1 systemd System and Service Manager
5 Announcements of Future Feature Removals and Incompatible Changes:
7 * The next release (v255) will remove support for split-usr (/usr/
8 mounted separately during late boot, instead of being mounted by the
9 initrd before switching to the rootfs) and unmerged-usr (parallel
10 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
12 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
14 * We intend to remove cgroup v1 support from a systemd release after
15 the end of 2023. If you run services that make explicit use of
16 cgroup v1 features (i.e. the "legacy hierarchy" with separate
17 hierarchies for each controller), please implement compatibility with
18 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
19 Most of Linux userspace has been ported over already.
21 * Support for System V service scripts is now deprecated and will be
22 removed in a future release. Please make sure to update your software
23 *now* to include a native systemd unit file instead of a legacy
24 System V script to retain compatibility with future systemd releases.
26 * Support for the SystemdOptions EFI variable is deprecated.
27 'bootctl systemd-efi-options' will emit a warning when used. It seems
28 that this feature is little-used and it is better to use alternative
29 approaches like credentials and confexts. The plan is to drop support
30 altogether at a later point, but this might be revisited based on
33 * EnvironmentFile= now treats the line following a comment line
34 trailing with escape as a non comment line. For details, see:
35 https://github.com/systemd/systemd/issues/27975
37 * PrivateNetwork=yes and NetworkNamespacePath= now imply
38 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
40 * Behaviour of sandboxing options for the per-user service manager
41 units has changed. They now imply PrivateUsers=yes, which means user
42 namespaces will be implicitly enabled when a sandboxing option is
43 enabled in a user unit. Enabling user namespaces has the drawback
44 that system users will no longer be visible (and processes/files will
45 appear as owned by 'nobody') in the user unit.
47 By definition a sandboxed user unit should run with reduced
48 privileges, so impact should be small. This will remove a great
49 source of confusion that has been reported by users over the years,
50 due to how these options require an extra setting to be manually
51 enabled when used in the per-user service manager, which is not
52 needed in the system service manager. For more details, see:
53 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
55 * systemd-run's switch --expand-environment= which currently is disabled
56 by default when combined with --scope, will be changed in a future
57 release to be enabled by default.
59 Security Relevant Changes:
61 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
62 process capability to invoked session processes of regular users on
63 local seats (as well as to systemd --user), unless configured
64 otherwise via data from JSON user records, or via the PAM module's
65 parameter list. This is useful in order allow desktop tools such as
66 GNOME's Alarm Clock application to set a timer for
67 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
68 per-user service unit file may thus use AmbientCapability= to pass
69 the capability to invoked processes. Note that this capability is
70 relatively narrow in focus (in particular compared to other process
71 capabilities such as CAP_SYS_ADMIN) and we already — by default —
72 permit more impactful operations such as system suspend to local
77 * Memory limits that apply while the unit is activating are now
78 supported. Previously IO and CPU settings were already supported via
79 StartupCPUWeight= and similar. The same logic has been added for the
80 various manager and unit memory settings (DefaultStartupMemoryLow=,
81 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
82 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
84 * The service manager gained support for enqueuing POSIX signals to
85 services that carry an additional integer value, exposing the
86 sigqueue() system call. This is accessible via new D-Bus calls
87 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
88 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
89 via the new --kill-value= option.
91 * systemctl gained a new "list-paths" verb, which shows all currently
92 active .path units, similarly to how "systemctl list-timers" shows
93 active timers, and "systemctl list-sockets" shows active sockets.
95 * systemctl gained a new --when= switch which is honoured by the various
96 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
97 scheduling these operations by time, similar in fashion to how this
98 has been supported by SysV shutdown.
100 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
101 supports the new PR_SET_MDWE prctl() call, it is used instead of the
102 seccomp()-based system call filter to achieve the same effect.
104 * A new set of kernel command line options is now understood:
105 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
106 systemd.tty.columns.<name>= allow configuring the TTY type and
107 dimensions for the tty specified via <name>. When systemd invokes a
108 service on a tty (via TTYName=) it will look for these and configure
109 the TTY accordingly. This is particularly useful in VM environments
110 to propagate host terminal settings into the appropriate TTYs of the
113 * A new RootEphemeral= setting is now understood in service units. It
114 takes a boolean argument. If enabled for services that use RootImage=
115 or RootDirectory= an ephemeral copy of the disk image or directory
116 tree is made when the service is started. It is removed automatically
117 when the service is stopped. That ephemeral copy is made using
118 btrfs/xfs reflinks or btrfs snapshots, if available.
120 * The service activation logic gained new settings RestartSteps= and
121 RestartMaxDelaySec= which allow exponentially-growing restart
122 intervals for Restart=.
124 * The service activation logic gained a new setting RestartMode= which
125 can be set to 'direct' to skip the inactive/failed states when
126 restarting, so that dependent units are not notified until the service
127 converges to a final (successful or failed) state. For example, this
128 means that OnSuccess=/OnFailure= units will not be triggered until the
129 service state has converged.
131 * PID 1 will now automatically load the virtio_console kernel module
132 during early initialization if running in a suitable VM. This is done
133 so that early-boot logging can be written to the console if available.
135 * Similarly, virtio-vsock support is loaded early in suitable VM
136 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
137 to the VMM if configured, thus loading this early is beneficial.
139 * A new verb "fdstore" has been added to systemd-analyze to show the
140 current contents of the file descriptor store of a unit. This is
141 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
144 * The service manager will now set a new $FDSTORE environment variable
145 when invoking processes for services that have the file descriptor
148 * A new service option FileDescriptorStorePreserve= has been added that
149 allows tuning the life-cycle of the per-service file descriptor
150 store. If set to "yes", the entries in the fd store are retained even
151 after the service has been fully stopped.
153 * The "systemctl clean" command may now be used to clear the fdstore of
156 * Unit *.preset files gained a new directive "ignore", in addition to
157 the existing "enable" and "disable". As the name suggests, matching
158 units are left unchanged, i.e. neither enabled nor disabled.
160 * Service units gained a new setting DelegateSubgroup=. It takes the
161 name of a sub-cgroup to place any processes the service manager forks
162 off in. Previously, the service manager would place all service
163 processes directly in the top-level cgroup it created for the
164 service. This usually meant that main process in a service with
165 delegation enabled would first have to create a subgroup and move
166 itself down into it, in order to not conflict with the "no processes
167 in inner cgroups" rule of cgroup v2. With this option, this step is
168 now handled by PID 1.
170 * The service manager will now look for .upholds/ directories,
171 similarly to the existing support for .wants/ and .requires/
172 directories. Symlinks in this directory result in Upholds=
175 The [Install] section of unit files gained support for a new
176 UpheldBy= directive to generate .upholds/ symlinks automatically when
179 * The service manager now supports a new kernel command line option
180 systemd.default_device_timeout_sec=, which may be used to override
181 the default timeout for .device units.
183 * A new "soft-reboot" mechanism has been added to the service manager.
184 A "soft reboot" is similar to a regular reboot, except that it
185 affects userspace only: the service manager shuts down any running
186 services and other units, then optionally switches into a new root
187 file system (mounted to /run/nextroot/), and then passes control to a
188 systemd instance in the new file system which then starts the system
189 up again. The kernel is not rebooted and neither is the hardware,
190 firmware or boot loader. This provides a fast, lightweight mechanism
191 to quickly reset or update userspace, without the latency that a full
192 system reset involves. Moreover, open file descriptors may be passed
193 across the soft reboot into the new system where they will be passed
194 back to the originating services. This allows pinning resources
195 across the reboot, thus minimizing grey-out time further. This new
196 reboot mechanism is accessible via the new "systemctl soft-reboot"
199 * Services using RootDirectory= or RootImage= will now have read-only
200 access to a copy of the host's os-release file under
201 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
202 This was already the case for Portable Services, and the feature has
203 now been extended to all services that do not run off the host's
206 * A new service setting MemoryKSM= has been added to enable kernel
207 same-page merging individually for services.
209 * A new service setting ImportCredentials= has been added that augments
210 LoadCredential= and LoadCredentialEncrypted= and searches for
211 credentials to import from the system, and supports globbing.
213 * A new job mode "restart-dependencies" has been added to the service
214 manager (exposed via systemctl --job-mode=). It is only valid when
215 used with "start" jobs, and has the effect that the "start" job will
216 be propagated as "restart" jobs to currently running units that have
217 a BindsTo= or Requires= dependency on the started unit.
219 * A new verb "whoami" has been added to "systemctl" which determines as
220 part of which unit the command is being invoked. It writes the unit
221 name to standard output. If one or more PIDs are specified reports
222 the unit names the processes referenced by the PIDs belong to.
224 * The system and service credential logic has been improved: there's
225 now a clearly defined place where system provisioning tools running
226 in the initrd can place credentials that will be imported into the
227 system's set of credentials during the initrd → host transition: the
228 /run/credentials/@initrd/ directory. Once the credentials placed
229 there are imported into the system credential set they are deleted
230 from this directory, and the directory itself is deleted afterwards
233 * A new kernel command line option systemd.set_credential_binary= has
234 been added, that is similar to the pre-existing
235 systemd.set_credential= but accepts arbitrary binary credential data,
236 encoded in Base64. Note that the kernel command line is not a
237 recommend way to transfer credentials into a system, since it is
238 world-readable from userspace.
240 * The default machine ID to use may now be configured via the
241 system.machine_id system credential. It will only be used if no
242 machine ID was set yet on the host.
244 * On Linux kernel 6.4 and newer system and service credentials will now
245 be placed in a tmpfs instance that has the "noswap" mount option
246 set. Previously, a "ramfs" instance was used. By switching to tmpfs
247 ACL support and overall size limits can now be enforced, without
248 compromising on security, as the memory is never paged out either
251 * The service manager now can detect when it is running in a
252 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
253 accepted by ConditionSecurity= for units that want to conditionalize
254 themselves on this. systemd-detect-virt gained new 'cvm' and
255 '--list-cvm' switches to respectively perform the detection or list
256 all known flavours of confidential VM, depending on the vendor. The
257 manager will publish a 'ConfidentialVirtualization' D-Bus property,
258 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
259 variable for unit generators. Finally, udev rules can match on a new
260 'cvm' key that will be set when in a confidential VM.
261 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
262 strings and QEMU's fw_cfg protocol will not be used to import
263 credentials and kernel command line parameters by the system manager,
264 systemd-boot and systemd-stub, because the hypervisor is considered
265 untrusted in this particular setting.
