rules.
14114 * A new unit file option AppArmorProfile= has been added to
14115 set the AppArmor profile for the processes of a unit.
14117 * A new condition check ConditionArchitecture= has been added
14118 to conditionalize units based on the system architecture, as
14119 reported by uname()'s "machine" field.
14121 * systemd-networkd now supports matching on the system
14122 virtualization, architecture, kernel command line, hostname
14125 * logind is now a lot more aggressive when suspending the
14126 machine due to a closed laptop lid. Instead of acting only
14127 on the lid close action, it will continuously watch the lid
14128 status and act on it. This is useful for laptops where the
14129 power button is on the outside of the chassis so that it can
14130 be reached without opening the lid (such as the Lenovo
14131 Yoga). On those machines, logind will now immediately
14132 re-suspend the machine if the power button has been
14133 accidentally pressed while the laptop was suspended and in a
14134 backpack or similar.
14136 * logind will now watch SW_DOCK switches and inhibit reaction
14137 to the lid switch if it is pressed. This means that logind
14138 will not suspend the machine anymore if the lid is closed
14139 and the system is docked, if the laptop supports SW_DOCK
14140 notifications via the input layer. Note that ACPI docking
14141 stations do not generate this currently. Also note that this
14142 logic is usually not fully sufficient and Desktop
14143 Environments should take a lid switch inhibitor lock when an
14144 external display is connected, as systemd will not watch
14147 * nspawn will now make use of the devices cgroup controller by
14148 default, and only permit creation of and access to the usual
14149 API device nodes like /dev/null or /dev/random, as well as
14150 access to (but not creation of) the pty devices.
14152 * We will now ship a default .network file for
14153 systemd-networkd that automatically configures DHCP for
14154 network interfaces created by nspawn's --network-veth or
14155 --network-bridge= switches.
14157 * systemd will now understand the usual M, K, G, T suffixes
14158 according to SI conventions (i.e. to the base 1000) when
14159 referring to throughput and hardware metrics. It will stay
14160 with IEC conventions (i.e. to the base 1024) for software
14161 metrics, according to what is customary according to
14162 Wikipedia. We explicitly document which base applies for
14163 each configuration option.
14165 * The DeviceAllow= setting in unit files now supports a syntax to
14166 allow-list an entire group of devices node majors at once, based on
14167 the /proc/devices listing. For example, with the string "char-pts",
14168 it is now possible to allow-list all current and future pseudo-TTYs
14171 * sd-event learned a new "post" event source. Event sources of
14172 this type are triggered by the dispatching of any event
14173 source of a type that is not "post". This is useful for
14174 implementing clean-up and check event sources that are
14175 triggered by other work being done in the program.
14177 * systemd-networkd is no longer statically enabled, but uses
14178 the usual [Install] sections so that it can be
14179 enabled/disabled using systemctl. It still is enabled by
14182 * When creating a veth interface pair with systemd-nspawn, the
14183 host side will now be prefixed with "vb-" if
14184 --network-bridge= is used, and with "ve-" if --network-veth
14185 is used. This way, it is easy to distinguish these cases on
14186 the host, for example to apply different configuration to
14187 them with systemd-networkd.
14189 * The compatibility libraries for libsystemd-journal.so,
14190 libsystem-id128.so, libsystemd-login.so and
14191 libsystemd-daemon.so do not make use of IFUNC
14192 anymore. Instead, we now build libsystemd.so multiple times
14193 under these alternative names. This means that the footprint
14194 is drastically increased, but given that these are
14195 transitional compatibility libraries, this should not matter
14196 much. This change has been made necessary to support the ARM
14197 platform for these compatibility libraries, as the ARM
14198 toolchain is not really at the same level as the toolchain
14199 for other architectures like x86 and does not support
14200 IFUNC. Please make sure to use --enable-compat-libs only
14201 during a transitional period!
14203 * The .include syntax has been deprecated and is not documented
14204 anymore. Drop-in files in .d directories should be used instead.
14206 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
14207 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14208 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
14209 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
14210 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
14211 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
14212 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
14213 Zbigniew Jędrzejewski-Szmek
14215 — Berlin, 2014-02-24
14219 * A new component "systemd-networkd" has been added that can
14220 be used to configure local network interfaces statically or
14221 via DHCP. It is capable of bringing up bridges, VLANs, and
14222 bonding. Currently, no hook-ups for interactive network
14223 configuration are provided. Use this for your initrd,
14224 container, embedded, or server setup if you need a simple,
14225 yet powerful, network configuration solution. This
14226 configuration subsystem is quite nifty, as it allows wildcard
14227 hotplug matching in interfaces. For example, with a single
14228 configuration snippet, you can configure that all Ethernet
14229 interfaces showing up are automatically added to a bridge,
14230 or similar. It supports link-sensing and more.
14232 * A new tool "systemd-socket-proxyd" has been added which can
14233 act as a bidirectional proxy for TCP sockets. This is
14234 useful for adding socket activation support to services that
14235 do not actually support socket activation, including virtual
14236 machines and the like.
14238 * Add a new tool to save/restore rfkill state on
14241 * Save/restore state of keyboard backlights in addition to
14242 display backlights on shutdown/boot.
14244 * udev learned a new SECLABEL{} construct to label device
14245 nodes with a specific security label when they appear. For
14246 now, only SECLABEL{selinux} is supported, but the syntax is
14247 prepared for additional security frameworks.
14249 * udev gained a new scheme to configure link-level attributes
14250 from files in /etc/systemd/network/*.link. These files can
14251 match against MAC address, device path, driver name and type,
14252 and will apply attributes like the naming policy, link speed,
14253 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
14254 address assignment policy (randomized, …).
14256 * The configuration of network interface naming rules for
14257 "permanent interface names" has changed: a new NamePolicy=
14258 setting in the [Link] section of .link files determines the
14259 priority of possible naming schemes (onboard, slot, MAC,
14260 path). The default value of this setting is determined by
14261 /usr/lib/net/links/99-default.link. Old
14262 80-net-name-slot.rules udev configuration file has been
14263 removed, so local configuration overriding this file should
14264 be adapted to override 99-default.link instead.
14266 * When the User= switch is used in a unit file, also
14267 initialize $SHELL= based on the user database entry.
14269 * systemd no longer depends on libdbus. All communication is
14270 now done with sd-bus, systemd's low-level bus library
14273 * kdbus support has been added to PID 1 itself. When kdbus is
14274 enabled, this causes PID 1 to set up the system bus and
14275 enable support for a new ".busname" unit type that
14276 encapsulates bus name activation on kdbus. It works a little
14277 bit like ".socket" units, except for bus names. A new
14278 generator has been added that converts classic dbus1 service
14279 activation files automatically into native systemd .busname
14280 and .service units.
14282 * sd-bus: add a light-weight vtable implementation that allows
14283 defining objects on the bus with a simple static const
14284 vtable array of its methods, signals and properties.
14286 * systemd will not generate or install static dbus
14287 introspection data anymore to /usr/share/dbus-1/interfaces,
14288 as the precise format of these files is unclear, and
14289 nothing makes use of it.
14291 * A proxy daemon is now provided to proxy clients connecting
14292 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
14293 compatibility with classic D-Bus.
14295 * A bus driver implementation has been added that supports the
14296 classic D-Bus bus driver calls on kdbus, also for
14297 compatibility purposes.
14299 * A new API "sd-event.h" has been added that implements a
14300 minimal event loop API built around epoll. It provides a
14301 couple of features that direct epoll usage is lacking:
14302 prioritization of events, scales to large numbers of timer
14303 events, per-event timer slack (accuracy), system-wide
14304 coalescing of timer events, exit handlers, watchdog
14305 supervision support using systemd's sd_notify() API, child
14308 * A new API "sd-rntl.h" has been added that provides an API
14309 around the route netlink interface of the kernel, similar in
14310 style to "sd-bus.h".
14312 * A new API "sd-dhcp-client.h" has been added that provides a
14313 small DHCPv4 client-side implementation. This is used by
14314 "systemd-networkd".
14316 * There is a new kernel command line option
14317 "systemd.restore_state=0|1". When set to "0", none of the
14318 systemd tools will restore saved runtime state to hardware
14319 devices. More specifically, the rfkill and backlight states
14322 * The FsckPassNo= compatibility option in mount/service units
14323 has been removed. The fstab generator will now add the
14324 necessary dependencies automatically, and does not require
14325 PID1's support for that anymore.
14327 * journalctl gained a new switch, --list-boots, that lists
14328 recent boots with their times and boot IDs.
14330 * The various tools like systemctl, loginctl, timedatectl,
14331 busctl, systemd-run, … have gained a new switch "-M" to
14332 connect to a specific, local OS container (as direct
14333 connection, without requiring SSH). This works on any
14334 container that is registered with machined, such as those
14335 created by libvirt-lxc or nspawn.
14337 * systemd-run and systemd-analyze also gained support for "-H"
14338 to connect to remote hosts via SSH. This is particularly
14339 useful for systemd-run because it enables queuing of jobs
14340 onto remote systems.
14342 * machinectl gained a new command "login" to open a getty
14343 login in any local container. This works with any container
14344 that is registered with machined (such as those created by
14345 libvirt-lxc or nspawn), and which runs systemd inside.
14347 * machinectl gained a new "reboot" command that may be used to
14348 trigger a reboot on a specific container that is registered
14349 with machined. This works on any container that runs an init
14350 system of some kind.
14352 * systemctl gained a new "list-timers" command to print a nice
14353 listing of installed timer units with the times they elapse
14356 * Alternative reboot() parameters may now be specified on the
14357 "systemctl reboot" command line and are passed to the
14358 reboot() system call.
14360 * systemctl gained a new --job-mode= switch to configure the
14361 mode to queue a job with. This is a more generic version of
14362 --fail, --irreversible, and --ignore-dependencies, which are
14363 still available but not advertised anymore.
14365 * /etc/systemd/system.conf gained new settings to configure
14366 various default timeouts of units, as well as the default
14367 start limit interval and burst. These may still be overridden
14370 * PID1 will now export on the bus profile data of the security
14371 policy upload process (such as the SELinux policy upload to
14374 * journald: when forwarding logs to the console, include
14375 timestamps (following the setting in
14376 /sys/module/printk/parameters/time).
14378 * OnCalendar= in timer units now understands the special
14379 strings "yearly" and "annually". (Both are equivalent)
14381 * The accuracy of timer units is now configurable with the new
14382 AccuracySec= setting. It defaults to 1min.
14384 * A new dependency type JoinsNamespaceOf= has been added that
14385 allows running two services within the same /tmp and network
14386 namespace, if PrivateNetwork= or PrivateTmp= are used.
14388 * A new command "cat" has been added to systemctl. It outputs
14389 the original unit file of a unit, and concatenates the
14390 contents of additional "drop-in" unit file snippets, so that
14391 the full configuration is shown.
14393 * systemctl now supports globbing on the various "list-xyz"
14394 commands, like "list-units" or "list-sockets", as well as on
14395 those commands which take multiple unit names.
14397 * journalctl's --unit= switch gained support for globbing.
14399 * All systemd daemons now make use of the watchdog logic so
14400 that systemd automatically notices when they hang.
14402 * If the $container_ttys environment variable is set,
14403 getty-generator will automatically spawn a getty for each
14404 listed tty. This is useful for container managers to request
14405 login gettys to be spawned on as many ttys as needed.
14407 * %h, %s, %U specifier support is not available anymore when
14408 used in unit files for PID 1. This is because NSS calls are
14409 not safe from PID 1. They stay available for --user
14410 instances of systemd, and as special case for the root user.
14412 * loginctl gained a new "--no-legend" switch to turn off output
14413 of the legend text.
14415 * The "sd-login.h" API gained three new calls:
14416 sd_session_is_remote(), sd_session_get_remote_user(),
14417 sd_session_get_remote_host() to query information about
14420 * The udev hardware database now also carries vendor/product
14421 information of SDIO devices.
14423 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
14424 determine whether watchdog notifications are requested by
14425 the system manager.
14427 * Socket-activated per-connection services now include a
14428 short description of the connection parameters in the
14431 * tmpfiles gained a new "--boot" option. When this is not used,
14432 only lines where the command character is not suffixed with
14433 "!" are executed. When this option is specified, those
14434 options are executed too. This partitions tmpfiles
14435 directives into those that can be safely executed at any
14436 time, and those which should be run only at boot (for
14437 example, a line that creates /run/nologin).
14439 * A new API "sd-resolve.h" has been added which provides a simple
14440 asynchronous wrapper around glibc NSS hostname resolution
14441 calls, such as getaddrinfo(). In contrast to glibc's
14442 getaddrinfo_a(), it does not use signals. In contrast to most
14443 other asynchronous name resolution libraries, this one does
14444 not reimplement DNS, but reuses NSS, so that alternate
14445 hostname resolution systems continue to work, such as mDNS,
14446 LDAP, etc. This API is based on libasyncns, but it has been
14447 cleaned up for inclusion in systemd.
14449 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
14450 "sd-daemon.h" are no longer found in individual libraries
14451 libsystemd-journal.so, libsystemd-login.so,
14452 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
14453 merged them into a single library, libsystemd.so, which
14454 provides all symbols. The reason for this is cyclic
14455 dependencies, as these libraries tend to use each other's
14456 symbols. So far, we have managed to workaround that by linking
14457 a copy of a good part of our code into each of these
14458 libraries again and again, which, however, makes certain
14459 things hard to do, like sharing static variables. Also, it
14460 substantially increases footprint. With this change, there
14461 is only one library for the basic APIs systemd
14462 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
14463 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
14464 library as well, however are subject to the --enable-kdbus
14465 switch (see below). Note that "sd-dhcp-client.h" is not part
14466 of this library (this is because it only consumes, never
14467 provides, services of/to other APIs). To make the transition
14468 easy from the separate libraries to the unified one, we
14469 provide the --enable-compat-libs compile-time switch which
14470 will generate stub libraries that are compatible with the
14471 old ones but redirect all calls to the new one.
14473 * All of the kdbus logic and the new APIs "sd-bus.h",
14474 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
14475 and "sd-utf8.h" are compile-time optional via the
14476 "--enable-kdbus" switch, and they are not compiled in by
14477 default. To make use of kdbus, you have to explicitly enable
14478 the switch. Note however, that neither the kernel nor the
14479 userspace API for all of this is considered stable yet. We
14480 want to maintain the freedom to still change the APIs for
14481 now. By specifying this build-time switch, you acknowledge
14482 that you are aware of the instability of the current
14485 * Also, note that while kdbus is pretty much complete,
14486 it lacks one thing: proper policy support. This means you
14487 can build a fully working system with all features; however,
14488 it will be highly insecure. Policy support will be added in
14489 one of the next releases, at the same time that we will
14490 declare the APIs stable.
14492 * When the kernel command line argument "kdbus" is specified,
14493 systemd will automatically load the kdbus.ko kernel module. At
14494 this stage of development, it is only useful for testing kdbus
14495 and should not be used in production. Note: if "--enable-kdbus"
14496 is specified, and the kdbus.ko kernel module is available, and
14497 "kdbus" is added to the kernel command line, the entire system
14498 runs with kdbus instead of dbus-daemon, with the above mentioned
14499 problem of missing the system policy enforcement. Also a future
14500 version of kdbus.ko or a newer systemd will not be compatible with
14501 each other, and will unlikely be able to boot the machine if only
14502 one of them is updated.
14504 * systemctl gained a new "import-environment" command which
14505 uploads the caller's environment (or parts thereof) into the
14506 service manager so that it is inherited by services started
14507 by the manager. This is useful to upload variables like
14508 $DISPLAY into the user service manager.
14510 * A new PrivateDevices= switch has been added to service units
14511 which allows running a service with a namespaced /dev
14512 directory that does not contain any device nodes for
14513 physical devices. More specifically, it only includes devices
14514 such as /dev/null, /dev/urandom, and /dev/zero which are API
14517 * logind has been extended to support behaviour like VT
14518 switching on seats that do not support a VT. This makes
14519 multi-session available on seats that are not the first seat
14520 (seat0), and on systems where kernel support for VTs has
14521 been disabled at compile-time.
14523 * If a process holds a delay lock for system sleep or shutdown
14524 and fails to release it in time, we will now log its
14525 identity. This makes it easier to identify processes that
14526 cause slow suspends or power-offs.
14528 * When parsing /etc/crypttab, support for a new key-slot=
14529 option as supported by Debian is added. It allows indicating
14530 which LUKS slot to use on disk, speeding up key loading.
14532 * The sd_journal_sendv() API call has been checked and
14533 officially declared to be async-signal-safe so that it may
14534 be invoked from signal handlers for logging purposes.
14536 * Boot-time status output is now enabled automatically after a
14537 short timeout if boot does not progress, in order to give
14538 the user an indication what she or he is waiting for.
14540 * The boot-time output has been improved to show how much time
14541 remains until jobs expire.
14543 * The KillMode= switch in service units gained a new possible
14544 value "mixed". If set, and the unit is shut down, then the
14545 initial SIGTERM signal is sent only to the main daemon
14546 process, while the following SIGKILL signal is sent to
14547 all remaining processes of the service.
14549 * When a scope unit is registered, a new property "Controller"
14550 may be set. If set to a valid bus name, systemd will send a
14551 RequestStop() signal to this name when it would like to shut
14552 down the scope. This may be used to hook manager logic into
14553 the shutdown logic of scope units. Also, scope units may now
14554 be put in a special "abandoned" state, in which case the
14555 manager process which created them takes no further
14556 responsibilities for it.
14558 * When reading unit files, systemd will now verify
14559 the access mode of these files, and warn about certain
14560 suspicious combinations. This has been added to make it
14561 easier to track down packaging bugs where unit files are
14562 marked executable or world-writable.
14564 * systemd-nspawn gained a new "--setenv=" switch to set
14565 container-wide environment variables. The similar option in
14566 systemd-activate was renamed from "--environment=" to
14567 "--setenv=" for consistency.
14569 * systemd-nspawn has been updated to create a new kdbus domain
14570 for each container that is invoked, thus allowing each
14571 container to have its own set of system and user buses,
14572 independent of the host.
14574 * systemd-nspawn gained a new --drop-capability= switch to run
14575 the container with less capabilities than the default. Both
14576 --drop-capability= and --capability= now take the special
14577 string "all" for dropping or keeping all capabilities.
14579 * systemd-nspawn gained new switches for executing containers
14580 with specific SELinux labels set.
14582 * systemd-nspawn gained a new --quiet switch to not generate
14583 any additional output but the container's own console
14586 * systemd-nspawn gained a new --share-system switch to run a
14587 container without PID namespacing enabled.
14589 * systemd-nspawn gained a new --register= switch to control
14590 whether the container is registered with systemd-machined or
14591 not. This is useful for containers that do not run full
14592 OS images, but only specific apps.
14594 * systemd-nspawn gained a new --keep-unit which may be used
14595 when invoked as the only program from a service unit, and
14596 results in registration of the unit service itself in
14597 systemd-machined, instead of a newly opened scope unit.
14599 * systemd-nspawn gained a new --network-interface= switch for
14600 moving arbitrary interfaces to the container. The new
14601 --network-veth switch creates a virtual Ethernet connection
14602 between host and container. The new --network-bridge=
14603 switch then allows assigning the host side of this virtual
14604 Ethernet connection to a bridge device.
14606 * systemd-nspawn gained a new --personality= switch for
14607 setting the kernel personality for the container. This is
14608 useful when running a 32-bit container on a 64-bit host. A
14609 similar option Personality= is now also available for service
14612 * logind will now also track a "Desktop" identifier for each
14613 session which encodes the desktop environment of it. This is
14614 useful for desktop environments that want to identify
14615 multiple running sessions of itself easily.
14617 * A new SELinuxContext= setting for service units has been
14618 added that allows setting a specific SELinux execution
14619 context for a service.
14621 * Most systemd client tools will now honour $SYSTEMD_LESS for
14622 settings of the "less" pager. By default, these tools will
14623 override $LESS to allow certain operations to work, such as
14624 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
14625 influence this logic.
14627 * systemd's "seccomp" hook-up has been changed to make use of
14628 the libseccomp library instead of using its own
14629 implementation. This has benefits for portability among
14632 * For usage together with SystemCallFilter=, a new
14633 SystemCallErrorNumber= setting has been introduced that
14634 allows configuration of a system error number to be returned
14635 on filtered system calls, instead of immediately killing the
14636 process. Also, SystemCallArchitectures= has been added to
14637 limit access to system calls of a particular architecture
14638 (in order to turn off support for unused secondary
14639 architectures). There is also a global
14640 SystemCallArchitectures= setting in system.conf now to turn
14641 off support for non-native system calls system-wide.
14643 * systemd requires a kernel with a working name_to_handle_at(),
14644 please see the kernel config requirements in the README file.
14646 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
14647 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
14648 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
14649 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
14650 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
14651 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
14652 Elia Pinto, Florian Weimer, George McCollister, Goffredo
14653 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
14654 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
14655 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
14656 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
14657 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
14658 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
14659 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
14660 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
14661 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
14662 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
14663 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
14664 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
14665 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
14666 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
14667 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
14668 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
14669 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
14671 — Berlin, 2014-02-20
14675 * logind has gained support for facilitating privileged input
14676 and drm device access for unprivileged clients. This work is
14677 useful to allow Wayland display servers (and similar
14678 programs, such as kmscon) to run under the user's ID and
14679 access input and drm devices which are normally
14680 protected. When this is used (and the kernel is new enough)
14681 logind will "mute" IO on the file descriptors passed to
14682 Wayland as long as it is in the background and "unmute" it
14683 if it returns into the foreground. This allows secure
14684 session switching without allowing background sessions to
14685 eavesdrop on input and display data. This also introduces
14686 session switching support if VT support is turned off in the
14687 kernel, and on seats that are not seat0.
14689 * A new kernel command line option luks.options= is understood
14690 now which allows specifying LUKS options for usage for LUKS
14691 encrypted partitions specified with luks.uuid=.
14693 * tmpfiles.d(5) snippets may now use specifier expansion in
14694 path names. More specifically %m, %b, %H, %v, are now
14695 replaced by the local machine id, boot id, hostname, and
14696 kernel version number.
14698 * A new tmpfiles.d(5) command "m" has been introduced which
14699 may be used to change the owner/group/access mode of a file
14700 or directory if it exists, but do nothing if it does not.
14702 * This release removes high-level support for the
14703 MemorySoftLimit= cgroup setting. The underlying kernel
14704 cgroup attribute memory.soft_limit= is currently badly
14705 designed and likely to be removed from the kernel API in its
14706 current form, hence we should not expose it for now.
14708 * The memory.use_hierarchy cgroup attribute is now enabled for
14709 all cgroups systemd creates in the memory cgroup
14710 hierarchy. This option is likely to be come the built-in
14711 default in the kernel anyway, and the non-hierarchical mode
14712 never made much sense in the intrinsically hierarchical
14715 * A new field _SYSTEMD_SLICE= is logged along with all journal
14716 messages containing the slice a message was generated
14717 from. This is useful to allow easy per-customer filtering of
14718 logs among other things.
14720 * systemd-journald will no longer adjust the group of journal
14721 files it creates to the "systemd-journal" group. Instead we
14722 rely on the journal directory to be owned by the
14723 "systemd-journal" group, and its setgid bit set, so that the
14724 kernel file system layer will automatically enforce that
14725 journal files inherit this group assignment. The reason for
14726 this change is that we cannot allow NSS look-ups from
14727 journald which would be necessary to resolve
14728 "systemd-journal" to a numeric GID, because this might
14729 create deadlocks if NSS involves synchronous queries to
14730 other daemons (such as nscd, or sssd) which in turn are
14731 logging clients of journald and might block on it, which
14732 would then dead lock. A tmpfiles.d(5) snippet included in
14733 systemd will make sure the setgid bit and group are
14734 properly set on the journal directory if it exists on every
14735 boot. However, we recommend adjusting it manually after
14736 upgrades too (or from RPM scriptlets), so that the change is
14737 not delayed until next reboot.
14739 * Backlight and random seed files in /var/lib/ have moved into
14740 the /var/lib/systemd/ directory, in order to centralize all
14741 systemd generated files in one directory.
14743 * Boot time performance measurements (as displayed by
14744 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
14745 performance information if that's available to determine how
14746 much time BIOS and boot loader initialization required. With
14747 a sufficiently new BIOS you hence no longer need to boot
14748 with Gummiboot to get access to such information.
14750 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
14751 Cristian Rodríguez, Dave Reisner, David Herrmann, David
14752 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
14753 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
14754 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
14755 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
14756 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
14758 — Berlin, 2013-10-02
14762 * The Restart= option for services now understands a new
14763 on-watchdog setting, which will restart the service
14764 automatically if the service stops sending out watchdog keep
14765 alive messages (as configured with WatchdogSec=).
14767 * The getty generator (which is responsible for bringing up a
14768 getty on configured serial consoles) will no longer only
14769 start a getty on the primary kernel console but on all
14770 others, too. This makes the order in which console= is
14771 specified on the kernel command line less important.
14773 * libsystemd-logind gained a new sd_session_get_vt() call to
14774 retrieve the VT number of a session.
14776 * If the option "tries=0" is set for an entry of /etc/crypttab
14777 its passphrase is queried indefinitely instead of any
14778 maximum number of tries.