269 * The sd-journal API gained a new call sd_journal_get_seqnum() to
270 retrieve the current log record's sequence number and sequence number
271 ID, which allows applications to order records the same way as
272 journal does internally. The sequence number is now also exported in
273 the JSON and "export" output of the journal.
275 * journalctl gained a new switch --truncate-newline. If specified
276 multi-line log records will be truncated at the first newline,
277 i.e. only the first line of each log message will be shown.
279 * systemd-journal-upload gained support for --namespace=, similar to
280 the switch of the same name of journalctl.
284 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
285 may be used to exclude certain files from the effect of CopyFiles=.
287 * systemd-repart's Verity support now implements the Minimize= setting
288 to minimize the size of the resulting partition.
290 * systemd-repart gained a new --offline= switch, which may be used to
291 control whether images shall be built "online" or "offline",
292 i.e. whether to make use of kernel facilities such as loopback block
293 devices and device mapper or not.
295 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
296 partition with some files, it will now default to VFAT rather than
299 * systemd-repart gained a new --architecture= switch. If specified, the
300 per-architecture GPT partition types (i.e. the root and /usr/
301 partitions) configured in the partition drop-in files are
302 automatically adjusted to match the specified CPU architecture, in
303 order to simplify cross-architecture DDI building.
305 * systemd-repart will now default to a minimum size of 300MB for XFS
306 filesystems if no size parameter is specified. This matches what the
307 XFS tools (xfsprogs) can support.
309 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
311 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
312 Instead, pyelftools is now needed, and it will be used to perform the
313 ELF -> PE relocations at build time.
315 * bootctl gained a new switch --print-root-device/-R that prints the
316 block device the root file system is backed by. If specified twice,
317 it returns the whole disk block device (as opposed to partition block
318 device) the root file system is on. It's useful for invocations such
319 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
322 * systemd-stub will now look for the SMBIOS Type 1 field
323 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
324 kernel command line it invokes. This is useful for VMMs such as qemu
325 to pass additional kernel command lines into the system even when
326 booting via full UEFI. The contents of the field are measured into
329 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
330 value "auto". With this value, a kernel will be automatically
331 analyzed, and if it qualifies as UKI, it will be installed as if the
332 setting was to set to "uki", otherwise as "bls".
334 * systemd-stub can now optionally load UEFI PE "add-on" images that may
335 contain additional kernel command line information. These "add-ons"
336 superficially look like a regular UEFI executable, and are expected
337 to be signed via SecureBoot/shim. However, they do not actually
338 contain code, but instead a subset of the PE sections that UKIs
339 support. They are supposed to provide a way to extend UKIs with
340 additional resources in a secure and authenticated way. Currently,
341 only the .cmdline PE section may be used in add-ons, in which case
342 any specified string is appended to the command line embedded into
343 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
344 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
345 the future we expect other sections to be made extensible like this as
348 * ukify has been updated to allow building these UEFI PE "add-on"
349 images, using the new 'addon<EFI-ARCH>.efi.stub'.
351 * ukify gained a new "genkey" verb for generating a set of of key pairs
352 to sign UKIs and their PCR data with.
354 * ukify now accepts SBAT information to place in the .sbat PE section
355 of UKIs and addons. If a UKI is built the SBAT information from the
356 inner kernel is merged with any SBAT information associated with
357 systemd-stub and the SBAT data specified on the ukify command line.
359 * The kernel-install script has been rewritten in C, and reuses much of
360 the infrastructure of existing tools such as bootctl. It also gained
361 --esp-path= and --boot-path= options to override the path to the ESP,
362 and the $BOOT partition. Options --make-entry-directory= and
363 --entry-token= have been added as well, similar to bootctl's options
366 * A new kernel-install plugin 60-ukify has been added which will
367 combine kernel/initrd locally into a UKI and optionally sign them
368 with a local key. This may be used to switch to UKI mode even on
369 systems where a local kernel or initrd is used. (Typically UKIs are
370 built and signed by the vendor.)
372 * The ukify tool now supports "pesign" in addition to the pre-existing
373 "sbsign" for signing UKIs.
375 * systemd-measure and systemd-stub now look for the .uname PE section
376 that should contain the kernel's "uname -r" string.
378 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
379 "offline", i.e. without access to a TPM (physical or
382 Memory Pressure & Control:
384 * The sd-event API gained new calls sd_event_add_memory_pressure(),
385 sd_event_source_set_memory_pressure_type(),
386 sd_event_source_set_memory_pressure_period() to create and configure
387 an event source that is called whenever the OS signals memory
388 pressure. Another call sd_event_trim_memory() is provided that
389 compacts the process' memory use by releasing allocated but unused
390 malloc() memory back to the kernel. Services can also provide their
391 own custom callback to do memory trimming. This should improve system
392 behaviour under memory pressure, as on Linux traditionally provided
393 no mechanism to return process memory back to the kernel if the
394 kernel was under memory pressure. This makes use of the kernel's PSI
395 interface. Most long-running services in systemd have been hooked up
396 with this, and in particular systems with low memory should benefit
399 * Service units gained new settings MemoryPressureWatch= and
400 MemoryPressureThresholdSec= to configure the PSI memory pressure
401 logic individually. If these options are used, the
402 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
403 variables will be set for the invoked processes to inform them about
404 the requested memory pressure behaviour. (This is used by the
405 aforementioned sd-events API additions, if set.)
407 * systemd-analyze gained a new "malloc" verb that shows the output
408 generated by glibc's malloc_info() on services that support it. Right
409 now, only the service manager has been updated accordingly. This
410 call requires privileges.
412 User & Session Management:
414 * The sd-login API gained a new call sd_session_get_username() to
415 return the user name of the owner of a login session. It also gained
416 a new call sd_session_get_start_time() to retrieve the time the login
417 session started. A new call sd_session_get_leader() has been added to
418 return the PID of the "leader" process of a session. A new call
419 sd_uid_get_login_time() returns the time since the specified user has
420 most recently been continuously logged in with at least one session.
422 * JSON user records gained a new set of fields capabilityAmbientSet and
423 capabilityBoundingSet which contain a list of POSIX capabilities to
424 set for the logged in users in the ambient and bounding sets,
425 respectively. homectl gained the ability to configure these two sets
426 for users via --capability-bounding-set=/--capability-ambient-set=.
428 * pam_systemd learnt two new module options
429 default-capability-bounding-set= and default-capability-ambient-set=,
430 which configure the default bounding sets for users as they are
431 logging in, if the JSON user record doesn't specify this explicitly
432 (see above). The built-in default for the ambient set now contains
433 the CAP_WAKE_ALARM, thus allowing regular users who may log in
434 locally to resume from a system suspend via a timer.
436 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
437 call to update the TTY of a session after it has been allocated. This
438 is useful for SSH sessions which are typically allocated first, and
439 for which a TTY is added later.
441 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
442 combines the various other sd_pid_notify() flavours into one: takes a
443 format string, an overriding PID, and a set of file descriptors to
444 send. It also gained a new call sd_pid_notify_barrier() call which is
445 equivalent to sd_notify_barrier() but allows the originating PID to
448 * "loginctl list-users" and "loginctl list-sessions" will now show the
449 state of each logged in user/session in their tabular output. It will
450 also show the current idle state of sessions.
454 * systemd-dissect will now show the intended CPU architecture of an
457 * systemd-dissect will now install itself as mount helper for the "ddi"
458 pseudo-file system type. This means you may now mount DDIs directly
459 via /bin/mount or /etc/fstab, making full use of embedded Verity
460 information and all other DDI features.
462 Example: mount -t ddi myimage.raw /some/where
464 * The systemd-dissect tool gained the new switches --attach/--detach to
465 attach/detach a DDI to a loopback block device without mounting it.
466 It will automatically derive the right sector size from the image
467 and set up Verity and similar, but not mount the file systems in it.
469 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
470 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
471 implied. Given that these file systems are typically untrusted, this
472 should make mounting them automatically have less of a security
475 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
476 systemd-tmpfiles, …) now understand a new switch --image-policy= which
477 takes a string encoding image dissection policy. With this mechanism
478 automatic discovery and use of specific partition types and the
479 cryptographic requirements on the partitions (Verity, LUKS, …) can be
480 restricted, permitting better control of the exposed attack surfaces
481 when mounting disk images. systemd-gpt-auto-generator will honour such
482 an image policy too, configurable via the systemd.image_policy= kernel
483 command line option. Unit files gained the RootImagePolicy=,
484 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
485 disk images a service runs off.
487 * systemd-analyze gained a new verb "image-policy" to validate and
488 parse image policy strings.
490 * systemd-dissect gained support for a new --validate switch to
491 superficially validate DDI structure, and check whether a specific
492 image policy allows the DDI.
494 * systemd-dissect gained support for a new --mtree-hash switch to
495 optionally disable calculating mtree hashes, which can be slow on
498 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
499 are now able to operate on directories too, other than images.
503 * networkd's GENEVE support as gained a new .network option
504 InheritInnerProtocol=.
506 * The [Tunnel] section in .netdev files has gained a new setting
507 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
509 * A new global IPv6PrivacyExtensions= setting has been added that
510 selects the default value of the per-network setting of the same
513 * The predictable network interface naming logic will now include
514 SR-IOV-R "representor" information in network interface names.
516 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
517 the RFC8910 captive portal option.
521 * udevadm gained the new "verify" verb for validating udev rules files
524 * udev gained a new tool "iocost" that can be used to configure QoS IO
525 cost data based on hwdb information onto suitable block devices. Also
526 see https://github.com/iocost-benchmark/iocost-benchmarks.
528 TPM2 Support + Disk Encryption & Authentication:
530 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
531 ("Storage Root Key") as first step in the TPM2, and then use that
532 for binding FDE to, if TPM2 support is used. This matches
533 recommendations of TCG (see
534 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
536 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
537 understand textual identifiers for these PCRs.
539 * systemd-veritysetup + /etc/veritytab gained support for a series of
540 new options: hash-offset=, superblock=, format=, data-block-size=,
541 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
542 fec-offset=, fec-roots= to configure various aspects of a Verity
545 * systemd-cryptsetup + /etc/crypttab gained support for a new
546 veracrypt-pim= option for setting the Personal Iteration Multiplier
547 of veracrypt volumes.
549 * systemd-integritysetup + /etc/integritytab gained support for a new
550 mode= setting for controlling the dm-integrity mode (journal, bitmap,
551 direct) for the volume.