14780 * If a service with a configure PID file terminates its PID
14781 file will now be removed automatically if it still exists
14782 afterwards. This should put an end to stale PID files.
14784 * systemd-run will now also take relative binary path names
14785 for execution and no longer insists on absolute paths.
14787 * InaccessibleDirectories= and ReadOnlyDirectories= now take
14788 paths that are optionally prefixed with "-" to indicate that
14789 it should not be considered a failure if they do not exist.
14791 * journalctl -o (and similar commands) now understands a new
14792 output mode "short-precise", it is similar to "short" but
14793 shows timestamps with usec accuracy.
14795 * The option "discard" (as known from Debian) is now
14796 synonymous to "allow-discards" in /etc/crypttab. In fact,
14797 "discard" is preferred now (since it is easier to remember
14800 * Some licensing clean-ups were made, so that more code is now
14801 LGPL-2.1 licensed than before.
14803 * A minimal tool to save/restore the display backlight
14804 brightness across reboots has been added. It will store the
14805 backlight setting as late as possible at shutdown, and
14806 restore it as early as possible during reboot.
14808 * A logic to automatically discover and enable home and swap
14809 partitions on GPT disks has been added. With this in place
14810 /etc/fstab becomes optional for many setups as systemd can
14811 discover certain partitions located on the root disk
14812 automatically. Home partitions are recognized under their
14813 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
14814 partitions are recognized under their GPT type ID
14815 0657fd6da4ab43c484e50933c84b4f4f.
14817 * systemd will no longer pass any environment from the kernel
14818 or initrd to system services. If you want to set an
14819 environment for all services, do so via the kernel command
14820 line systemd.setenv= assignment.
14822 * The systemd-sysctl tool no longer natively reads the file
14823 /etc/sysctl.conf. If desired, the file should be symlinked
14824 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
14825 legacy support by a symlink rather than built-in code, it
14826 also makes the otherwise hidden order of application of the
14827 different files visible. (Note that this partly reverts to a
14828 pre-198 application order of sysctl knobs!)
14830 * The "systemctl set-log-level" and "systemctl dump" commands
14831 have been moved to systemd-analyze.
14833 * systemd-run learned the new --remain-after-exit switch,
14834 which causes the scope unit not to be cleaned up
14835 automatically after the process terminated.
14837 * tmpfiles learned a new --exclude-prefix= switch to exclude
14838 certain paths from operation.
14840 * journald will now automatically flush all messages to disk
14841 as soon as a message at the log level CRIT, ALERT or EMERG
14844 Contributions from: Andrew Cook, Brandon Philips, Christian
14845 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
14846 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
14847 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
14848 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
14849 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
14850 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
14851 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
14852 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
14853 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
14854 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
14855 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
14856 William Giokas, Zbigniew Jędrzejewski-Szmek
14858 — Berlin, 2013-09-13
14862 * The documentation has been updated to cover the various new
14863 concepts introduced with 205.
14865 * Unit files now understand the new %v specifier which
14866 resolves to the kernel version string as returned by "uname
14869 * systemctl now supports filtering the unit list output by
14870 load state, active state and sub state, using the new
14871 --state= parameter.
14873 * "systemctl status" will now show the results of the
14874 condition checks (like ConditionPathExists= and similar) of
14875 the last start attempts of the unit. They are also logged to
14878 * "journalctl -b" may now be used to look for boot output of a
14879 specific boot. Try "journalctl -b -1" for the previous boot,
14880 but the syntax is substantially more powerful.
14882 * "journalctl --show-cursor" has been added which prints the
14883 cursor string the last shown log line. This may then be used
14884 with the new "journalctl --after-cursor=" switch to continue
14885 browsing logs from that point on.
14887 * "journalctl --force" may now be used to force regeneration
14890 * Creation of "dead" device nodes has been moved from udev
14891 into kmod and tmpfiles. Previously, udev would read the kmod
14892 databases to pre-generate dead device nodes based on meta
14893 information contained in kernel modules, so that these would
14894 be auto-loaded on access rather then at boot. As this
14895 does not really have much to do with the exposing actual
14896 kernel devices to userspace this has always been slightly
14897 alien in the udev codebase. Following the new scheme kmod
14898 will now generate a runtime snippet for tmpfiles from the
14899 module meta information and it now is tmpfiles' job to the
14900 create the nodes. This also allows overriding access and
14901 other parameters for the nodes using the usual tmpfiles
14902 facilities. As side effect this allows us to remove the
14903 CAP_SYS_MKNOD capability bit from udevd entirely.
14905 * logind's device ACLs may now be applied to these "dead"
14906 devices nodes too, thus finally allowing managed access to
14907 devices such as /dev/snd/sequencer without loading the
14908 backing module right-away.
14910 * A new RPM macro has been added that may be used to apply
14911 tmpfiles configuration during package installation.
14913 * systemd-detect-virt and ConditionVirtualization= now can
14914 detect User-Mode-Linux machines (UML).
14916 * journald will now implicitly log the effective capabilities
14917 set of processes in the message metadata.
14919 * systemd-cryptsetup has gained support for TrueCrypt volumes.
14921 * The initrd interface has been simplified (more specifically,
14922 support for passing performance data via environment
14923 variables and fsck results via files in /run has been
14924 removed). These features were non-essential, and are
14925 nowadays available in a much nicer way by having systemd in
14926 the initrd serialize its state and have the hosts systemd
14927 deserialize it again.
14929 * The udev "keymap" data files and tools to apply keyboard
14930 specific mappings of scan to key codes, and force-release
14931 scan code lists have been entirely replaced by a udev
14932 "keyboard" builtin and a hwdb data file.
14934 * systemd will now honour the kernel's "quiet" command line
14935 argument also during late shutdown, resulting in a
14936 completely silent shutdown when used.
14938 * There's now an option to control the SO_REUSEPORT socket
14939 option in .socket units.
14941 * Instance units will now automatically get a per-template
14942 subslice of system.slice unless something else is explicitly
14943 configured. For example, instances of sshd@.service will now
14944 implicitly be placed in system-sshd.slice rather than
14945 system.slice as before.
14947 * Test coverage support may now be enabled at build time.
14949 Contributions from: Dave Reisner, Frederic Crozat, Harald
14950 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
14951 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
14952 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
14953 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
14954 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
14955 Giokas, Zbigniew Jędrzejewski-Szmek
14957 — Berlin, 2013-07-23
14961 * Two new unit types have been introduced:
14963 Scope units are very similar to service units, however, are
14964 created out of pre-existing processes — instead of PID 1
14965 forking off the processes. By using scope units it is
14966 possible for system services and applications to group their
14967 own child processes (worker processes) in a powerful way
14968 which then maybe used to organize them, or kill them
14969 together, or apply resource limits on them.
14971 Slice units may be used to partition system resources in an
14972 hierarchical fashion and then assign other units to them. By
14973 default there are now three slices: system.slice (for all
14974 system services), user.slice (for all user sessions),
14975 machine.slice (for VMs and containers).
14977 Slices and scopes have been introduced primarily in
14978 context of the work to move cgroup handling to a
14979 single-writer scheme, where only PID 1
14980 creates/removes/manages cgroups.
14982 * There's a new concept of "transient" units. In contrast to
14983 normal units these units are created via an API at runtime,
14984 not from configuration from disk. More specifically this
14985 means it is now possible to run arbitrary programs as
14986 independent services, with all execution parameters passed
14987 in via bus APIs rather than read from disk. Transient units
14988 make systemd substantially more dynamic then it ever was,
14989 and useful as a general batch manager.
14991 * logind has been updated to make use of scope and slice units
14992 for managing user sessions. As a user logs in he will get
14993 his own private slice unit, to which all sessions are added
14994 as scope units. We also added support for automatically
14995 adding an instance of user@.service for the user into the
14996 slice. Effectively logind will no longer create cgroup
14997 hierarchies on its own now, it will defer entirely to PID 1
14998 for this by means of scope, service and slice units. Since
14999 user sessions this way become entities managed by PID 1
15000 the output of "systemctl" is now a lot more comprehensive.
15002 * A new mini-daemon "systemd-machined" has been added which
15003 may be used by virtualization managers to register local
15004 VMs/containers. nspawn has been updated accordingly, and
15005 libvirt will be updated shortly. machined will collect a bit
15006 of meta information about the VMs/containers, and assign
15007 them their own scope unit (see above). The collected
15008 meta-data is then made available via the "machinectl" tool,
15009 and exposed in "ps" and similar tools. machined/machinectl
15010 is compile-time optional.
15012 * As discussed earlier, the low-level cgroup configuration
15013 options ControlGroup=, ControlGroupModify=,
15014 ControlGroupPersistent=, ControlGroupAttribute= have been
15015 removed. Please use high-level attribute settings instead as
15016 well as slice units.
15018 * A new bus call SetUnitProperties() has been added to alter
15019 various runtime parameters of a unit. This is primarily
15020 useful to alter cgroup parameters dynamically in a nice way,
15021 but will be extended later on to make more properties
15022 modifiable at runtime. systemctl gained a new set-properties
15023 command that wraps this call.
15025 * A new tool "systemd-run" has been added which can be used to
15026 run arbitrary command lines as transient services or scopes,
15027 while configuring a number of settings via the command
15028 line. This tool is currently very basic, however already
15029 very useful. We plan to extend this tool to even allow
15030 queuing of execution jobs with time triggers from the
15031 command line, similar in fashion to "at".
15033 * nspawn will now inform the user explicitly that kernels with
15034 audit enabled break containers, and suggest the user to turn
15037 * Support for detecting the IMA and AppArmor security
15038 frameworks with ConditionSecurity= has been added.
15040 * journalctl gained a new "-k" switch for showing only kernel
15041 messages, mimicking dmesg output; in addition to "--user"
15042 and "--system" switches for showing only user's own logs
15045 * systemd-delta can now show information about drop-in
15046 snippets extending unit files.
15048 * libsystemd-bus has been substantially updated but is still
15049 not available as public API.
15051 * systemd will now look for the "debug" argument on the kernel
15052 command line and enable debug logging, similar to what
15053 "systemd.log_level=debug" already did before.
15055 * "systemctl set-default", "systemctl get-default" has been
15056 added to configure the default.target symlink, which
15057 controls what to boot into by default.
15059 * "systemctl set-log-level" has been added as a convenient
15060 way to raise and lower systemd logging threshold.
15062 * "systemd-analyze plot" will now show the time the various
15063 generators needed for execution, as well as information
15064 about the unit file loading.
15066 * libsystemd-journal gained a new sd_journal_open_files() call
15067 for opening specific journal files. journactl also gained a
15068 new switch to expose this new functionality. Previously we
15069 only supported opening all files from a directory, or all
15070 files from the system, as opening individual files only is
15071 racy due to journal file rotation.
15073 * systemd gained the new DefaultEnvironment= setting in
15074 /etc/systemd/system.conf to set environment variables for
15077 * If a privileged process logs a journal message with the
15078 OBJECT_PID= field set, then journald will automatically
15079 augment this with additional OBJECT_UID=, OBJECT_GID=,
15080 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
15081 system services want to log events about specific client
15082 processes. journactl/systemctl has been updated to make use
15083 of this information if all log messages regarding a specific
15086 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
15087 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
15088 Reisner, David Coppa, David King, David Strauss, Eelco
15089 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
15090 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
15091 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
15092 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
15093 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
15094 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
15095 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
15096 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
15097 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
15098 Łukasz Stelmach, 장동준
15102 * The Python bindings gained some minimal support for the APIs
15103 exposed by libsystemd-logind.
15105 * ConditionSecurity= gained support for detecting SMACK. Since
15106 this condition already supports SELinux and AppArmor we only
15107 miss IMA for this. Patches welcome!
15109 Contributions from: Karol Lewandowski, Lennart Poettering,
15110 Zbigniew Jędrzejewski-Szmek
15114 * systemd-nspawn will now create /etc/resolv.conf if
15115 necessary, before bind-mounting the host's file onto it.
15117 * systemd-nspawn will now store meta information about a
15118 container on the container's cgroup as extended attribute
15119 fields, including the root directory.
15121 * The cgroup hierarchy has been reworked in many ways. All
15122 objects any of the components systemd creates in the cgroup
15123 tree are now suffixed. More specifically, user sessions are
15124 now placed in cgroups suffixed with ".session", users in
15125 cgroups suffixed with ".user", and nspawn containers in
15126 cgroups suffixed with ".nspawn". Furthermore, all cgroup
15127 names are now escaped in a simple scheme to avoid collision
15128 of userspace object names with kernel filenames. This work
15129 is preparation for making these objects relocatable in the
15130 cgroup tree, in order to allow easy resource partitioning of
15131 these objects without causing naming conflicts.