553 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
554 registers, their symbolic names and current values.
558 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
559 access right is specified this is equivalent to "x" but only if the
560 inode in question already has the executable bit set for at least
561 some user/group. Otherwise the "x" bit will be turned off.
563 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
564 with C+ will result in a "merge" copy, i.e. all files of the source
565 tree are copied into the target tree, even if that tree already
566 exists, resulting in a combined tree of files already present in the
567 target tree and those copied in.
569 * systemd-tmpfiles gained a new --graceful switch. If specified lines
570 with unknown users/groups will silently be skipped.
574 * systemd-notify gained two new options --fd= and --fdname= for sending
575 arbitrary file descriptors to the service manager (while specifying an
576 explicit name for it).
578 * systemd-notify gained a new --exec switch, which makes it execute the
579 specified command line after sending the requested messages. This is
580 useful for sending out READY=1 first, and then continuing invocation
581 without changing process ID, so that the tool can be nicely used
582 within an ExecStart= line of a unit file that uses Type=notify.
584 sd-event + sd-bus APIs:
586 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
587 which may be used to explicitly end a rate-limit state an event
588 source might be in, resetting all rate limiting counters.
590 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
591 sockets, it will now encode the "description" set via
592 sd_bus_set_description() into the source socket address. It will also
593 look for this information when accepting a connection. This is useful
594 to track individual D-Bus connections on a D-Bus broker for debug
599 * systemd-resolved gained a new resolved.conf setting
600 StateRetentionSec= which may be used to retain cached DNS records
601 even after their nominal TTL, and use them in case upstream DNS
602 servers cannot be reached. This can be sued to make name resolution
603 more resilient in case of network problems.
605 * resolvectl gained a new verb "show-cache" to show the current cache
606 contents of systemd-resolved. This verb communicates with the
607 systemd-resolved daemon and requires privileges.
611 * Meson >= 0.60.0 is now required to build systemd.
613 * The default keymap to apply may now be chosen at build-time via the
614 new -Ddefault-keymap= meson option.
616 * Most of systemd's long-running services now have a generic handler of
617 the SIGRTMIN+18 signal handler which executes various operations
618 depending on the sigqueue() parameter sent along. For example, values
619 0x100…0x107 allow changing the maximum log level of such
620 services. 0x200…0x203 allow changing the log target of such
621 services. 0x300 make the services trim their memory similarly to the
622 automatic PSI-triggered action, see above. 0x301 make the services
623 output their malloc_info() data to the logs.
625 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
626 files, inspired by systemctl's verbs of the same name which edit unit
627 files. Similarly, networkctl gained the same verbs for editing
628 .network, .netdev, .link files.
630 * A new syscall filter group "@sandbox" has been added that contains
631 syscalls for sandboxing system calls such as those for seccomp and
634 * New documentation has been added:
636 https://systemd.io/COREDUMP
637 https://systemd.io/MEMORY_PRESSURE
640 * systemd-firstboot gained a new --reset option. If specified, the
641 settings in /etc/ it knows how to initialize are reset.
643 * systemd-sysext is now a multi-call binary and is also installed under
644 the systemd-confext alias name (via a symlink). When invoked that way
645 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
646 powerful, atomic, secure configuration management of sorts, that
647 locally can merge configuration from multiple confext configuration
648 images into a single immutable tree.
650 * The --network-macvlan=, --network-ipvlan=, --network-interface=
651 switches of systemd-nspawn may now optionally take the intended
652 network interface inside the container.
654 * All our programs will now send an sd_notify() message with their exit
655 status in the EXIT_STATUS= field when exiting, using the usual
656 protocol, including PID 1. This is useful for VMMs and container
657 managers to collect an exit status from a system as it shuts down, as
658 set via "systemctl exit …". This is particularly useful in test cases
659 and similar, as invocations via a VM can now nicely propagate an exit
660 status to the host, similar to local processes.
662 * systemd-run gained a new switch --expand-environment=no to disable
663 server-side environment variable expansion in specified command
664 lines. Expansion defaults to enabled for all execution types except
665 --scope, where it defaults to off (and prints a warning) for backward
666 compatibility reasons. --scope will be flipped to enabled by default
667 too in a future release. If you are using --scope and passing a '$'
668 character in the payload you should start explicitly using
669 --expand-environment=yes/no according to the use case.
671 * The systemd-system-update-generator has been updated to also look for
672 the special flag file /etc/system-update in addition to the existing
673 support for /system-update to decide whether to enter system update
676 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
677 mount options by default.
679 * systemd-fstab-generator now understands two new kernel command line
680 options systemd.mount-extra= and systemd.swap-extra=, which configure
681 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
682 will be ran on these block devices, like it already happens for
683 'root='. It also now supports the new fstab.extra and
684 fstab.extra.initrd credentials that may contain additional /etc/fstab
685 lines to apply at boot.
687 * systemd-getty-generator now understands two new credentials
688 getty.ttys.container and getty.ttys.serial. These credentials may
689 contain a list of TTY devices – one per line – to instantiate
690 container-getty@.service and serial-getty@.service on.
692 * The getty/serial-getty/container-getty units now import the 'agetty.*'
693 and 'login.*' credentials, which are consumed by the 'login' and
694 'agetty' programs starting from util-linux v2.40.
696 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
697 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
698 which case the Path= setting is taken relative to the ESP or XBOOTLDR
699 partitions, rather than the system's root directory /. The relevant
700 directories are automatically discovered.
702 * The systemd-ac-power tool gained a new switch --low, which reports
703 whether the battery charge is considered "low", similar to how the
704 s2h suspend logic checks this state to decide whether to enter system
705 suspend or hibernation.
707 * The /etc/os-release file can now have two new optional fields
708 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
711 * When the system hibernates, information about the device and offset
712 used is now written to a non-volatile EFI variable. On next boot the
713 system will attempt to resume from the location indicated in this EFI
714 variable. This should make hibernation a lot more robust, while
715 requiring no manual configuration of the resume location.
717 * The $XDG_STATE_HOME environment variable (added in more recent
718 versions of the XDG basedir specification) is now honoured to
719 implement the StateDirectory= setting in user services.
721 * A new component "systemd-battery-check" has been added. It may run
722 during early boot (usually in the initrd), and checks the battery
723 charge level of the system. In case the charge level is very low the
724 user is notified (graphically via Plymouth – if available – as well
725 as in text form on the console), and the system is turned off after a
726 10s delay. The feature can be disabled by passing
727 systemd.battery-check=0 through the kernel command line.
729 * The 'passwdqc' library is now supported as an alternative to the
730 'pwquality' library and can be selected at build time.
732 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
733 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
734 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
735 Arian van Putten, Arthur Shau, A S Alam,
736 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
737 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
738 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
739 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
740 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
741 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
742 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
743 David Schroeder, David Tardon, dependabot[bot],
744 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
745 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
746 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
747 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
748 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
749 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
750 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
751 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
752 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
753 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
754 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
755 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
756 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
757 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
758 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
759 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
760 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
761 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
762 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
763 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
764 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
765 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
766 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
767 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
768 saikat0511, Samanta Navarro, Sam James, Sam Morris,
769 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
770 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
771 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
772 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
773 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
774 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
775 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
776 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
777 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
778 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
779 Дамјан Георгиевски, наб
781 — Edinburgh, 2023-07-28
785 Announcements of Future Feature Removals and Incompatible Changes:
787 * We intend to remove cgroup v1 support from systemd release after the
788 end of 2023. If you run services that make explicit use of cgroup v1
789 features (i.e. the "legacy hierarchy" with separate hierarchies for
790 each controller), please implement compatibility with cgroup v2 (i.e.
791 the "unified hierarchy") sooner rather than later. Most of Linux
792 userspace has been ported over already.
794 * We intend to remove support for split-usr (/usr mounted separately
795 during boot) and unmerged-usr (parallel directories /bin and
796 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
797 half of 2023, in the first release that falls into that time window.
798 For more details, see:
799 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
801 * We intend to change behaviour w.r.t. units of the per-user service
802 manager and sandboxing options, so that they work without having to
803 manually enable PrivateUsers= as well, which is not required for
804 system units. To make this work, we will implicitly enable user
805 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
806 user unit. The drawback is that system users will no longer be visible
807 (and appear as 'nobody') to the user unit when a sandboxing option is
808 enabled. By definition a sandboxed user unit should run with reduced
809 privileges, so impact should be small. This will remove a great source
810 of confusion that has been reported by users over the years, due to
811 how these options require an extra setting to be manually enabled when
812 used in the per-user service manager, as opposed as to the system
813 service manager. We plan to enable this change in the next release
814 later this year. For more details, see:
815 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
817 Deprecations and incompatible changes:
819 * systemctl will now warn when invoked without /proc/ mounted
820 (e.g. when invoked after chroot() into an directory tree without the
821 API mount points like /proc/ being set up.) Operation in such an
822 environment is not fully supported.
824 * The return value of 'systemctl is-active|is-enabled|is-failed' for
825 unknown units is changed: previously 1 or 3 were returned, but now 4
826 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
828 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
829 systemd-hwdb (added in 2014) should be used instead.
831 * 'bootctl --json' now outputs a single JSON array, instead of a stream
832 of newline-separated JSON objects.
834 * Udev rules in 60-evdev.rules have been changed to load hwdb
835 properties for all modalias patterns. Previously only the first
836 matching pattern was used. This could change what properties are
837 assigned if the user has more and less specific patterns that could
838 match the same device, but it is expected that the change will have
839 no effect for most users.
841 * systemd-networkd-wait-online exits successfully when all interfaces
842 are ready or unmanaged. Previously, if neither '--any' nor
843 '--interface=' options were used, at least one interface had to be in
844 configured state. This change allows the case where systemd-networkd
845 is enabled, but no interfaces are configured, to be handled
846 gracefully. It may occur in particular when a different network
847 manager is also enabled and used.
849 * Some compatibility helpers were dropped: EmergencyAction= in the user
850 manager, as well as measuring kernel command line into PCR 8 in
851 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
854 * The '-Dupdate-helper-user-timeout=' build-time option has been
855 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
856 integer as parameter instead of a string.
858 * The DDI image dissection logic (which backs RootImage= in service
859 unit files, the --image= switch in various tools such as
860 systemd-nspawn, as well as systemd-dissect) will now only mount file
861 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
862 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
863 variable. These file systems are fairly well supported and maintained
864 in current kernels, while others are usually more niche, exotic or
865 legacy and thus typically do not receive the same level of security
868 * The default per-link multicast DNS mode is changed to "yes"
869 (that was previously "no"). As the default global multicast DNS mode
870 has been "yes" (but can be changed by the build option), now the
871 multicast DNS is enabled on all links by default. You can disable the
872 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
873 or on an interface by calling "resolvectl mdns INTERFACE no".
877 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
878 (UKIs) has been added. This replaces functionality provided by
879 'dracut --uefi' and extends it with automatic calculation of PE file
880 offsets, insertion of signed PCR policies generated by
881 systemd-measure, support for initrd concatenation, signing of the
882 embedded Linux image and the combined image with sbsign, and
883 heuristics to autodetect the kernel uname and verify the splash
886 Changes in systemd and units:
888 * A new service type Type=notify-reload is defined. When such a unit is
889 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
890 service process. The manager will then wait until it receives a
891 "RELOADING=1" followed by a "READY=1" notification from the unit as
892 response (via sd_notify()). Otherwise, this type is the same as
893 Type=notify. A new setting ReloadSignal= may be used to change the
894 signal to send from the default of SIGHUP.
896 user@.service, systemd-networkd.service, systemd-udevd.service, and
897 systemd-logind have been updated to this type.
899 * Initrd environments which are not on a pure memory file system (e.g.
900 overlayfs combination as opposed to tmpfs) are now supported. With
901 this change, during the initrd → host transition ("switch root")
902 systemd will erase all files of the initrd only when the initrd is
903 backed by a memory file system such as tmpfs.
905 * New per-unit MemoryZSwapMax= option has been added to configure
906 memory.zswap.max cgroup properties (the maximum amount of zswap
909 * A new LogFilterPatterns= option has been added for units. It may be
910 used to specify accept/deny regular expressions for log messages
911 generated by the unit, that shall be enforced by systemd-journald.
912 Rejected messages are neither stored in the journal nor forwarded.
913 This option may be used to suppress noisy or uninteresting messages
916 * The manager has a new
917 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
918 query process ownership via a PIDFD, which is more resilient against
919 PID recycling issues.
921 * Scope units now support OOMPolicy=. Login session scopes default to
922 OOMPolicy=continue, allowing login scopes to survive the OOM killer
923 terminating some processes in the scope.
925 * systemd-fstab-generator now supports x-systemd.makefs option for
926 /sysroot/ (in the initrd).
928 * The maximum rate at which daemon reloads are executed can now be
929 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
930 options. (Or the equivalent on the kernel command line:
931 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
932 addition, systemd now logs the originating unit and PID when a reload
933 request is received over D-Bus.
935 * When enabling a swap device systemd will now reinitialize the device
936 when the page size of the swap space does not match the page size of
937 the running kernel. Note that this requires the 'swapon' utility to
938 provide the '--fixpgsz' option, as implemented by util-linux, and it
939 is not supported by busybox at the time of writing.
941 * systemd now executes generator programs in a mount namespace
942 "sandbox" with most of the file system read-only and write access
943 restricted to the output directories, and with a temporary /tmp/
944 mount provided. This provides a safeguard against programming errors
945 in the generators, but also fixes here-docs in shells, which
946 previously didn't work in early boot when /tmp/ wasn't available
947 yet. (This feature has no security implications, because the code is
948 still privileged and can trivially exit the sandbox.)
950 * The system manager will now parse a new "vmm.notify_socket"
951 system credential, which may be supplied to a VM via SMBIOS. If
952 found, the manager will send a "READY=1" notification on the
953 specified socket after boot is complete. This allows readiness
954 notification to be sent from a VM guest to the VM host over a VSOCK
957 * The sample PAM configuration file for systemd-user@.service now
958 includes a call to pam_namespace. This puts children of user@.service
959 in the expected namespace. (Many distributions replace their file
960 with something custom, so this change has limited effect.)
962 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
963 can be used to override the mount units burst late limit for
964 parsing '/proc/self/mountinfo', which was introduced in v249.
967 * Drop-ins for init.scope changing control group resource limits are
968 now applied, while they were previously ignored.
970 * New build-time configuration options '-Ddefault-timeout-sec=' and
971 '-Ddefault-user-timeout-sec=' have been added, to let distributions
972 choose the default timeout for starting/stopping/aborting system and
973 user units respectively.
975 * Service units gained a new setting OpenFile= which may be used to
976 open arbitrary files in the file system (or connect to arbitrary
977 AF_UNIX sockets in the file system), and pass the open file
978 descriptor to the invoked process via the usual file descriptor
979 passing protocol. This is useful to give unprivileged services access
980 to select files which have restrictive access modes that would
981 normally not allow this. It's also useful in case RootDirectory= or
982 RootImage= is used to allow access to files from the host environment
983 (which is after all not visible from the service if these two options
988 * The new net naming scheme "v253" has been introduced. In the new
989 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
990 a PCI bus. This extends the coverage of predictable interface names
991 in some embedded systems.
993 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
994 a more informative path on some embedded systems.
996 * Partition block devices will now also get symlinks in
997 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
998 block device nodes via the kernel's "diskseq" value. Previously those
999 symlinks were only created for the main block device.
1001 * A new operator '-=' is supported for SYMLINK variables. This allows
1002 symlinks to be unconfigured even if an earlier rule added them.
1004 * 'udevadm --trigger --settle' now also works for network devices
1005 that are being renamed.
1007 Changes in sd-boot, bootctl, and the Boot Loader Specification:
1009 * systemd-boot now passes its random seed directly to the kernel's RNG
1010 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
1011 means the RNG gets seeded very early in boot before userspace has
1014 * systemd-boot will pass a disk-backed random seed – even when secure
1015 boot is enabled – if it can additionally get a random seed from EFI
1016 itself (via EFI's RNG protocol), or a prior seed in
1017 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
1019 * systemd-boot-system-token.service was renamed to
1020 systemd-boot-random-seed.service and extended to always save a random
1021 seed to ESP on every boot when a compatible boot loader is used. This
1022 allows a refreshed random seed to be used in the boot loader.
1024 * systemd-boot handles various seed inputs using a domain- and
1025 field-separated hashing scheme.
1027 * systemd-boot's 'random-seed-mode' option has been removed. A system
1028 token is now always required to be present for random seeds to be
1031 * systemd-boot now supports being loaded from other locations than the
1032 ESP, for example for direct kernel boot under QEMU or when embedded
1035 * systemd-boot now parses SMBIOS information to detect
1036 virtualization. This information is used to skip some warnings which
1037 are not useful in a VM and to conditionalize other aspects of
1040 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
1041 Secure Boot automated certificate enrollment from the ESP only if it
1042 is considered 'safe' to do so. At the moment 'safe' means running in
1045 * systemd-stub now processes random seeds in the same way as
1046 systemd-boot already does, in case a unified kernel image is being
1047 used from a different bootloader than systemd-boot, or without any
1050 * bootctl will now generate a system token on all EFI systems, even
1051 virtualized ones, and is activated in the case that the system token
1052 is missing from either sd-boot and sd-stub booted systems.
1054 * bootctl now implements two new verbs: 'kernel-identify' prints the
1055 type of a kernel image file, and 'kernel-inspect' provides
1056 information about the embedded command line and kernel version of
1059 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
1060 as for kernel-install.
1062 * The JSON output of "bootctl list" will now contain two more fields:
1063 isDefault and isSelected are boolean fields set to true on the
1064 default and currently booted boot menu entries.
1066 * bootctl gained a new verb "unlink" for removing a boot loader entry
1067 type #1 file from disk in a safe and robust way.
1069 * bootctl also gained a new verb "cleanup" that automatically removes
1070 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
1071 is not referenced anymore by any installed Type #1 boot loader
1072 specification entry. This is particularly useful in environments where
1073 a large number of entries reference the same or partly the same
1074 resources (for example, for snapshot-based setups).
1076 Changes in kernel-install:
1078 * A new "installation layout" can be configured as layout=uki. With
1079 this setting, a Boot Loader Specification Type#1 entry will not be
1080 created. Instead, a new kernel-install plugin 90-uki-copy.install
1081 will copy any .efi files from the staging area into the boot
1082 partition. A plugin to generate the UKI .efi file must be provided
1085 Changes in systemctl:
1087 * 'systemctl reboot' has dropped support for accepting a positional
1088 argument as the argument to the reboot(2) syscall. Please use the
1089 --reboot-argument= option instead.
1091 * 'systemctl disable' will now warn when called on units without
1092 install information. A new --no-warn option has been added that
1093 silences this warning.
1095 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
1096 of the drop-in to edit. (Previously, 'override.conf' was always
1099 * 'systemctl list-dependencies' now respects --type= and --state=.
1101 * 'systemctl kexec' now supports XEN VMM environments.
1103 * 'systemctl edit' will now tell the invoked editor to jump into the
1104 first line with actual unit file data, skipping over synthesized
1107 Changes in systemd-networkd and related tools:
1109 * The [DHCPv4] section in .network file gained new SocketPriority=
1110 setting that assigns the Linux socket priority used by the DHCPv4 raw
1111 socket. This may be used in conjunction with the
1112 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
1113 desired ethernet 802.1Q frame priority for DHCPv4 initial
1114 packets. This cannot be achieved with netfilter mangle tables because
1115 of the raw socket bypass.
1117 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
1118 new QuickAck= boolean setting that enables the TCP quick ACK mode for
1119 the routes configured by the acquired DHCPv4 lease or received router
1120 advertisements (RAs).
1122 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
1123 routes) now accepts three values, for high, medium, and low preference
1124 of the router (which can be set with the RouterPreference=) setting.
1126 * systemd-networkd-wait-online now supports matching via alternative
1129 * The [DHCPv6] section in .network file gained new SendRelease=
1130 setting which enables the DHCPv6 client to send release when
1131 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
1132 It is enabled by default.
1134 * If the Address= setting in [Network] or [Address] sections in .network
1135 specified without its prefix length, then now systemd-networkd assumes
1136 /32 for IPv4 or /128 for IPv6 addresses.
1138 * networkctl shows network and link file dropins in status output.
1140 Changes in systemd-dissect:
1142 * systemd-dissect gained a new option --list, to print the paths of
1143 all files and directories in a DDI.
1145 * systemd-dissect gained a new option --mtree, to generate a file
1146 manifest compatible with BSD mtree(5) of a DDI
1148 * systemd-dissect gained a new option --with, to execute a command with
1149 the specified DDI temporarily mounted and used as working
1150 directory. This is for example useful to convert a DDI to "tar"
1151 simply by running it within a "systemd-dissect --with" invocation.