15133 * systemctl list-dependencies gained the new switches
15134 --plain, --reverse, --after and --before.
15136 * systemd-inhibit now shows the process name of processes that
15137 have taken an inhibitor lock.
15139 * nss-myhostname will now also resolve "localhost"
15140 implicitly. This makes /etc/hosts an optional file and
15141 nicely handles that on IPv6 ::1 maps to both "localhost" and
15142 the local hostname.
15144 * libsystemd-logind.so gained a new call
15145 sd_get_machine_names() to enumerate running containers and
15146 VMs (currently only supported by very new libvirt and
15147 nspawn). sd_login_monitor can now be used to watch
15148 VMs/containers coming and going.
15150 * .include is not allowed recursively anymore, and only in
15151 unit files. Usually it is better to use drop-in snippets in
15152 .d/*.conf anyway, as introduced with systemd 198.
15154 * systemd-analyze gained a new "critical-chain" command that
15155 determines the slowest chain of units run during system
15156 boot-up. It is very useful for tracking down where
15157 optimizing boot time is the most beneficial.
15159 * systemd will no longer allow manipulating service paths in
15160 the name=systemd:/system cgroup tree using ControlGroup= in
15161 units. (But is still fine with it in all other dirs.)
15163 * There's a new systemd-nspawn@.service service file that may
15164 be used to easily run nspawn containers as system
15165 services. With the container's root directory in
15166 /var/lib/container/foobar it is now sufficient to run
15167 "systemctl start systemd-nspawn@foobar.service" to boot it.
15169 * systemd-cgls gained a new parameter "--machine" to list only
15170 the processes within a certain container.
15172 * ConditionSecurity= now can check for "apparmor". We still
15173 are lacking checks for SMACK and IMA for this condition
15174 check though. Patches welcome!
15176 * A new configuration file /etc/systemd/sleep.conf has been
15177 added that may be used to configure which kernel operation
15178 systemd is supposed to execute when "suspend", "hibernate"
15179 or "hybrid-sleep" is requested. This makes the new kernel
15180 "freeze" state accessible to the user.
15182 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
15183 the passed argument if applicable.
15185 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
15186 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
15187 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
15188 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
15189 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
15190 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
15191 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
15196 * The output of 'systemctl list-jobs' got some polishing. The
15197 '--type=' argument may now be passed more than once. A new
15198 command 'systemctl list-sockets' has been added which shows
15199 a list of kernel sockets systemd is listening on with the
15200 socket units they belong to, plus the units these socket
15203 * The experimental libsystemd-bus library got substantial
15204 updates to work in conjunction with the (also experimental)
15205 kdbus kernel project. It works well enough to exchange
15206 messages with some sophistication. Note that kdbus is not
15207 ready yet, and the library is mostly an elaborate test case
15208 for now, and not installable.
15210 * systemd gained a new unit 'systemd-static-nodes.service'
15211 that generates static device nodes earlier during boot, and
15212 can run in conjunction with udev.
15214 * libsystemd-login gained a new call sd_pid_get_user_unit()
15215 to retrieve the user systemd unit a process is running
15216 in. This is useful for systems where systemd is used as
15219 * systemd-nspawn now places all containers in the new /machine
15220 top-level cgroup directory in the name=systemd
15221 hierarchy. libvirt will soon do the same, so that we get a
15222 uniform separation of /system, /user and /machine for system
15223 services, user processes and containers/virtual
15224 machines. This new cgroup hierarchy is also useful to stick
15225 stable names to specific container instances, which can be
15226 recognized later this way (this name may be controlled
15227 via systemd-nspawn's new -M switch). libsystemd-login also
15228 gained a new call sd_pid_get_machine_name() to retrieve the
15229 name of the container/VM a specific process belongs to.
15231 * bootchart can now store its data in the journal.
15233 * libsystemd-journal gained a new call
15234 sd_journal_add_conjunction() for AND expressions to the
15235 matching logic. This can be used to express more complex
15236 logical expressions.
15238 * journactl can now take multiple --unit= and --user-unit=
15241 * The cryptsetup logic now understands the "luks.key=" kernel
15242 command line switch for specifying a file to read the
15243 decryption key from. Also, if a configured key file is not
15244 found the tool will now automatically fall back to prompting
15247 * Python systemd.journal module was updated to wrap recently
15248 added functions from libsystemd-journal. The interface was
15249 changed to bring the low level interface in s.j._Reader
15250 closer to the C API, and the high level interface in
15251 s.j.Reader was updated to wrap and convert all data about
15254 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
15255 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
15256 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
15257 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
15258 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
15259 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15263 * journalctl --update-catalog now understands a new --root=
15264 option to operate on catalogs found in a different root
15267 * During shutdown after systemd has terminated all running
15268 services a final killing loop kills all remaining left-over
15269 processes. We will now print the name of these processes
15270 when we send SIGKILL to them, since this usually indicates a
15273 * If /etc/crypttab refers to password files stored on
15274 configured mount points automatic dependencies will now be
15275 generated to ensure the specific mount is established first
15276 before the key file is attempted to be read.
15278 * 'systemctl status' will now show information about the
15279 network sockets a socket unit is listening on.
15281 * 'systemctl status' will also shown information about any
15282 drop-in configuration file for units. (Drop-In configuration
15283 files in this context are files such as
15284 /etc/systemd/system/foobar.service.d/*.conf)
15286 * systemd-cgtop now optionally shows summed up CPU times of
15287 cgroups. Press '%' while running cgtop to switch between
15288 percentage and absolute mode. This is useful to determine
15289 which cgroups use up the most CPU time over the entire
15290 runtime of the system. systemd-cgtop has also been updated
15291 to be 'pipeable' for processing with further shell tools.
15293 * 'hostnamectl set-hostname' will now allow setting of FQDN
15296 * The formatting and parsing of time span values has been
15297 changed. The parser now understands fractional expressions
15298 such as "5.5h". The formatter will now output fractional
15299 expressions for all time spans under 1min, i.e. "5.123456s"
15300 rather than "5s 123ms 456us". For time spans under 1s
15301 millisecond values are shown, for those under 1ms
15302 microsecond values are shown. This should greatly improve
15303 all time-related output of systemd.
15305 * libsystemd-login and libsystemd-journal gained new
15306 functions for querying the poll() events mask and poll()
15307 timeout value for integration into arbitrary event
15310 * localectl gained the ability to list available X11 keymaps
15311 (models, layouts, variants, options).
15313 * 'systemd-analyze dot' gained the ability to filter for
15314 specific units via shell-style globs, to create smaller,
15315 more useful graphs. I.e. it is now possible to create simple
15316 graphs of all the dependencies between only target units, or
15317 of all units that Avahi has dependencies with.
15319 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
15320 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
15321 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
15322 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
15323 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
15324 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
15325 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
15329 * The boot-time readahead implementation for rotating media
15330 will now read the read-ahead data in multiple passes which
15331 consist of all read requests made in equidistant time
15332 intervals. This means instead of strictly reading read-ahead
15333 data in its physical order on disk we now try to find a
15334 middle ground between physical and access time order.
15336 * /etc/os-release files gained a new BUILD_ID= field for usage
15337 on operating systems that provide continuous builds of OS
15340 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
15341 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
15342 William Douglas, Zbigniew Jędrzejewski-Szmek
15346 * systemd-python gained an API exposing libsystemd-daemon.
15348 * The SMACK setup logic gained support for uploading CIPSO
15351 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
15352 ReadOnlyDirectories= and InaccessibleDirectories= has
15353 changed. The private /tmp and /var/tmp directories are now
15354 shared by all processes of a service (which means
15355 ExecStartPre= may now leave data in /tmp that ExecStart= of
15356 the same service can still access). When a service is
15357 stopped its temporary directories are immediately deleted
15358 (normal clean-up with tmpfiles is still done in addition to
15361 * By default, systemd will now set a couple of sysctl
15362 variables in the kernel: the safe sysrq options are turned
15363 on, IP route verification is turned on, and source routing
15364 disabled. The recently added hardlink and softlink
15365 protection of the kernel is turned on. These settings should
15366 be reasonably safe, and good defaults for all new systems.
15368 * The predictable network naming logic may now be turned off
15369 with a new kernel command line switch: net.ifnames=0.
15371 * A new libsystemd-bus module has been added that implements a
15372 pretty complete D-Bus client library. For details see:
15374 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
15376 * journald will now explicitly flush the journal files to disk
15377 at the latest 5min after each write. The file will then also
15378 be marked offline until the next write. This should increase
15379 reliability in case of a crash. The synchronization delay
15380 can be configured via SyncIntervalSec= in journald.conf.
15382 * There's a new remote-fs-setup.target unit that can be used
15383 to pull in specific services when at least one remote file
15384 system is to be mounted.
15386 * There are new targets timers.target and paths.target as
15387 canonical targets to pull user timer and path units in
15388 from. This complements sockets.target with a similar
15389 purpose for socket units.
15391 * libudev gained a new call udev_device_set_attribute_value()
15392 to set sysfs attributes of a device.
15394 * The udev daemon now sets the default number of worker
15395 processes executed in parallel based on the number of available
15396 CPUs instead of the amount of available RAM. This is supposed
15397 to provide a more reliable default and limit a too aggressive
15398 parallelism for setups with 1000s of devices connected.
15400 Contributions from: Auke Kok, Colin Walters, Cristian
15401 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
15402 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
15403 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
15404 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
15405 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
15406 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
15407 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
15408 Zbigniew Jędrzejewski-Szmek
15412 * Configuration of unit files may now be extended via drop-in
15413 files without having to edit/override the unit files
15414 themselves. More specifically, if the administrator wants to
15415 change one value for a service file foobar.service he can
15416 now do so by dropping in a configuration snippet into
15417 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
15418 will load all these snippets and apply them on top of the
15419 main unit configuration file, possibly extending or
15420 overriding its settings. Using these drop-in snippets is
15421 generally nicer than the two earlier options for changing
15422 unit files locally: copying the files from
15423 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
15424 them there; or creating a new file in /etc/systemd/system/
15425 that incorporates the original one via ".include". Drop-in
15426 snippets into these .d/ directories can be placed in any
15427 directory systemd looks for units in, and the usual
15428 overriding semantics between /usr/lib, /etc and /run apply
15431 * Most unit file settings which take lists of items can now be
15432 reset by assigning the empty string to them. For example,
15433 normally, settings such as Environment=FOO=BAR append a new
15434 environment variable assignment to the environment block,
15435 each time they are used. By assigning Environment= the empty
15436 string the environment block can be reset to empty. This is
15437 particularly useful with the .d/*.conf drop-in snippets
15438 mentioned above, since this adds the ability to reset list
15439 settings from vendor unit files via these drop-ins.
15441 * systemctl gained a new "list-dependencies" command for
15442 listing the dependencies of a unit recursively.
15444 * Inhibitors are now honored and listed by "systemctl
15445 suspend", "systemctl poweroff" (and similar) too, not only
15446 GNOME. These commands will also list active sessions by
15449 * Resource limits (as exposed by the various control group
15450 controllers) can now be controlled dynamically at runtime
15451 for all units. More specifically, you can now use a command
15452 like "systemctl set-cgroup-attr foobar.service cpu.shares
15453 2000" to alter the CPU shares a specific service gets. These
15454 settings are stored persistently on disk, and thus allow the
15455 administrator to easily adjust the resource usage of
15456 services with a few simple commands. This dynamic resource
15457 management logic is also available to other programs via the
15458 bus. Almost any kernel cgroup attribute and controller is
15461 * systemd-vconsole-setup will now copy all font settings to
15462 all allocated VTs, where it previously applied them only to
15465 * libsystemd-login gained the new sd_session_get_tty() API
15468 * This release drops support for a few legacy or
15469 distribution-specific LSB facility names when parsing init
15470 scripts: $x-display-manager, $mail-transfer-agent,
15471 $mail-transport-agent, $mail-transfer-agent, $smtp,
15472 $null. Also, the mail-transfer-agent.target unit backing
15473 this has been removed. Distributions which want to retain
15474 compatibility with this should carry the burden for
15475 supporting this themselves and patch support for these back
15476 in, if they really need to. Also, the facilities $syslog and
15477 $local_fs are now ignored, since systemd does not support
15478 early-boot LSB init scripts anymore, and these facilities
15479 are implied anyway for normal services. syslog.target has
15482 * There are new bus calls on PID1's Manager object for
15483 cancelling jobs, and removing snapshot units. Previously,
15484 both calls were only available on the Job and Snapshot
15485 objects themselves.
15487 * systemd-journal-gatewayd gained SSL support.
15489 * The various "environment" files, such as /etc/locale.conf
15490 now support continuation lines with a backslash ("\") as
15491 last character in the line, similarly in style (but different)
15492 to how this is supported in shells.