1153 * systemd-dissect gained a new option --discover, to search for
1154 Discoverable Disk Images (DDIs) in well-known directories of the
1155 system. This will list machine, portable service and system extension
1158 * systemd-dissect now understands 2nd stage initrd images stored as a
1159 Discoverable Disk Image (DDI).
1161 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
1162 disk UUID stored in the GPT header) among the other data it can show.
1164 * systemd-dissect gained a new --in-memory switch to operate on an
1165 in-memory copy of the specified DDI file. This is useful to access a
1166 DDI with write access without persisting any changes. It's also
1167 useful for accessing a DDI without keeping the originating file
1170 * The DDI dissection logic will now automatically detect the intended
1171 sector size of disk images stored in files, based on the GPT
1172 partition table arrangement. Loopback block devices for such DDIs
1173 will then be configured automatically for the right sector size. This
1174 is useful to make dealing with modern 4K sector size DDIs fully
1175 automatic. The systemd-dissect tool will now show the detected sector
1176 size among the other DDI information in its output.
1178 Changes in systemd-repart:
1180 * systemd-repart gained new options --include-partitions= and
1181 --exclude-partitions= to filter operation on partitions by type UUID.
1182 This allows systemd-repart to be used to build images in which the
1183 type of one partition is set based on the contents of another
1184 partition (for example when the boot partition shall include a verity
1185 hash of the root partition).
1187 * systemd-repart also gained a --defer-partitions= option that is
1188 similar to --exclude-partitions=, but the size of the partition is
1189 still taken into account when sizing partitions, but without
1192 * systemd-repart gained a new --sector-size= option to specify what
1193 sector size should be used when an image is created.
1195 * systemd-repart now supports generating erofs file systems via
1196 CopyFiles= (a read-only file system similar to squashfs).
1198 * The Minimize= option was extended to accept "best" (which means the
1199 most minimal image possible, but may require multiple attempts) and
1200 "guess" (which means a reasonably small image).
1202 * The systemd-growfs binary now comes with a regular unit file template
1203 systemd-growfs@.service which can be instantiated directly for any
1204 desired file system. (Previously, the unit was generated dynamically
1205 by various generators, but no regular unit file template was
1208 Changes in journal tools:
1210 * Various systemd tools will append extra fields to log messages when
1211 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
1212 this includes information about D-Bus messages when sd-bus is used,
1213 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
1214 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
1215 Details of what is logged and when are subject to change.
1217 * The systemd-journald-audit.socket can now be disabled via the usual
1218 "systemctl disable" mechanism to stop collection of audit
1219 messages. Please note that it is not enabled statically anymore and
1220 must be handled by the preset/enablement logic in package
1221 installation scripts.
1223 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
1224 be used to curtail disk use by systemd-journal-remote. This is
1225 similar to the options supported by systemd-journald.
1227 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
1230 * When enrolling new keys systemd-cryptenroll now supports unlocking
1231 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
1232 password was strictly required to be specified.
1234 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
1235 (except for tokens with user verification, UV) to identify tokens
1236 before authentication. Multiple FIDO2 tokens can now be enrolled at
1237 the same time, and systemd-cryptsetup will automatically select one
1238 that corresponds to one of the available LUKS key slots.
1240 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
1241 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
1242 bank and number into which the volume key should be measured. This is
1243 automatically enabled for the encrypted root volume discovered and
1244 activated by systemd-gpt-auto-generator.
1246 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
1247 "noexec,nosuid,nodev".
1249 * systemd-gpt-auto-generator will now honour the rootfstype= and
1250 rootflags= kernel command line switches for root file systems it
1251 discovers, to match behaviour in case an explicit root fs is
1252 specified via root=.
1254 * systemd-pcrphase gained new options --machine-id and --file-system=
1255 to measure the machine-id and mount point information into PCR 15. New
1256 service unit files systemd-pcrmachine.service and
1257 systemd-pcrfs@.service have been added that invoke the tool with
1258 these switches during early boot.
1260 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
1261 with a success exit code even if no TPM device is detected.
1263 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
1264 making it harder to brute-force.
1266 Changes in other tools:
1268 * systemd-homed gained support for luksPbkdfForceIterations (the
1269 intended number of iterations for the PBKDF operation on LUKS).
1271 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
1272 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
1273 may now be used to specify additional arguments for mkfs when
1274 systemd-homed formats a file system.
1276 * systemd-hostnamed now exports the contents of
1277 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
1278 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
1279 unprivileged code to access those values.
1281 systemd-hostnamed also exports the SUPPORT_END= field from
1282 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
1283 this to show the status of the installed system.
1285 * systemd-measure gained an --append= option to sign multiple phase
1286 paths with different signing keys. This allows secrets to be
1287 accessible only in certain parts of the boot sequence. Note that
1288 'ukify' provides similar functionality in a more accessible form.
1290 * systemd-timesyncd will now write a structured log message with
1291 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
1292 on a on-disk timestamp, similarly to what it did when reaching
1293 synchronization via NTP.
1295 * systemd-timesyncd will now update the on-disk timestamp file on each
1296 boot at least once, making it more likely that the system time
1297 increases in subsequent boots.
1299 * systemd-vconsole-setup gained support for system/service credentials:
1300 vconsole.keymap/vconsole.keymap_toggle and
1301 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
1302 the similarly-named options in vconsole.conf.
1304 * systemd-localed will now save the XKB keyboard configuration to
1305 /etc/vconsole.conf, and also read it from there with a higher
1306 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
1307 file. Previously, this information was stored in the former file in
1308 converted form, and only in latter file in the original form. Tools
1309 which want to access keyboard configuration can now do so from a
1312 * systemd-resolved gained support for configuring the nameservers and
1313 search domains via kernel command line (nameserver=, domain=) and
1314 credentials (network.dns, network.search_domains).
1316 * systemd-resolved will now synthesize host names for the DNS stub
1317 addresses it supports. Specifically when "_localdnsstub" is resolved,
1318 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
1319 127.0.0.54 is returned.
1321 * systemd-notify will now send a "RELOADING=1" notification when called
1322 with --reloading, and "STOPPING=1" when called with --stopping. This
1323 can be used to implement notifications from units where it's easier
1324 to call a program than to use the sd-daemon library.
1326 * systemd-analyze's 'plot' command can now output its information in
1327 JSON, controlled via the --json= switch. Also, new --table, and
1328 --no-legend options have been added.
1330 * 'machinectl enable' will now automatically enable machines.target
1331 unit in addition to adding the machine unit to the target.
1333 Similarly, 'machinectl start|stop' gained a --now option to enable or
1334 disable the machine unit when starting or stopping it.
1336 * systemd-sysusers will now create /etc/ if it is missing.
1338 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
1339 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
1340 added to provide the new initial value for the new automated battery
1341 estimation functionality. If 'HibernateDelaySec=' is set to any value,
1342 the automated estimate (and thus the automated hibernation on low
1343 battery to avoid data loss) functionality will be disabled.
1345 * Default tmpfiles.d/ configuration will now automatically create
1346 credentials storage directory '/etc/credstore/' with the appropriate,
1347 secure permissions. If '/run/credstore/' exists, its permissions will
1348 be fixed too in case they are not correct.
1350 Changes in libsystemd and shared code:
1352 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
1353 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
1355 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
1356 128-bit ID in files such as /etc/machine-id has an invalid
1357 format. They also accept NULL as output parameter in more places,
1358 which is useful when the caller only wants to validate the inputs and
1359 does not need the output value.
1361 * sd-login gained new functions sd_pidfd_get_session(),
1362 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
1363 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
1364 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
1365 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
1366 but accept a PIDFD instead of a PID.
1368 * sd-path (and systemd-path) now export four new paths:
1369 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
1370 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
1371 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
1372 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
1374 * sd_notify() now supports AF_VSOCK as transport for notification
1375 messages (in addition to the existing AF_UNIX support). This is
1376 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
1378 * Detection of chroot() environments now works if /proc/ is not
1379 mounted. This affects systemd-detect-virt --chroot, but also means
1380 that systemd tools will silently skip various operations in such an
1383 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
1384 virtualization is now detected.
1386 Changes in the build system:
1388 * Standalone variants of systemd-repart and systemd-shutdown may now be
1389 built (if -Dstandalone=true).
1391 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
1392 example, allow scripts to conditionalize execution on AC power
1395 * The libp11kit library is now loaded through dlopen(3).
1397 Changes in the documentation:
1399 * Specifications that are not closely tied to systemd have moved to
1400 https://uapi-group.org/specifications/: the Boot Loader Specification
1401 and the Discoverable Partitions Specification.
1403 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
1404 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
1405 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
1406 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
1407 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
1408 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
1409 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
1410 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
1411 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
1412 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
1413 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
1414 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
1415 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
1416 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
1417 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
1418 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
1419 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
1420 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
1421 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
1422 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
1423 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
1424 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
1425 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
1426 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
1427 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
1428 reuben olinsky, Richard E. van der Luit, Richard Phibel,
1429 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
1430 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
1431 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
1432 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
1433 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
1434 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
1435 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
1436 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
1439 — Warsaw, 2023-02-15
1443 Announcements of Future Feature Removals:
1445 * We intend to remove cgroup v1 support from systemd release after the
1446 end of 2023. If you run services that make explicit use of cgroup v1
1447 features (i.e. the "legacy hierarchy" with separate hierarchies for
1448 each controller), please implement compatibility with cgroup v2 (i.e.
1449 the "unified hierarchy") sooner rather than later. Most of Linux
1450 userspace has been ported over already.
1452 * We intend to remove support for split-usr (/usr mounted separately
1453 during boot) and unmerged-usr (parallel directories /bin and
1454 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
1455 half of 2023, in the first release that falls into that time window.
1456 For more details, see:
1457 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1459 Compatibility Breaks:
1461 * ConditionKernelVersion= checks that use the '=' or '!=' operators
1462 will now do simple string comparisons (instead of version comparisons
1463 à la stverscmp()). Version comparisons are still done for the
1464 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
1465 specified, a shell-style glob match is now done. This creates a minor
1466 incompatibility compared to older systemd versions when the '*', '?',
1467 '[', ']' characters are used, as these will now match as shell globs
1468 instead of literally. Given that kernel version strings typically do
1469 not include these characters we expect little breakage through this
1472 * The service manager will now read the SELinux label used for SELinux
1473 access checks from the unit file at the time it loads the file.
1474 Previously, the label would be read at the moment of the access
1475 check, which was problematic since at that time the unit file might
1476 already have been updated or removed.
1480 * systemd-measure is a new tool for calculating and signing expected
1481 TPM2 PCR values for a given unified kernel image (UKI) booted via
1482 sd-stub. The public key used for the signature and the signed
1483 expected PCR information can be embedded inside the UKI. This
1484 information can be extracted from the UKI by external tools and code
1485 in the image itself and is made available to userspace in the booted
1488 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
1489 updated to make use of this information if available in the booted
1490 kernel: when locking an encrypted volume/credential to the TPM
1491 systemd-cryptenroll/systemd-creds will use the public key to bind the
1492 volume/credential to any kernel that carries PCR information signed
1493 by the same key pair. When unlocking such volumes/credentials
1494 systemd-cryptsetup/systemd-creds will use the signature embedded in
1495 the booted UKI to gain access.
1497 Binding TPM-based disk encryption to public keys/signatures of PCR
1498 values — instead of literal PCR values — addresses the inherent
1499 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
1500 disks remain accessible even if the UKI is updated, without any TPM
1501 specific preparation during the OS update — as long as each UKI
1502 carries the necessary PCR signature information.
1504 Net effect: if you boot a properly prepared kernel, TPM-bound disk
1505 encryption now defaults to be locked to kernels which carry PCR
1506 signatures from the same key pair. Example: if a hypothetical distro
1507 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
1508 by default – bound to only FooOS kernels, and encrypted volumes bound
1509 to the TPM cannot be unlocked on kernels from other sources. (But do
1510 note this behaviour requires preparation/enabling in the UKI, and of
1511 course users can always enroll non-TPM ways to unlock the volume.)
1513 * systemd-pcrphase is a new tool that is invoked at six places during
1514 system runtime, and measures additional words into TPM2 PCR 11, to
1515 mark milestones of the boot process. This allows binding access to
1516 specific TPM2-encrypted secrets to specific phases of the boot
1517 process. (Example: LUKS2 disk encryption key only accessible in the
1518 initrd, but not later.)
1520 Changes in systemd itself, i.e. the manager and units
1522 * The cpu controller is delegated to user manager units by default, and
1523 CPUWeight= settings are applied to the top-level user slice units
1524 (app.slice, background.slice, session.slice). This provides a degree
1525 of resource isolation between different user services competing for
1528 * Systemd can optionally do a full preset in the "first boot" condition
1529 (instead of just enable-only). This behaviour is controlled by the
1530 compile-time option -Dfirst-boot-full-preset. Right now it defaults
1531 to 'false', but the plan is to switch it to 'true' for the subsequent
1534 * Drop-ins are now allowed for transient units too.
1536 * Systemd will set the taint flag 'support-ended' if it detects that
1537 the OS image is past its end-of-support date. This date is declared
1538 in a new /etc/os-release field SUPPORT_END= described below.
1540 * Two new settings ConditionCredential= and AssertCredential= can be
1541 used to skip or fail units if a certain system credential is not
1544 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
1546 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
1547 specify the SMACK security label to use when not specified in a unit
1550 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
1551 specify the default timeout when waiting for device units to
1554 * C.UTF-8 is used as the default locale if nothing else has been
1557 * [Condition|Assert]Firmware= have been extended to support certain
1558 SMBIOS fields. For example
1560 ConditionFirmware=smbios-field(board_name = "Custom Board")
1562 conditionalizes the unit to run only when
1563 /sys/class/dmi/id/board_name contains "Custom Board" (without the
1566 * ConditionFirstBoot= now correctly evaluates as true only during the
1567 boot phase of the first boot. A unit executed later, after booting
1568 has completed, will no longer evaluate this condition as true.
1570 * Socket units will now create sockets in the SELinuxContext= of the
1571 associated service unit, if any.
1573 * Boot phase transitions (start initrd → exit initrd → boot complete →
1574 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
1575 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
1576 unsealed only in the initrd.
1578 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
1579 also be provided to ExecStartPre= processes.
1581 * Various units are now correctly ordered against
1582 initrd-switch-root.target where previously a conflict without
1583 ordering was configured. A stop job for those units would be queued,
1584 but without the ordering it could be executed only after
1585 initrd-switch-root.service, leading to units not being restarted in
1586 the host system as expected.
1588 * In order to fully support the IPMI watchdog driver, which has not yet
1589 been ported to the new common watchdog device interface,
1590 /dev/watchdog0 will be tried first and systemd will silently fallback
1591 to /dev/watchdog if it is not found.
1593 * New watchdog-related D-Bus properties are now published by systemd:
1594 WatchdogDevice, WatchdogLastPingTimestamp,
1595 WatchdogLastPingTimestampMonotonic.
1597 * At shutdown, API virtual files systems (proc, sys, etc.) will be
1600 * At shutdown, systemd will now log about processes blocking unmounting
1603 * A new meson build option 'clock-valid-range-usec-max' was added to
1604 allow disabling system time correction if RTC returns a timestamp far
1607 * Propagated restart jobs will no longer be discarded while a unit is
1610 * PID 1 will now import system credentials from SMBIOS Type 11 fields
1611 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
1612 simple, fast and generic path for supplying credentials to a VM,
1613 without involving external tools such as cloud-init/ignition.
1615 * The CPUWeight= setting of unit files now accepts a new special value
1616 "idle", which configures "idle" level scheduling for the unit.
1618 * Service processes that are activated due to a .timer or .path unit
1619 triggering will now receive information about this via environment
1620 variables. Note that this is information is lossy, as activation
1621 might be coalesced and only one of the activating triggers will be
1622 reported. This is hence more suited for debugging or tracing rather
1623 than for behaviour decisions.
1625 * The riscv_flush_icache(2) system call has been added to the list of
1626 system calls allowed by default when SystemCallFilter= is used.
1628 * The selinux context derived from the target executable, instead of
1629 'init_t' used for the manager itself, is now used when creating
1630 listening sockets for units that specify SELinuxContextFromNet=yes.
1632 Changes in sd-boot, bootctl, and the Boot Loader Specification:
1634 * The Boot Loader Specification has been cleaned up and clarified.
1635 Various corner cases in version string comparisons have been fixed
1636 (e.g. comparisons for empty strings). Boot counting is now part of
1637 the main specification.
1639 * New PCRs measurements are performed during boot: PCR 11 for the
1640 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
1641 took place this is now reported to userspace via the new
1642 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
1644 * As before, systemd-stub will measure kernel parameters and system
1645 credentials into PCR 12. It will now report this fact via the
1646 StubPcrKernelParameters EFI variable to userspace.
1648 * The UEFI monotonic boot counter is now included in the updated random
1649 seed file maintained by sd-boot, providing some additional entropy.
1651 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
1652 of arranging the image manually and jumping to the kernel entry
1653 point. sd-stub also installs a temporary UEFI SecurityOverride to
1654 allow the (unsigned) nested image to be booted. This is safe because
1655 the outer (signed) stub+kernel binary must have been verified before
1656 the stub was executed.
1658 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
1659 is now supported by sd-boot.
1661 * bootctl gained a bunch of new options: --all-architectures to install
1662 binaries for all supported EFI architectures, --root= and --image=
1663 options to operate on a directory or disk image, and
1664 --install-source= to specify the source for binaries to install,
1665 --efi-boot-option-description= to control the name of the boot entry.
1667 * The sd-boot stub exports a StubFeatures flag, which is used by
1668 bootctl to show features supported by the stub that was used to boot.
1670 * The PE section offsets that are used by tools that assemble unified
1671 kernel images have historically been hard-coded. This may lead to
1672 overlapping PE sections which may break on boot. The UKI will now try
1673 to detect and warn about this.
1675 Any tools that assemble UKIs must update to calculate these offsets
1676 dynamically. Future sd-stub versions may use offsets that will not
1677 work with the currently used set of hard-coded offsets!
1679 * sd-stub now accepts (and passes to the initrd and then to the full
1680 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
1681 signatures of expected PCR values, to allow sealing secrets via the
1682 TPM2 against pre-calculated PCR measurements.
1684 Changes in the hardware database:
1686 * 'systemd-hwdb query' now supports the --root= option.
1688 Changes in systemctl:
1690 * systemctl now supports --state= and --type= options for the 'show'
1693 * systemctl gained a new verb 'list-automounts' to list automount
1696 * systemctl gained support for a new --image= switch to be able to
1697 operate on the specified disk image (similar to the existing --root=
1698 which operates relative to some directory).
1700 Changes in systemd-networkd:
1702 * networkd can set Linux NetLabel labels for integration with the
1703 network control in security modules via a new NetLabel= option.
1705 * The RapidCommit= is (re-)introduced to enable faster configuration
1706 via DHCPv6 (RFC 3315).
1708 * networkd gained a new option TCPCongestionControlAlgorithm= that
1709 allows setting a per-route TCP algorithm.
1711 * networkd gained a new option KeepFileDescriptor= to allow keeping a
1712 reference (file descriptor) open on TUN/TAP interfaces, which is
1713 useful to avoid link flaps while the underlying service providing the
1714 interface is being serviced.
1716 * RouteTable= now also accepts route table names.
1718 Changes in systemd-nspawn:
1720 * The --bind= and --overlay= options now support relative paths.
1722 * The --bind= option now supports a 'rootidmap' value, which will
1723 use id-mapped mounts to map the root user inside the container to the
1724 owner of the mounted directory on the host.
1726 Changes in systemd-resolved:
1728 * systemd-resolved now persists DNSOverTLS in its state file too. This
1729 fixes a problem when used in combination with NetworkManager, which
1730 sends the setting only once, causing it to be lost if resolved was
1731 restarted at any point.
1733 * systemd-resolved now exposes a varlink socket at
1734 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
1735 root. Processed DNS requests in a JSON format will be published to
1736 any clients connected to this socket.
1738 resolvectl gained a 'monitor' verb to make use of this.
1740 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
1741 instead of returning SERVFAIL, as per RFC:
1742 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
1744 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
1745 is still supported.)
1747 Changes in libsystemd and other libraries:
1749 * libsystemd now exports sd_bus_error_setfv() (a convenience function
1750 for setting bus errors), sd_id128_string_equal (a convenience
1751 function for 128-bit ID string comparisons), and
1752 sd_bus_message_read_strv_extend() (a function to incrementally read
1755 * libsystemd now exports sd_device_get_child_first()/_next() as a
1756 high-level interface for enumerating child devices. It also supports
1757 sd_device_new_child() for opening a child device given a device
1760 * libsystemd now exports sd_device_monitor_set()/get_description()
1761 which allow setting a custom description that will be used in log
1762 messages by sd_device_monitor*.