15494 * For normal user processes the _SYSTEMD_USER_UNIT= field is
15495 now implicitly appended to every log entry logged. systemctl
15496 has been updated to filter by this field when operating on a
15497 user systemd instance.
15499 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
15500 CAP_AUDIT_CONTROL capabilities to the capabilities set for
15501 the container. This makes it easier to boot unmodified
15502 Fedora systems in a container, which however still requires
15503 audit=0 to be passed on the kernel command line. Auditing in
15504 kernel and userspace is unfortunately still too broken in
15505 context of containers, hence we recommend compiling it out
15506 of the kernel or using audit=0. Hopefully this will be fixed
15507 one day for good in the kernel.
15509 * nspawn gained the new --bind= and --bind-ro= parameters to
15510 bind mount specific directories from the host into the
15513 * nspawn will now mount its own devpts file system instance
15514 into the container, in order not to leak pty devices from
15515 the host into the container.
15517 * systemd will now read the firmware boot time performance
15518 information from the EFI variables, if the used boot loader
15519 supports this, and takes it into account for boot performance
15520 analysis via "systemd-analyze". This is currently supported
15521 only in conjunction with Gummiboot, but could be supported
15522 by other boot loaders too. For details see:
15524 https://systemd.io/BOOT_LOADER_INTERFACE
15526 * A new generator has been added that automatically mounts the
15527 EFI System Partition (ESP) to /boot, if that directory
15528 exists, is empty, and no other file system has been
15529 configured to be mounted there.
15531 * logind will now send out PrepareForSleep(false) out
15532 unconditionally, after coming back from suspend. This may be
15533 used by applications as asynchronous notification for
15534 system resume events.
15536 * "systemctl unlock-sessions" has been added, that allows
15537 unlocking the screens of all user sessions at once, similar
15538 to how "systemctl lock-sessions" already locked all users
15539 sessions. This is backed by a new D-Bus call UnlockSessions().
15541 * "loginctl seat-status" will now show the master device of a
15542 seat. (i.e. the device of a seat that needs to be around for
15543 the seat to be considered available, usually the graphics
15546 * tmpfiles gained a new "X" line type, that allows
15547 configuration of files and directories (with wildcards) that
15548 shall be excluded from automatic cleanup ("aging").
15550 * udev default rules set the device node permissions now only
15551 at "add" events, and do not change them any longer with a
15552 later "change" event.
15554 * The log messages for lid events and power/sleep keypresses
15555 now carry a message ID.
15557 * We now have a substantially larger unit test suite, but this
15558 continues to be work in progress.
15560 * udevadm hwdb gained a new --root= parameter to change the
15561 root directory to operate relative to.
15563 * logind will now issue a background sync() request to the kernel
15564 early at shutdown, so that dirty buffers are flushed to disk early
15565 instead of at the last moment, in order to optimize shutdown
15568 * A new bootctl tool has been added that is an interface for
15569 certain boot loader operations. This is currently a preview
15570 and is likely to be extended into a small mechanism daemon
15571 like timedated, localed, hostnamed, and can be used by
15572 graphical UIs to enumerate available boot options, and
15573 request boot into firmware operations.
15575 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
15576 the rest of the package. It also has been updated to work
15577 correctly in initrds.
15579 * polkit previously has been runtime optional, and is now also
15580 compile time optional via a configure switch.
15582 * systemd-analyze has been reimplemented in C. Also "systemctl
15583 dot" has moved into systemd-analyze.
15585 * "systemctl status" with no further parameters will now print
15586 the status of all active or failed units.
15588 * Operations such as "systemctl start" can now be executed
15589 with a new mode "--irreversible" which may be used to queue
15590 operations that cannot accidentally be reversed by a later
15591 job queuing. This is by default used to make shutdown
15592 requests more robust.
15594 * The Python API of systemd now gained a new module for
15595 reading journal files.
15597 * A new tool kernel-install has been added that can install
15598 kernel images according to the Boot Loader Specification:
15600 https://systemd.io/BOOT_LOADER_SPECIFICATION
15602 * Boot time console output has been improved to provide
15603 animated boot time output for hanging jobs.
15605 * A new tool systemd-activate has been added which can be used
15606 to test socket activation with, directly from the command
15607 line. This should make it much easier to test and debug
15608 socket activation in daemons.
15610 * journalctl gained a new "--reverse" (or -r) option to show
15611 journal output in reverse order (i.e. newest line first).
15613 * journalctl gained a new "--pager-end" (or -e) option to jump
15614 to immediately jump to the end of the journal in the
15615 pager. This is only supported in conjunction with "less".
15617 * journalctl gained a new "--user-unit=" option, that works
15618 similarly to "--unit=" but filters for user units rather than
15621 * A number of unit files to ease adoption of systemd in
15622 initrds has been added. This moves some minimal logic from
15623 the various initrd implementations into systemd proper.
15625 * The journal files are now owned by a new group
15626 "systemd-journal", which exists specifically to allow access
15627 to the journal, and nothing else. Previously, we used the
15628 "adm" group for that, which however possibly covers more
15629 than just journal/log file access. This new group is now
15630 already used by systemd-journal-gatewayd to ensure this
15631 daemon gets access to the journal files and as little else
15632 as possible. Note that "make install" will also set FS ACLs
15633 up for /var/log/journal to give "adm" and "wheel" read
15634 access to it, in addition to "systemd-journal" which owns
15635 the journal files. We recommend that packaging scripts also
15636 add read access to "adm" + "wheel" to /var/log/journal, and
15637 all existing/future journal files. To normal users and
15638 administrators little changes, however packagers need to
15639 ensure to create the "systemd-journal" system group at
15640 package installation time.
15642 * The systemd-journal-gatewayd now runs as unprivileged user
15643 systemd-journal-gateway:systemd-journal-gateway. Packaging
15644 scripts need to create these system user/group at
15647 * timedated now exposes a new boolean property CanNTP that
15648 indicates whether a local NTP service is available or not.
15650 * systemd-detect-virt will now also detect xen PVs
15652 * The pstore file system is now mounted by default, if it is
15655 * In addition to the SELinux and IMA policies we will now also
15656 load SMACK policies at early boot.
15658 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
15659 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
15660 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
15661 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
15662 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
15663 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
15664 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
15665 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
15666 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
15667 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
15668 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
15669 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
15670 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
15671 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
15675 * Timer units now support calendar time events in addition to
15676 monotonic time events. That means you can now trigger a unit
15677 based on a calendar time specification such as "Thu,Fri
15678 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
15679 or fifth day of any month of the year 2013, given that it is
15680 a Thursday or a Friday. This brings timer event support
15681 considerably closer to cron's capabilities. For details on
15682 the supported calendar time specification language see
15685 * udev now supports a number of different naming policies for
15686 network interfaces for predictable names, and a combination
15687 of these policies is now the default. Please see this wiki
15688 document for details:
15690 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
15692 * Auke Kok's bootchart implementation has been added to the
15693 systemd tree. It is an optional component that can graph the
15694 boot in quite some detail. It is one of the best bootchart
15695 implementations around and minimal in its code and
15698 * nss-myhostname has been integrated into the systemd source
15699 tree. nss-myhostname guarantees that the local hostname
15700 always stays resolvable via NSS. It has been a weak
15701 requirement of systemd-hostnamed since a long time, and
15702 since its code is actually trivial we decided to just
15703 include it in systemd's source tree. It can be turned off
15704 with a configure switch.
15706 * The read-ahead logic is now capable of properly detecting
15707 whether a btrfs file system is on SSD or rotating media, in
15708 order to optimize the read-ahead scheme. Previously, it was
15709 only capable of detecting this on traditional file systems
15712 * In udev, additional device properties are now read from the
15713 IAB in addition to the OUI database. Also, Bluetooth company
15714 identities are attached to the devices as well.
15716 * In service files %U may be used as specifier that is
15717 replaced by the configured user name of the service.
15719 * nspawn may now be invoked without a controlling TTY. This
15720 makes it suitable for invocation as its own service. This
15721 may be used to set up a simple containerized server system
15722 using only core OS tools.
15724 * systemd and nspawn can now accept socket file descriptors
15725 when they are started for socket activation. This enables
15726 implementation of socket activated nspawn
15727 containers. i.e. think about autospawning an entire OS image
15728 when the first SSH or HTTP connection is received. We expect
15729 that similar functionality will also be added to libvirt-lxc
15732 * journalctl will now suppress ANSI color codes when
15733 presenting log data.
15735 * systemctl will no longer show control group information for
15736 a unit if the control group is empty anyway.
15738 * logind can now automatically suspend/hibernate/shutdown the
15741 * /etc/machine-info and hostnamed now also expose the chassis
15742 type of the system. This can be used to determine whether
15743 the local system is a laptop, desktop, handset or
15744 tablet. This information may either be configured by the
15745 user/vendor or is automatically determined from ACPI and DMI
15746 information if possible.
15748 * A number of polkit actions are now bound together with "imply"
15749 rules. This should simplify creating UIs because many actions
15750 will now authenticate similar ones as well.
15752 * Unit files learnt a new condition ConditionACPower= which
15753 may be used to conditionalize a unit depending on whether an
15754 AC power source is connected or not, of whether the system
15755 is running on battery power.
15757 * systemctl gained a new "is-failed" verb that may be used in
15758 shell scripts and suchlike to check whether a specific unit
15759 is in the "failed" state.
15761 * The EnvironmentFile= setting in unit files now supports file
15762 globbing, and can hence be used to easily read a number of
15763 environment files at once.
15765 * systemd will no longer detect and recognize specific
15766 distributions. All distribution-specific #ifdeffery has been
15767 removed, systemd is now fully generic and
15768 distribution-agnostic. Effectively, not too much is lost as
15769 a lot of the code is still accessible via explicit configure
15770 switches. However, support for some distribution specific
15771 legacy configuration file formats has been dropped. We
15772 recommend distributions to simply adopt the configuration
15773 files everybody else uses now and convert the old
15774 configuration from packaging scripts. Most distributions
15775 already did that. If that's not possible or desirable,
15776 distributions are welcome to forward port the specific
15777 pieces of code locally from the git history.
15779 * When logging a message about a unit systemd will now always
15780 log the unit name in the message meta data.
15782 * localectl will now also discover system locale data that is
15783 not stored in locale archives, but directly unpacked.
15785 * logind will no longer unconditionally use framebuffer
15786 devices as seat masters, i.e. as devices that are required
15787 to be existing before a seat is considered preset. Instead,
15788 it will now look for all devices that are tagged as
15789 "seat-master" in udev. By default, framebuffer devices will
15790 be marked as such, but depending on local systems, other
15791 devices might be marked as well. This may be used to
15792 integrate graphics cards using closed source drivers (such
15793 as NVidia ones) more nicely into logind. Note however, that
15794 we recommend using the open source NVidia drivers instead,
15795 and no udev rules for the closed-source drivers will be
15796 shipped from us upstream.
15798 Contributions from: Adam Williamson, Alessandro Crismani, Auke
15799 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
15800 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
15801 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
15802 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
15803 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
15804 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
15805 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
15806 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
15807 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
15808 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
15809 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
15814 * udev gained support for loading additional device properties
15815 from an indexed database that is keyed by vendor/product IDs
15816 and similar device identifiers. For the beginning this
15817 "hwdb" is populated with data from the well-known PCI and
15818 USB database, but also includes PNP, ACPI and OID data. In
15819 the longer run this indexed database shall grow into
15820 becoming the one central database for non-essential
15821 userspace device metadata. Previously, data from the PCI/USB
15822 database was only attached to select devices, since the
15823 lookup was a relatively expensive operation due to O(n) time
15824 complexity (with n being the number of entries in the
15825 database). Since this is now O(1), we decided to add in this
15826 data for all devices where this is available, by
15827 default. Note that the indexed database needs to be rebuilt
15828 when new data files are installed. To achieve this you need
15829 to update your packaging scripts to invoke "udevadm hwdb
15830 --update" after installation of hwdb data files. For
15831 RPM-based distributions we introduced the new
15832 %udev_hwdb_update macro for this purpose.
15834 * The Journal gained support for the "Message Catalog", an
15835 indexed database to link up additional information with
15836 journal entries. For further details please check:
15838 https://www.freedesktop.org/wiki/Software/systemd/catalog
15840 The indexed message catalog database also needs to be
15841 rebuilt after installation of message catalog files. Use
15842 "journalctl --update-catalog" for this. For RPM-based
15843 distributions we introduced the %journal_catalog_update
15844 macro for this purpose.