1764 * Private shared libraries (libsystemd-shared-nnn.so,
1765 libsystemd-core-nnn.so) are now installed into arch-specific
1766 directories to allow multi-arch installs.
1768 * A new sd-gpt.h header is now published, listing GUIDs from the
1769 Discoverable Partitions specification. For more details see:
1770 https://systemd.io/DISCOVERABLE_PARTITIONS/
1772 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
1773 database given an explicit path to the file.
1775 * The signal number argument to sd_event_add_signal() now can now be
1776 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
1777 be automatically invoked to block the specified signal. This is
1778 useful to simplify invocations as the caller doesn't have to do this
1781 * A new convenience call sd_event_set_signal_exit() has been added to
1782 sd-event to set up signal handling so that the event loop
1783 automatically terminates cleanly on SIGTERM/SIGINT.
1785 Changes in other components:
1787 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
1788 can now be provided via the credential mechanism.
1790 * systemd-analyze gained a new verb 'compare-versions' that implements
1791 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
1792 'dpkg --compare-versions').
1794 * 'systemd-analyze dump' is extended to accept glob patterns for unit
1795 names to limit the output to matching units.
1797 * tmpfiles.d/ lines can read file contents to write from a credential.
1798 The new modifier char '^' is used to specify that the argument is a
1799 credential name. This mechanism is used to automatically populate
1800 /etc/motd, /etc/issue, and /etc/hosts from credentials.
1802 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
1803 an inode if the specification is prefixed with ':' and the inode
1806 * Default tmpfiles.d/ configuration now carries a line to automatically
1807 use an 'ssh.authorized_keys.root' credential if provided to set up
1808 the SSH authorized_keys file for the root user.
1810 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
1813 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
1814 in base64. This is useful to write arbitrary binary data into files.
1816 * The pkgconfig and rpm macros files now export the directory for user
1817 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
1819 * Detection of Apple Virtualization and detection of Parallels and
1820 KubeVirt virtualization on non-x86 archs have been added.
1822 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
1823 user when their system will become unsupported.
1825 * When performing suspend-then-hibernate, the system will estimate the
1826 discharge rate and use that to set the delay until hibernation and
1827 hibernate immediately instead of suspending when running from a
1828 battery and the capacity is below 5%.
1830 * systemd-sysctl gained a --strict option to fail when a sysctl
1831 setting is unknown to the kernel.
1833 * machinectl supports --force for the 'copy-to' and 'copy-from'
1836 * coredumpctl gained the --root and --image options to look for journal
1837 files under the specified root directory, image, or block device.
1839 * 'journalctl -o' and similar commands now implement a new output mode
1840 "short-delta". It is similar to "short-monotonic", but also shows the
1841 time delta between subsequent messages.
1843 * journalctl now respects the --quiet flag when verifying consistency
1846 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
1847 will indicate whether a message was logged in the 'initrd' phase or
1848 in the 'system' phase of the boot process.
1850 * Journal files gained a new compatibility flag
1851 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
1852 to the storage format that allow reducing size on disk. As with other
1853 compatibility flags, older journalctl versions will not be able to
1854 read journal files using this new format. The environment variable
1855 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
1856 disable this functionality. It is enabled by default.
1858 * systemd-run's --working-directory= switch now works when used in
1859 combination with --scope.
1861 * portablectl gained a --force flag to skip certain sanity checks. This
1862 is implemented using new flags accepted by systemd-portabled for the
1863 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
1864 flag now means that the attach/detach checks whether the units are
1865 already present and running will be skipped. Similarly,
1866 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
1867 image name matches the name declared inside of the image will be
1868 skipped. Callers must be sure to do those checks themselves if
1871 * systemd-portabled will now use the original filename to check
1872 extension-release.NAME for correctness, in case it is passed a
1875 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
1878 * sysext's extension-release files now support '_any' as a special
1879 value for the ID= field, to allow distribution-independent extensions
1880 (e.g.: fully statically compiled binaries, scripts). It also gained
1881 support for a new ARCHITECTURE= field that may be used to explicitly
1882 restrict an image to hosts of a specific architecture.
1884 * systemd-repart now supports creating squashfs partitions. This
1885 requires mksquashfs from squashfs-tools.
1887 * systemd-repart gained a --split flag to also generate split
1888 artifacts, i.e. a separate file for each partition. This is useful in
1889 conjunction with systemd-sysupdate or other tools, or to generate
1890 split dm-verity artifacts.
1892 * systemd-repart is now able to generate dm-verity partitions, including
1895 * systemd-repart can now set a partition UUID to zero, allowing it to
1896 be filled in later, such as when using verity partitions.
1898 * systemd-repart now supports drop-ins for its configuration files.
1900 * Package metadata logged by systemd-coredump in the system journal is
1903 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
1905 * systemd-oomd now automatically links against libatomic, if available.
1907 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
1910 * scope units now also provide oom-kill status.
1912 * systemd-pstore will now try to load only the efi_pstore kernel module
1913 before running, ensuring that pstore can be used.
1915 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
1916 session after a preconfigure timeout.
1918 * systemd-homed will now wait up to 30 seconds for workers to terminate,
1919 rather than indefinitely.
1921 * homectl gained a new '--luks-sector-size=' flag that allows users to
1922 select the preferred LUKS sector size. Must be a power of 2 between 512
1923 and 4096. systemd-userdbd records gained a corresponding field.
1925 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
1926 variable when generating the 'sp_lstchg' field, to ensure an image
1927 build can be reproducible.
1929 * 'udevadm wait' will now listen to kernel uevents too when called with
1932 * When naming network devices udev will now consult the Devicetree
1933 "alias" fields for the device.
1935 * systemd-udev will now create infiniband/by-path and
1936 infiniband/by-ibdev links for Infiniband verbs devices.
1938 * systemd-udev-trigger.service will now also prioritize input devices.
1940 * ConditionACPower= and systemd-ac-power will now assume the system is
1941 running on AC power if no battery can be found.
1943 * All features and tools using the TPM2 will now communicate with it
1944 using a bind key. Beforehand, the tpm2 support used encrypted sessions
1945 by creating a primary key that was used to encrypt traffic. This
1946 creates a problem as the key created for encrypting the traffic could
1947 be faked by an active interposer on the bus. In cases when a pin is
1948 used, a bind key will be used. The pin is used as the auth value for
1949 the seal key, aka the disk encryption key, and that auth value will be
1950 used in the session establishment. An attacker would need the pin
1951 value to create the secure session and thus an active interposer
1952 without the pin cannot interpose on TPM2 traffic.
1954 * systemd-growfs no longer requires udev to run.
1956 * systemd-backlight now will better support systems with multiple
1959 * systemd-cryptsetup's keyfile-timeout= option now also works when a
1960 device is used as a keyfile.
1962 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
1963 unlocking key from a key file (instead of prompting the user). Note
1964 that this is the key for unlocking the volume in order to be able to
1965 enroll a new key, but it is not the key that is enrolled.
1967 * systemd-dissect gained a new --umount switch that will safely and
1968 synchronously unmount all partitions of an image previously mounted
1969 with 'systemd-dissect --mount'.
1971 * When using gcrypt, all systemd tools and services will now configure
1972 it to prefer the OS random number generator if present.
1974 * All example code shipped with documentation has been relicensed from CC0
1977 * Unit tests will no longer fail when running on a system without
1980 Experimental features:
1982 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
1983 and bpftool >= 7.0).
1985 * sd-boot can automatically enroll SecureBoot keys from files found on
1986 the ESP. This enrollment can be either automatic ('force' mode) or
1987 controlled by the user ('manual' mode). It is sufficient to place the
1988 SecureBoot keys in the right place in the ESP and they will be picked
1989 up by sd-boot and shown in the boot menu.
1991 * The mkosi config in systemd gained support for automatically
1992 compiling a kernel with the configuration appropriate for testing
1993 systemd. This may be useful when developing or testing systemd in
1994 tandem with the kernel.
1996 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
1997 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
1998 Alexander Graf, Alexander Shopov, Alexander Wilson,
1999 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
2000 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
2001 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
2002 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
2003 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
2004 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
2005 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
2006 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
2007 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
2008 David Jaša, David Rheinsberg, David Seifert, David Tardon,
2009 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
2010 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
2011 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
2012 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
2013 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
2014 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
2015 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
2016 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
2017 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
2018 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
2019 JeroenHD, jiangchuangang, João Loureiro,
2020 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
2021 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
2022 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
2023 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
2024 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
2025 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
2026 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
2027 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
2028 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
2029 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
2030 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
2031 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
2032 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
2033 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
2034 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
2035 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
2036 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
2037 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
2038 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
2039 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
2040 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
2041 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
2042 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
2043 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
2045 – The Great Beyond, 2022-10-31 👻
2049 Backwards-incompatible changes:
2051 * The minimum kernel version required has been bumped from 3.13 to 4.15,
2052 and CLOCK_BOOTTIME is now assumed to always exist.
2054 * C11 with GNU extensions (aka "gnu11") is now used to build our
2055 components. Public API headers are still restricted to ISO C89.
2057 * In v250, a systemd-networkd feature that automatically configures
2058 routes to addresses specified in AllowedIPs= was added and enabled by
2059 default. However, this causes network connectivity issues in many
2060 existing setups. Hence, it has been disabled by default since
2061 systemd-stable 250.3. The feature can still be used by explicitly
2062 configuring RouteTable= setting in .netdev files.
2064 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
2065 when a Condition*= check does not succeed, restoring the JobRemoved
2066 signal to the behaviour it had before v250.
2068 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
2069 GetImageMetadataWithExtensions() have been fixed to provide an extra
2070 return parameter, containing the actual extension release metadata.
2071 The current implementation was judged to be broken and unusable, and
2072 thus the usual procedure of adding a new set of methods was skipped,
2073 and backward compatibility broken instead on the assumption that
2074 nobody can be affected given the current state of this interface.