15846 * The Python Journal bindings gained support for the standard
15847 Python logging framework.
15849 * The Journal API gained new functions for checking whether
15850 the underlying file system of a journal file is capable of
15851 properly reporting file change notifications, or whether
15852 applications that want to reflect journal changes "live"
15853 need to recheck journal files continuously in appropriate
15856 * It is now possible to set the "age" field for tmpfiles
15857 entries to 0, indicating that files matching this entry
15858 shall always be removed when the directories are cleaned up.
15860 * coredumpctl gained a new "gdb" verb which invokes gdb
15861 right-away on the selected coredump.
15863 * There's now support for "hybrid sleep" on kernels that
15864 support this, in addition to "suspend" and "hibernate". Use
15865 "systemctl hybrid-sleep" to make use of this.
15867 * logind's HandleSuspendKey= setting (and related settings)
15868 now gained support for a new "lock" setting to simply
15869 request the screen lock on all local sessions, instead of
15870 actually executing a suspend or hibernation.
15872 * systemd will now mount the EFI variables file system by
15875 * Socket units now gained support for configuration of the
15876 SMACK security label.
15878 * timedatectl will now output the time of the last and next
15879 daylight saving change.
15881 * We dropped support for various legacy and distro-specific
15882 concepts, such as insserv, early-boot SysV services
15883 (i.e. those for non-standard runlevels such as 'b' or 'S')
15884 or ArchLinux /etc/rc.conf support. We recommend the
15885 distributions who still need support this to either continue
15886 to maintain the necessary patches downstream, or find a
15887 different solution. (Talk to us if you have questions!)
15889 * Various systemd components will now bypass polkit checks for
15890 root and otherwise handle properly if polkit is not found to
15891 be around. This should fix most issues for polkit-less
15892 systems. Quite frankly this should have been this way since
15893 day one. It is absolutely our intention to make systemd work
15894 fine on polkit-less systems, and we consider it a bug if
15895 something does not work as it should if polkit is not around.
15897 * For embedded systems it is now possible to build udev and
15898 systemd without blkid and/or kmod support.
15900 * "systemctl switch-root" is now capable of switching root
15901 more than once. I.e. in addition to transitions from the
15902 initrd to the host OS it is now possible to transition to
15903 further OS images from the host. This is useful to implement
15904 offline updating tools.
15906 * Various other additions have been made to the RPM macros
15907 shipped with systemd. Use %udev_rules_update() after
15908 installing new udev rules files. %_udevhwdbdir,
15909 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
15910 %_sysctldir are now available which resolve to the right
15911 directories for packages to place various data files in.
15913 * journalctl gained the new --full switch (in addition to
15914 --all, to disable ellipsation for long messages.
15916 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
15917 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
15918 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
15919 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
15920 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
15921 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
15922 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
15923 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
15924 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
15928 * journalctl gained new --since= and --until= switches to
15929 filter by time. It also now supports nice filtering for
15930 units via --unit=/-u.
15932 * Type=oneshot services may use ExecReload= and do the
15935 * The journal daemon now supports time-based rotation and
15936 vacuuming, in addition to the usual disk-space based
15939 * The journal will now index the available field values for
15940 each field name. This enables clients to show pretty drop
15941 downs of available match values when filtering. The bash
15942 completion of journalctl has been updated
15943 accordingly. journalctl gained a new switch -F to list all
15944 values a certain field takes in the journal database.
15946 * More service events are now written as structured messages
15947 to the journal, and made recognizable via message IDs.
15949 * The timedated, localed and hostnamed mini-services which
15950 previously only provided support for changing time, locale
15951 and hostname settings from graphical DEs such as GNOME now
15952 also have a minimal (but very useful) text-based client
15953 utility each. This is probably the nicest way to changing
15954 these settings from the command line now, especially since
15955 it lists available options and is fully integrated with bash
15958 * There's now a new tool "systemd-coredumpctl" to list and
15959 extract coredumps from the journal.
15961 * We now install a README each in /var/log/ and
15962 /etc/rc.d/init.d explaining where the system logs and init
15963 scripts went. This hopefully should help folks who go to
15964 that dirs and look into the otherwise now empty void and
15965 scratch their heads.
15967 * When user-services are invoked (by systemd --user) the
15968 $MANAGERPID env var is set to the PID of systemd.
15970 * SIGRTMIN+24 when sent to a --user instance will now result
15971 in immediate termination of systemd.
15973 * gatewayd received numerous feature additions such as a
15974 "follow" mode, for live syncing and filtering.
15976 * browse.html now allows filtering and showing detailed
15977 information on specific entries. Keyboard navigation and
15978 mouse screen support has been added.
15980 * gatewayd/journalctl now supports HTML5/JSON
15981 Server-Sent-Events as output.
15983 * The SysV init script compatibility logic will now
15984 heuristically determine whether a script supports the
15985 "reload" verb, and only then make this available as
15986 "systemctl reload".
15988 * "systemctl status --follow" has been removed, use "journalctl
15991 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
15992 have been removed since they are hardly useful to be
15995 * And I'd like to take the opportunity to specifically mention
15996 Zbigniew for his great contributions. Zbigniew, you rock!
15998 Contributions from: Andrew Eikum, Christian Hesse, Colin
15999 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
16000 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
16001 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
16002 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
16003 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
16004 Jędrzejewski-Szmek, Сковорода Никита Андреевич
16008 * If /etc/vconsole.conf is non-existent or empty we will no
16009 longer load any console font or key map at boot by
16010 default. Instead the kernel defaults will be left
16011 intact. This is definitely the right thing to do, as no
16012 configuration should mean no configuration, and hard-coding
16013 font names that are different on all archs is probably a bad
16014 idea. Also, the kernel default key map and font should be
16015 good enough for most cases anyway, and mostly identical to
16016 the userspace fonts/key maps we previously overloaded them
16017 with. If distributions want to continue to default to a
16018 non-kernel font or key map they should ship a default
16019 /etc/vconsole.conf with the appropriate contents.
16021 Contributions from: Colin Walters, Daniel J Walsh, Dave
16022 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
16023 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
16027 * journalctl gained a new --cursor= switch to show entries
16028 starting from the specified location in the journal.
16030 * We now enforce a size limit on journal entry fields exported
16031 with "-o json" in journalctl. Fields larger than 4K will be
16032 assigned null. This can be turned off with --all.
16034 * An (optional) journal gateway daemon is now available as
16035 "systemd-journal-gatewayd.service". This service provides
16036 access to the journal via HTTP and JSON. This functionality
16037 will be used to implement live log synchronization in both
16038 pull and push modes, but has various other users too, such
16039 as easy log access for debugging of embedded devices. Right
16040 now it is already useful to retrieve the journal via HTTP:
16042 # systemctl start systemd-journal-gatewayd.service
16043 # wget http://localhost:19531/entries
16045 This will download the journal contents in a
16046 /var/log/messages compatible format. The same as JSON:
16048 # curl -H"Accept: application/json" http://localhost:19531/entries
16050 This service is also accessible via a web browser where a
16051 single static HTML5 app is served that uses the JSON logic
16052 to enable the user to do some basic browsing of the
16053 journal. This will be extended later on. Here's an example
16054 screenshot of this app in its current state:
16056 https://0pointer.de/public/journal-gatewayd
16058 Contributions from: Kay Sievers, Lennart Poettering, Robert
16059 Milasan, Tom Gundersen
16063 * The bash completion logic is now available for journalctl
16066 * We do not mount the "cpuset" controller anymore together with
16067 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
16068 started if no parameters are assigned to it. "cpuset" hence
16069 broke code that assumed it could create "cpu" groups and
16072 * journalctl -f will now subscribe to terminal size changes,
16073 and line break accordingly.
16075 Contributions from: Dave Reisner, Kay Sievers, Lennart
16076 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
16080 * nspawn will now create a symlink /etc/localtime in the
16081 container environment, copying the host's timezone
16082 setting. Previously this has been done via a bind mount, but
16083 since symlinks cannot be bind mounted this has now been
16084 changed to create/update the appropriate symlink.
16086 * journalctl -n's line number argument is now optional, and
16087 will default to 10 if omitted.
16089 * journald will now log the maximum size the journal files may
16090 take up on disk. This is particularly useful if the default
16091 built-in logic of determining this parameter from the file
16092 system size is used. Use "systemctl status
16093 systemd-journald.service" to see this information.
16095 * The multi-seat X wrapper tool has been stripped down. As X
16096 is now capable of enumerating graphics devices via udev in a
16097 seat-aware way the wrapper is not strictly necessary
16098 anymore. A stripped down temporary stop-gap is still shipped
16099 until the upstream display managers have been updated to
16100 fully support the new X logic. Expect this wrapper to be
16101 removed entirely in one of the next releases.
16103 * HandleSleepKey= in logind.conf has been split up into
16104 HandleSuspendKey= and HandleHibernateKey=. The old setting
16105 is not available anymore. X11 and the kernel are
16106 distinguishing between these keys and we should too. This
16107 also means the inhibition lock for these keys has been split
16110 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
16111 Poettering, Lukas Nykryn, Václav Pavlín
16115 * Whenever a unit changes state we will now log this to the
16116 journal and show along the unit's own log output in
16117 "systemctl status".
16119 * ConditionPathIsMountPoint= can now properly detect bind
16120 mount points too. (Previously, a bind mount of one file
16121 system to another place in the same file system could not be
16122 detected as mount, since they shared struct stat's st_dev
16125 * We will now mount the cgroup controllers cpu, cpuacct,
16126 cpuset and the controllers net_cls, net_prio together by
16129 * nspawn containers will now have a virtualized boot
16130 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
16131 over with a randomized ID at container initialization). This
16132 has the effect of making "journalctl -b" do the right thing
16135 * The JSON output journal serialization has been updated not
16136 to generate "endless" list objects anymore, but rather one
16137 JSON object per line. This is more in line how most JSON
16138 parsers expect JSON objects. The new output mode
16139 "json-pretty" has been added to provide similar output, but
16140 neatly aligned for readability by humans.
16142 * We dropped all explicit sync() invocations in the shutdown
16143 code. The kernel does this implicitly anyway in the kernel
16144 reboot() syscall. halt(8)'s -n option is now a compatibility
16147 * We now support virtualized reboot() in containers, as
16148 supported by newer kernels. We will fall back to exit() if
16149 CAP_SYS_REBOOT is not available to the container. Also,
16150 nspawn makes use of this now and will actually reboot the
16151 container if the containerized OS asks for that.
16153 * journalctl will only show local log output by default
16154 now. Use --merge (-m) to show remote log output, too.
16156 * libsystemd-journal gained the new sd_journal_get_usage()
16157 call to determine the current disk usage of all journal
16158 files. This is exposed in the new "journalctl --disk-usage"
16161 * journald gained a new configuration setting SplitMode= in
16162 journald.conf which may be used to control how user journals
16163 are split off. See journald.conf(5) for details.
16165 * A new condition type ConditionFileNotEmpty= has been added.
16167 * tmpfiles' "w" lines now support file globbing, to write
16168 multiple files at once.
16170 * We added Python bindings for the journal submission
16171 APIs. More Python APIs for a number of selected APIs will
16172 likely follow. Note that we intend to add native bindings
16173 only for the Python language, as we consider it common
16174 enough to deserve bindings shipped within systemd. There are
16175 various projects outside of systemd that provide bindings
16176 for languages such as PHP or Lua.
16178 * Many conditions will now resolve specifiers such as %i. In
16179 addition, PathChanged= and related directives of .path units
16180 now support specifiers as well.
16182 * There's now a new RPM macro definition for the system preset
16185 * journald will now warn if it ca not forward a message to the
16186 syslog daemon because its socket is full.
16188 * timedated will no longer write or process /etc/timezone,
16189 except on Debian. As we do not support late mounted /usr
16190 anymore /etc/localtime always being a symlink is now safe,
16191 and hence the information in /etc/timezone is not necessary
16194 * logind will now always reserve one VT for a text getty (VT6
16195 by default). Previously if more than 6 X sessions where
16196 started they took up all the VTs with auto-spawned gettys,
16197 so that no text gettys were available anymore.
16199 * udev will now automatically inform the btrfs kernel logic
16200 about btrfs RAID components showing up. This should make
16201 simple hotplug based btrfs RAID assembly work.
16203 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
16204 (but not for its children which will stay at the kernel
16205 default). This should allow setups with a lot more listening
16208 * systemd will now always pass the configured timezone to the
16209 kernel at boot. timedated will do the same when the timezone
16212 * logind's inhibition logic has been updated. By default,
16213 logind will now handle the lid switch, the power and sleep
16214 keys all the time, even in graphical sessions. If DEs want
16215 to handle these events on their own they should take the new
16216 handle-power-key, handle-sleep-key and handle-lid-switch
16217 inhibitors during their runtime. A simple way to achieve
16218 that is to invoke the DE wrapped in an invocation of:
16220 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
16222 * Access to unit operations is now checked via SELinux taking
16223 the unit file label and client process label into account.
16225 * systemd will now notify the administrator in the journal
16226 when he over-mounts a non-empty directory.
16228 * There are new specifiers that are resolved in unit files,
16229 for the hostname (%H), the machine ID (%m) and the boot ID
16232 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
16233 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
16234 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
16235 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
16236 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
16237 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
16238 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
16242 * Support for reading structured kernel messages from
16243 /dev/kmsg has now been added and is enabled by default.
16245 * Support for reading kernel messages from /proc/kmsg has now
16246 been removed. If you want kernel messages in the journal
16247 make sure to run a recent kernel (>= 3.5) that supports
16248 reading structured messages from /dev/kmsg (see
16249 above). /proc/kmsg is now exclusive property of classic
16250 syslog daemons again.
16252 * The libudev API gained the new
16253 udev_device_new_from_device_id() call.
16255 * The logic for file system namespace (ReadOnlyDirectory=,
16256 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
16257 require pivot_root() anymore. This means fewer temporary
16258 directories are created below /tmp for this feature.
16260 * nspawn containers will now see and receive all submounts
16261 made on the host OS below the root file system of the
16264 * Forward Secure Sealing is now supported for Journal files,
16265 which provide cryptographical sealing of journal files so
16266 that attackers cannot alter log history anymore without this
16267 being detectable. Lennart will soon post a blog story about
16268 this explaining it in more detail.
16270 * There are two new service settings RestartPreventExitStatus=
16271 and SuccessExitStatus= which allow configuration of exit
16272 status (exit code or signal) which will be excepted from the
16273 restart logic, resp. consider successful.
16275 * journalctl gained the new --verify switch that can be used
16276 to check the integrity of the structure of journal files and
16277 (if Forward Secure Sealing is enabled) the contents of
16280 * nspawn containers will now be run with /dev/stdin, /dev/fd/
16281 and similar symlinks pre-created. This makes running shells
16282 as container init process a lot more fun.
16284 * The fstab support can now handle PARTUUID= and PARTLABEL=
16287 * A new ConditionHost= condition has been added to match
16288 against the hostname (with globs) and machine ID. This is
16289 useful for clusters where a single OS image is used to
16290 provision a large number of hosts which shall run slightly
16291 different sets of services.
16293 * Services which hit the restart limit will now be placed in a
16296 Contributions from: Bertram Poettering, Dave Reisner, Huang
16297 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
16298 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
16302 * When running in --user mode systemd will now become a
16303 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
16304 tree a lot more organized.
16306 * A new PartOf= unit dependency type has been introduced that
16307 may be used to group services in a natural way.
16309 * "systemctl enable" may now be used to enable instances of
16312 * journalctl now prints error log levels in red, and
16313 warning/notice log levels in bright white. It also supports
16314 filtering by log level now.
16316 * cgtop gained a new -n switch (similar to top), to configure
16317 the maximum number of iterations to run for. It also gained
16318 -b, to run in batch mode (accepting no input).
16320 * The suffix ".service" may now be omitted on most systemctl
16321 command lines involving service unit names.
16323 * There's a new bus call in logind to lock all sessions, as
16324 well as a loginctl verb for it "lock-sessions".
16326 * libsystemd-logind.so gained a new call sd_journal_perror()
16327 that works similar to libc perror() but logs to the journal
16328 and encodes structured information about the error number.
16330 * /etc/crypttab entries now understand the new keyfile-size=
16333 * shutdown(8) now can send a (configurable) wall message when
16334 a shutdown is cancelled.
16336 * The mount propagation mode for the root file system will now
16337 default to "shared", which is useful to make containers work
16338 nicely out-of-the-box so that they receive new mounts from
16339 the host. This can be undone locally by running "mount
16340 --make-rprivate /" if needed.
16342 * The prefdm.service file has been removed. Distributions
16343 should maintain this unit downstream if they intend to keep
16344 it around. However, we recommend writing normal unit files
16345 for display managers instead.
16347 * Since systemd is a crucial part of the OS we will now
16348 default to a number of compiler switches that improve
16349 security (hardening) such as read-only relocations, stack
16350 protection, and suchlike.
16352 * The TimeoutSec= setting for services is now split into
16353 TimeoutStartSec= and TimeoutStopSec= to allow configuration
16354 of individual time outs for the start and the stop phase of
16357 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
16358 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
16359 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
16360 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
16361 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
16362 Gundersen, Zbigniew Jędrzejewski-Szmek
16366 * The journal and id128 C APIs are now fully documented as man
16369 * Extra safety checks have been added when transitioning from
16370 the initial RAM disk to the main system to avoid accidental
16373 * /etc/crypttab entries now understand the new keyfile-offset=
16376 * systemctl -t can now be used to filter by unit load state.
16378 * The journal C API gained the new sd_journal_wait() call to
16379 make writing synchronous journal clients easier.
16381 * journalctl gained the new -D switch to show journals from a
16382 specific directory.
16384 * journalctl now displays a special marker between log
16385 messages of two different boots.
16387 * The journal is now explicitly flushed to /var via a service
16388 systemd-journal-flush.service, rather than implicitly simply
16389 by seeing /var/log/journal to be writable.
16391 * journalctl (and the journal C APIs) can now match for much
16392 more complex expressions, with alternatives and
16395 * When transitioning from the initial RAM disk to the main
16396 system we will now kill all processes in a killing spree to
16397 ensure no processes stay around by accident.
16399 * Three new specifiers may be used in unit files: %u, %h, %s
16400 resolve to the user name, user home directory resp. user
16401 shell. This is useful for running systemd user instances.
16403 * We now automatically rotate journal files if their data
16404 object hash table gets a fill level > 75%. We also size the
16405 hash table based on the configured maximum file size. This
16406 together should lower hash collisions drastically and thus
16407 speed things up a bit.
16409 * journalctl gained the new "--header" switch to introspect
16410 header data of journal files.
16412 * A new setting SystemCallFilters= has been added to services which may
16413 be used to apply deny lists or allow lists to system calls. This is
16414 based on SECCOMP Mode 2 of Linux 3.5.
16416 * nspawn gained a new --link-journal= switch (and quicker: -j)
16417 to link the container journal with the host. This makes it
16418 very easy to centralize log viewing on the host for all
16419 guests while still keeping the journal files separated.
16421 * Many bugfixes and optimizations
16423 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
16424 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
16425 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
16430 * Several tools now understand kernel command line arguments,
16431 which are only read when run in an initial RAM disk. They
16432 usually follow closely their normal counterparts, but are
16435 * There's a new tool to analyze the readahead files that are
16436 automatically generated at boot. Use:
16438 /usr/lib/systemd/systemd-readahead analyze /.readahead
16440 * We now provide an early debug shell on tty9 if this enabled. Use:
16442 systemctl enable debug-shell.service
16444 * All plymouth related units have been moved into the Plymouth
16445 package. Please make sure to upgrade your Plymouth version
16448 * systemd-tmpfiles now supports getting passed the basename of
16449 a configuration file only, in which case it will look for it
16450 in all appropriate directories automatically.
16452 * udevadm info now takes a /dev or /sys path as argument, and
16453 does the right thing. Example:
16455 udevadm info /dev/sda
16456 udevadm info /sys/class/block/sda
16458 * systemctl now prints a warning if a unit is stopped but a
16459 unit that might trigger it continues to run. Example: a
16460 service is stopped but the socket that activates it is left
16463 * "systemctl status" will now mention if the log output was
16464 shortened due to rotation since a service has been started.
16466 * The journal API now exposes functions to determine the
16467 "cutoff" times due to rotation.
16469 * journald now understands SIGUSR1 and SIGUSR2 for triggering
16470 immediately flushing of runtime logs to /var if possible,
16471 resp. for triggering immediate rotation of the journal
16474 * It is now considered an error if a service is attempted to
16475 be stopped that is not loaded.
16477 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
16479 * systemd-analyze now supports Python 3
16481 * tmpfiles now supports cleaning up directories via aging
16482 where the first level dirs are always kept around but
16483 directories beneath it automatically aged. This is enabled
16484 by prefixing the age field with '~'.
16486 * Seat objects now expose CanGraphical, CanTTY properties
16487 which is required to deal with very fast bootups where the
16488 display manager might be running before the graphics drivers
16489 completed initialization.
16491 * Seat objects now expose a State property.
16493 * We now include RPM macros for service enabling/disabling
16494 based on the preset logic. We recommend RPM based
16495 distributions to make use of these macros if possible. This
16496 makes it simpler to reuse RPM spec files across
16499 * We now make sure that the collected systemd unit name is
16500 always valid when services log to the journal via
16503 * There's a new man page kernel-command-line(7) detailing all
16504 command line options we understand.
16506 * The fstab generator may now be disabled at boot by passing
16507 fstab=0 on the kernel command line.
16509 * A new kernel command line option modules-load= is now understood
16510 to load a specific kernel module statically, early at boot.
16512 * Unit names specified on the systemctl command line are now
16513 automatically escaped as needed. Also, if file system or
16514 device paths are specified they are automatically turned
16515 into the appropriate mount or device unit names. Example:
16517 systemctl status /home
16518 systemctl status /dev/sda
16520 * The SysVConsole= configuration option has been removed from
16521 system.conf parsing.
16523 * The SysV search path is no longer exported on the D-Bus
16526 * The Names= option has been removed from unit file parsing.
16528 * There's a new man page bootup(7) detailing the boot process.
16530 * Every unit and every generator we ship with systemd now
16531 comes with full documentation. The self-explanatory boot is
16534 * A couple of services gained "systemd-" prefixes in their
16535 name if they wrap systemd code, rather than only external
16536 code. Among them fsck@.service which is now
16537 systemd-fsck@.service.
16539 * The HaveWatchdog property has been removed from the D-Bus
16542 * systemd.confirm_spawn= on the kernel command line should now
16545 * There's a new man page crypttab(5) which details all options
16546 we actually understand.
16548 * systemd-nspawn gained a new --capability= switch to pass
16549 additional capabilities to the container.
16551 * timedated will now read known NTP implementation unit names
16552 from /usr/lib/systemd/ntp-units.d/*.list,
16553 systemd-timedated-ntp.target has been removed.
16555 * journalctl gained a new switch "-b" that lists log data of
16556 the current boot only.
16558 * The notify socket is in the abstract namespace again, in
16559 order to support daemons which chroot() at start-up.
16561 * There is a new Storage= configuration option for journald
16562 which allows configuration of where log data should go. This
16563 also provides a way to disable journal logging entirely, so
16564 that data collected is only forwarded to the console, the
16565 kernel log buffer or another syslog implementation.
16567 * Many bugfixes and optimizations
16569 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
16570 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
16571 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
16572 Shawn Landden, Tom Gundersen
16576 * "systemctl help <unit>" now shows the man page if one is
16579 * Several new man pages have been added.
16581 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
16582 MaxLevelConsole= can now be specified in
16583 journald.conf. These options allow reducing the amount of
16584 data stored on disk or forwarded by the log level.
16586 * TimerSlackNSec= can now be specified in system.conf for
16587 PID1. This allows system-wide power savings.
16589 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
16590 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
16595 * logind is now capable of (optionally) handling power and
16596 sleep keys as well as the lid switch.
16598 * journalctl now understands the syntax "journalctl
16599 /usr/bin/avahi-daemon" to get all log output of a specific
16602 * CapabilityBoundingSet= in system.conf now also influences
16603 the capability bound set of usermode helpers of the kernel.
16605 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
16606 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
16607 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
16608 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
16612 * Note that we skipped 139 releases here in order to set the
16613 new version to something that is greater than both udev's
16614 and systemd's most recent version number.
16616 * udev: all udev sources are merged into the systemd source tree now.
16617 All future udev development will happen in the systemd tree. It
16618 is still fully supported to use the udev daemon and tools without
16619 systemd running, like in initramfs or other init systems. Building
16620 udev though, will require the *build* of the systemd tree, but
16621 udev can be properly *run* without systemd.
16623 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
16624 should be used to create dead device nodes as workarounds for broken
16627 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
16628 no longer supported. udev_monitor_new_from_netlink() needs to be
16629 used to subscribe to events.
16631 * udev: when udevd is started by systemd, processes which are left
16632 behind by forking them off of udev rules, are unconditionally cleaned
16633 up and killed now after the event handling has finished. Services or
16634 daemons must be started as systemd services. Services can be
16635 pulled-in by udev to get started, but they can no longer be directly
16636 forked by udev rules.
16638 * udev: the daemon binary is called systemd-udevd now and installed
16639 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
16640 to adapt to that, create symlink, or rename the binary after building
16643 * libudev no longer provides these symbols:
16644 udev_monitor_from_socket()
16645 udev_queue_get_failed_list_entry()
16646 udev_get_{dev,sys,run}_path()
16647 The versions number was bumped and symbol versioning introduced.
16649 * systemd-loginctl and systemd-journalctl have been renamed
16650 to loginctl and journalctl to match systemctl.
16652 * The config files: /etc/systemd/systemd-logind.conf and
16653 /etc/systemd/systemd-journald.conf have been renamed to
16654 logind.conf and journald.conf. Package updates should rename
16655 the files to the new names on upgrade.
16657 * For almost all files the license is now LGPL2.1+, changed
16658 from the previous GPL2.0+. Exceptions are some minor stuff
16659 of udev (which will be changed to LGPL2.1 eventually, too),
16660 and the MIT licensed sd-daemon.[ch] library that is suitable
16661 to be used as drop-in files.
16663 * systemd and logind now handle system sleep states, in
16664 particular suspending and hibernating.
16666 * logind now implements a sleep/shutdown/idle inhibiting logic
16667 suitable for a variety of uses. Soonishly Lennart will blog
16668 about this in more detail.
16670 * var-run.mount and var-lock.mount are no longer provided
16671 (which previously bind mounted these directories to their new
16672 places). Distributions which have not converted these
16673 directories to symlinks should consider stealing these files
16674 from git history and add them downstream.
16676 * We introduced the Documentation= field for units and added
16677 this to all our shipped units. This is useful to make it
16678 easier to explore the boot and the purpose of the various
16681 * All smaller setup units (such as
16682 systemd-vconsole-setup.service) now detect properly if they
16683 are run in a container and are skipped when
16684 appropriate. This guarantees an entirely noise-free boot in
16685 Linux container environments such as systemd-nspawn.
16687 * A framework for implementing offline system updates is now
16688 integrated, for details see:
16689 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
16691 * A new service type Type=idle is available now which helps us
16692 avoiding ugly interleaving of getty output and boot status
16695 * There's now a system-wide CapabilityBoundingSet= option to
16696 globally reduce the set of capabilities for the
16697 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
16698 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
16699 even CAP_NET_ADMIN system-wide for secure systems.
16701 * There are now system-wide DefaultLimitXXX= options to
16702 globally change the defaults of the various resource limits
16703 for all units started by PID 1.
16705 * Harald Hoyer's systemd test suite has been integrated into
16706 systemd which allows easy testing of systemd builds in qemu
16707 and nspawn. (This is really awesome! Ask us for details!)
16709 * The fstab parser is now implemented as generator, not inside
16712 * systemctl will now warn you if .mount units generated from
16713 /etc/fstab are out of date due to changes in fstab that
16714 have not been read by systemd yet.
16716 * systemd is now suitable for usage in initrds. Dracut has
16717 already been updated to make use of this. With this in place
16718 initrds get a slight bit faster but primarily are much
16719 easier to introspect and debug since "systemctl status" in
16720 the host system can be used to introspect initrd services,
16721 and the journal from the initrd is kept around too.
16723 * systemd-delta has been added, a tool to explore differences
16724 between user/admin configuration and vendor defaults.
16726 * PrivateTmp= now affects both /tmp and /var/tmp.
16728 * Boot time status messages are now much prettier and feature
16729 proper english language. Booting up systemd has never been
16732 * Read-ahead pack files now include the inode number of all
16733 files to pre-cache. When the inode changes the pre-caching
16734 is not attempted. This should be nicer to deal with updated
16735 packages which might result in changes of read-ahead
16738 * We now temporaritly lower the kernel's read_ahead_kb variable
16739 when collecting read-ahead data to ensure the kernel's
16740 built-in read-ahead does not add noise to our measurements
16741 of necessary blocks to pre-cache.
16743 * There's now RequiresMountsFor= to add automatic dependencies
16744 for all mounts necessary for a specific file system path.
16746 * MountAuto= and SwapAuto= have been removed from
16747 system.conf. Mounting file systems at boot has to take place
16750 * nspawn now learned a new switch --uuid= to set the machine
16751 ID on the command line.
16753 * nspawn now learned the -b switch to automatically search
16754 for an init system.
16756 * vt102 is now the default TERM for serial TTYs, upgraded from
16759 * systemd-logind now works on VT-less systems.
16761 * The build tree has been reorganized. The individual
16762 components now have directories of their own.
16764 * A new condition type ConditionPathIsReadWrite= is now available.
16766 * nspawn learned the new -C switch to create cgroups for the
16767 container in other hierarchies.
16769 * We now have support for hardware watchdogs, configurable in
16772 * The scheduled shutdown logic now has a public API.
16774 * We now mount /tmp as tmpfs by default, but this can be
16775 masked and /etc/fstab can override it.
16777 * Since udisks does not make use of /media anymore we are not
16778 mounting a tmpfs on it anymore.
16780 * journalctl gained a new --local switch to only interleave
16781 locally generated journal files.
16783 * We can now load the IMA policy at boot automatically.
16785 * The GTK tools have been split off into a systemd-ui.
16787 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
16788 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
16789 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
16790 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
16791 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
16792 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
16793 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
16794 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
16795 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
16800 * This is mostly a bugfix release
16802 * Support optional initialization of the machine ID from the
16803 KVM or container configured UUID.
16805 * Support immediate reboots with "systemctl reboot -ff"
16807 * Show /etc/os-release data in systemd-analyze output
16809 * Many bugfixes for the journal, including endianness fixes and
16810 ensuring that disk space enforcement works
16812 * sd-login.h is C++ compatible again
16814 * Extend the /etc/os-release format on request of the Debian
16817 * We now refuse non-UTF8 strings used in various configuration
16818 and unit files. This is done to ensure we do not pass invalid
16819 data over D-Bus or expose it elsewhere.
16821 * Register Mimo USB Screens as suitable for automatic seat
16824 * Read SELinux client context from journal clients in a race
16827 * Reorder configuration file lookup order. /etc now always
16828 overrides /run in order to allow the administrator to always
16829 and unconditionally override vendor-supplied or
16830 automatically generated data.
16832 * The various user visible bits of the journal now have man
16833 pages. We still lack man pages for the journal API calls
16836 * We now ship all man pages in HTML format again in the
16839 Contributions from: Dave Reisner, Dirk Eibach, Frederic
16840 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
16841 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
16846 * This is mostly a bugfix release
16848 * systems lacking /etc/os-release are no longer supported.
16850 * Various functionality updates to libsystemd-login.so
16852 * Track class of PAM logins to distinguish greeters from
16853 normal user logins.
16855 Contributions from: Kay Sievers, Lennart Poettering, Michael
16860 * This is an important bugfix release for v41.
16862 * Building man pages is now optional which should be useful
16863 for those building systemd from git but unwilling to install
16866 * Watchdog support for supervising services is now usable. In
16867 a future release support for hardware watchdogs
16868 (i.e. /dev/watchdog) will be added building on this.
16870 * Service start rate limiting is now configurable and can be
16871 turned off per service. When a start rate limit is hit a
16872 reboot can automatically be triggered.
16874 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
16876 Contributions from: Benjamin Franzke, Bill Nottingham,
16877 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
16878 Schmidt, Michał Górny, Piotr Drąg
16882 * The systemd binary is installed /usr/lib/systemd/systemd now;
16883 An existing /sbin/init symlink needs to be adapted with the
16886 * The code that loads kernel modules has been ported to invoke
16887 libkmod directly, instead of modprobe. This means we do not
16888 support systems with module-init-tools anymore.
16890 * Watchdog support is now already useful, but still not
16893 * A new kernel command line option systemd.setenv= is
16894 understood to set system wide environment variables
16895 dynamically at boot.
16897 * We now limit the set of capabilities of systemd-journald.
16899 * We now set SIGPIPE to ignore by default, since it only is
16900 useful in shell pipelines, and has little use in general
16901 code. This can be disabled with IgnoreSIPIPE=no in unit
16904 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
16905 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
16910 * This is mostly a bugfix release
16912 * We now expose the reason why a service failed in the
16913 "Result" D-Bus property.
16915 * Rudimentary service watchdog support (will be completed over
16916 the next few releases.)
16918 * When systemd forks off in order execute some service we will
16919 now immediately changes its argv[0] to reflect which process
16920 it will execute. This is useful to minimize the time window
16921 with a generic argv[0], which makes bootcharts more useful
16923 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
16924 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
16925 Mike Kazantsev, Ray Strode
16929 * This is mostly a test release, but incorporates many
16932 * New systemd-cgtop tool to show control groups by their
16935 * Linking against libacl for ACLs is optional again. If
16936 disabled, support tracking device access for active logins
16937 goes becomes unavailable, and so does access to the user
16938 journals by the respective users.
16940 * If a group "adm" exists, journal files are automatically
16941 owned by them, thus allow members of this group full access
16942 to the system journal as well as all user journals.
16944 * The journal now stores the SELinux context of the logging
16945 client for all entries.
16947 * Add C++ inclusion guards to all public headers
16949 * New output mode "cat" in the journal to print only text
16950 messages, without any meta data like date or time.
16952 * Include tiny X server wrapper as a temporary stop-gap to
16953 teach XOrg udev display enumeration. This is used by display
16954 managers such as gdm, and will go away as soon as XOrg
16955 learned native udev hotplugging for display devices.
16957 * Add new systemd-cat tool for executing arbitrary programs
16958 with STDERR/STDOUT connected to the journal. Can also act as
16959 BSD logger replacement, and does so by default.
16961 * Optionally store all locally generated coredumps in the
16962 journal along with meta data.
16964 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
16965 writing short strings to files (for usage for /sys), and for
16966 creating symlinks, character and block device nodes.
16968 * New unit file option ControlGroupPersistent= to make cgroups
16969 persistent, following the mechanisms outlined in
16970 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
16972 * Support multiple local RTCs in a sane way
16974 * No longer monopolize IO when replaying readahead data on
16975 rotating disks, since we might starve non-file-system IO to
16976 death, since fanotify() will not see accesses done by blkid,
16979 * Do not show kernel threads in systemd-cgls anymore, unless
16980 requested with new -k switch.
16982 Contributions from: Dan Horák, Kay Sievers, Lennart
16983 Poettering, Michal Schmidt
16987 * This is mostly a test release, but incorporates many
16990 * The git repository moved to:
16991 git://anongit.freedesktop.org/systemd/systemd
16992 ssh://git.freedesktop.org/git/systemd/systemd
16994 * First release with the journal
16995 https://0pointer.de/blog/projects/the-journal.html
16997 * The journal replaces both systemd-kmsg-syslogd and
16998 systemd-stdout-bridge.
17000 * New sd_pid_get_unit() API call in libsystemd-logind
17002 * Many systemadm clean-ups
17004 * Introduce remote-fs-pre.target which is ordered before all
17005 remote mounts and may be used to start services before all
17008 * Added Mageia support
17010 * Add bash completion for systemd-loginctl
17012 * Actively monitor PID file creation for daemons which exit in
17013 the parent process before having finished writing the PID
17014 file in the daemon process. Daemons which do this need to be
17015 fixed (i.e. PID file creation must have finished before the
17016 parent exits), but we now react a bit more gracefully to them.
17018 * Add colourful boot output, mimicking the well-known output
17019 of existing distributions.
17021 * New option PassCredentials= for socket units, for
17022 compatibility with a recent kernel ABI breakage.
17024 * /etc/rc.local is now hooked in via a generator binary, and
17025 thus will no longer act as synchronization point during
17028 * systemctl list-unit-files now supports --root=.
17030 * systemd-tmpfiles now understands two new commands: z, Z for
17031 relabelling files according to the SELinux database. This is
17032 useful to apply SELinux labels to specific files in /sys,
17033 among other things.
17035 * Output of SysV services is now forwarded to both the console
17036 and the journal by default, not only just the console.
17038 * New man pages for all APIs from libsystemd-login.
17040 * The build tree got reorganized and the build system is a
17041 lot more modular allowing embedded setups to specifically
17042 select the components of systemd they are interested in.
17044 * Support for Linux systems lacking the kernel VT subsystem is
17047 * configure's --with-rootdir= got renamed to
17048 --with-rootprefix= to follow the naming used by udev and
17051 * Unless specified otherwise we will now install to /usr instead
17052 of /usr/local by default.
17054 * Processes with '@' in argv[0][0] are now excluded from the
17055 final shut-down killing spree, following the logic explained
17057 https://systemd.io/ROOT_STORAGE_DAEMONS/
17059 * All processes remaining in a service cgroup when we enter
17060 the START or START_PRE states are now killed with
17061 SIGKILL. That means it is no longer possible to spawn
17062 background processes from ExecStart= lines (which was never
17063 supported anyway, and bad style).
17065 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
17066 reloading of units together.
17068 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
17069 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
17070 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
17071 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
17072 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / thirdparty / systemd.git / blob