2076 * All kernels supported by systemd mix bytes returned by RDRAND (or
2077 similar) into the entropy pool at early boot. This means that on
2078 those systems, even if /dev/urandom is not yet initialized, it still
2079 returns bytes that are of at least RDRAND quality. For that reason,
2080 we no longer have reason to invoke RDRAND from systemd itself, which
2081 has historically been a source of bugs. Furthermore, kernels ≥5.6
2082 provide the getrandom(GRND_INSECURE) interface for returning random
2083 bytes before the entropy pool is initialized without warning into
2084 kmsg, which is what we attempt to use if available. systemd's direct
2085 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
2086 running an older kernel may experience kmsg warnings that were not
2087 seen with 250. For newer kernels, non-x86 systems, or older x86
2088 systems, there should be no visible changes.
2090 * sd-boot will now measure the kernel command line into TPM PCR 12
2091 rather than PCR 8. This improves usefulness of the measurements on
2092 systems where sd-boot is chainloaded from Grub. Grub measures all
2093 commands its executes into PCR 8, which makes it very hard to use
2094 reasonably, hence separate ourselves from that and use PCR 12
2095 instead, which is what certain Ubuntu editions already do. To retain
2096 compatibility with systems running older systemd systems a new meson
2097 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
2098 If enabled, the measurement is done twice: into the new-style PCR 12
2099 *and* the old-style PCR 8. It's strongly advised to migrate all users
2100 to PCR 12 for this purpose in the long run, as we intend to remove
2101 this compatibility feature in two years' time.
2103 * busctl capture now writes output in the newer pcapng format instead
2106 * A udev rule that imported hwdb matches for USB devices with lowercase
2107 hexadecimal vendor/product ID digits was added in systemd 250. This
2108 has been reverted, since uppercase hexadecimal digits are supposed to
2109 be used, and we already had a rule with the appropriate match.
2111 Users might need to adjust their local hwdb entries.
2113 * arch_prctl(2) has been moved to the @default set in the syscall filters
2114 (as exposed via the SystemCallFilter= setting in service unit files).
2115 It is apparently used by the linker now.
2117 * The tmpfiles entries that create the /run/systemd/netif directory and
2118 its subdirectories were moved from tmpfiles.d/systemd.conf to
2119 tmpfiles.d/systemd-network.conf.
2121 Users might need to adjust their files that override tmpfiles.d/systemd.conf
2122 to account for this change.
2124 * The requirement for Portable Services images to contain a well-formed
2125 os-release file (i.e.: contain at least an ID field) is now enforced.
2126 This applies to base images and extensions, and also to systemd-sysext.
2128 Changes in the Boot Loader Specification, kernel-install and sd-boot:
2130 * kernel-install's and bootctl's Boot Loader Specification Type #1
2131 entry generation logic has been reworked. The user may now pick
2132 explicitly by which "token" string to name the installation's boot
2133 entries, via the new /etc/kernel/entry-token file or the new
2134 --entry-token= switch to bootctl. By default — as before — the
2135 entries are named after the local machine ID. However, in "golden
2136 image" environments, where the machine ID shall be initialized on
2137 first boot (as opposed to at installation time before first boot) the
2138 machine ID will not be available at build time. In this case the
2139 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
2140 file) may be used to override the "token" for the entries, for
2141 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
2142 make the OS images independent of any machine ID, and ensure that the
2143 images will not carry any identifiable information before first boot,
2144 but on the other hand means that multiple parallel installations of
2145 the very same image on the same disk cannot be supported.
2147 Summary: if you are building golden images that shall acquire
2148 identity information exclusively on first boot, make sure to both
2149 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
2150 value of the IMAGE_ID= or ID= field of /etc/os-release or another
2151 suitable identifier before deploying the image.
2153 * The Boot Loader Specification has been extended with
2154 /loader/entries.srel file located in the EFI System Partition (ESP)
2155 that disambiguates the format of the entries in the /loader/entries/
2156 directory (in order to discern them from incompatible uses of this
2157 directory by other projects). For entries that follow the
2158 Specification, the string "type1" is stored in this file.
2160 bootctl will now write this file automatically when installing the
2161 systemd-boot boot loader.
2163 * kernel-install supports a new initrd_generator= setting in
2164 /etc/kernel/install.conf, that is exported as
2165 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
2166 allows choosing different initrd generators.
2168 * kernel-install will now create a "staging area" (an initially-empty
2169 directory to gather files for a Boot Loader Specification Type #1
2170 entry). The path to this directory is exported as
2171 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
2172 drop files there instead of writing them directly to the final
2173 location. kernel-install will move them when all files have been
2174 prepared successfully.
2176 * New option sort-key= has been added to the Boot Loader Specification
2177 to override the sorting order of the entries in the boot menu. It is
2178 read by sd-boot and bootctl, and will be written by kernel-install,
2179 with the default value of IMAGE_ID= or ID= fields from
2180 os-release. Together, this means that on multiboot installations,
2181 entries should be grouped and sorted in a predictable way.
2183 * The sort order of boot entries has been updated: entries which have
2184 the new field sort-key= are sorted by it first, and all entries
2185 without it are ordered later. After that, entries are sorted by
2186 version so that newest entries are towards the beginning of the list.
2188 * The kernel-install tool gained a new 'inspect' verb which shows the
2189 paths and other settings used.
2191 * sd-boot can now optionally beep when the menu is shown and menu
2192 entries are selected, which can be useful on machines without a
2193 working display. (Controllable via a loader.conf setting.)
2195 * The --make-machine-id-directory= switch to bootctl has been replaced
2196 by --make-entry-directory=, given that the entry directory is not
2197 necessarily named after the machine ID, but after some other suitable
2198 ID as selected via --entry-token= described above. The old name of
2199 the option is still understood to maximize compatibility.
2201 * 'bootctl list' gained support for a new --json= switch to output boot
2202 menu entries in JSON format.
2204 * 'bootctl is-installed' now supports the --graceful, and various verbs
2205 omit output with the new option --quiet.
2207 Changes in systemd-homed:
2209 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
2210 of activated home directories it manages (if the kernel and selected
2211 file systems support it). So far it mapped three UID ranges: the
2212 range from 0…60000, the user's own UID, and the range 60514…65534,
2213 leaving everything else unmapped (in other words, the 16-bit UID range
2214 is mapped almost fully, with the exception of the UID subrange used
2215 for systemd-homed users, with one exception: the user's own UID).
2216 Unmapped UIDs may not be used for file ownership in the home
2217 directory — any chown() attempts with them will fail. With this
2218 release a fourth range is added to these mappings:
2219 524288…1879048191. This range is the UID range intended for container
2222 https://systemd.io/UIDS-GIDS
2224 This range may be used for container managers that place container OS
2225 trees in the home directory (which is a questionable approach, for
2226 quota, permission, SUID handling and network file system
2227 compatibility reasons, but nonetheless apparently commonplace). Note
2228 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
2229 UID assignments from the range are not managed or mapped by
2230 `systemd-homed`, and must be managed with other mechanisms, in the
2231 context of the local system.
2233 Typically, a better approach to user namespacing in relevant
2234 container managers would be to leave container OS trees on disk at
2235 UID offset 0, but then map them to a dynamically allocated runtime
2236 UID range via another UID mount map at container invocation
2237 time. That way user namespace UID ranges become strictly a runtime
2238 concept, and do not leak into persistent file systems, persistent
2239 user databases or persistent configuration, thus greatly simplifying
2240 handling, and improving compatibility with home directories intended
2241 to be portable like the ones managed by systemd-homed.
2243 Changes in shared libraries:
2245 * A new libsystemd-core-<version>.so private shared library is
2246 installed under /usr/lib/systemd/system, mirroring the existing
2247 libsystemd-shared-<version>.so library. This allows the total
2248 installation size to be reduced by binary code reuse.
2250 * The <version> tag used in the name of libsystemd-shared.so and
2251 libsystemd-core.so can be configured via the meson option
2252 'shared-lib-tag'. Distributions may build subsequent versions of the
2253 systemd package with unique tags (e.g. the full package version),
2254 thus allowing multiple installations of those shared libraries to be
2255 available at the same time. This is intended to fix an issue where
2256 programs that link to those libraries would fail to execute because
2257 they were installed earlier or later than the appropriate version of
2260 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
2261 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
2262 format instead of as a simple series of hex characters.
2264 * The sd-device API gained two new calls sd_device_new_from_devname()
2265 and sd_device_new_from_path() which permit allocating an sd_device
2266 object from a device node name or file system path.
2268 * sd-device also gained a new call sd_device_open() which will open the
2269 device node associated with a device for which an sd_device object
2270 has been allocated. The call is supposed to address races around
2271 device nodes being removed/recycled due to hotplug events, or media
2272 change events: the call checks internally whether the major/minor of
2273 the device node and the "diskseq" (in case of block devices) match
2274 with the metadata loaded in the sd_device object, thus ensuring that
2275 the device once opened really matches the provided sd_device object.
2277 Changes in PID1, systemctl, and systemd-oomd:
2279 * A new set of service monitor environment variables will be passed to
2280 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
2281 handler unit as OnFailure=/OnSuccess=. The variables are:
2282 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
2283 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
2284 handler needs to watch multiple units, use a templated handler.
2286 * A new ExtensionDirectories= setting in service unit files allows
2287 system extensions to be loaded from a directory. (It is similar to
2288 ExtensionImages=, but takes paths to directories, instead of
2291 'portablectl attach --extension=' now also accepts directory paths.
2293 * The user.delegate and user.invocation_id extended attributes on
2294 cgroups are used in addition to trusted.delegate and
2295 trusted.invocation_id. The latter pair requires privileges to set,
2296 but the former doesn't and can be also set by the unprivileged user
2299 (Only supported on kernels ≥5.6.)
2301 * Units that were killed by systemd-oomd will now have a service result
2302 of 'oom-kill'. The number of times a service was killed is tallied
2303 in the 'user.oomd_ooms' extended attribute.
2305 The OOMPolicy= unit file setting is now also honoured by
2308 * In unit files the new %y/%Y specifiers can be used to refer to
2309 normalized unit file path, which is particularly useful for symlinked
2312 The new %q specifier resolves to the pretty hostname
2313 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
2315 The new %d specifier resolves to the credentials directory of a
2316 service (same as $CREDENTIALS_DIRECTORY).
2318 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
2319 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
2320 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
2321 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
2322 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
2323 MountFlags= service settings now also work in unprivileged user
2324 services, i.e. those run by the user's --user service manager, as long
2325 as user namespaces are enabled on the system.
2327 * Services with Restart=always and a failing ExecCondition= will no
2328 longer be restarted, to bring ExecCondition= behaviour in line with
2329 Condition*= settings.
2331 * LoadCredential= now accepts a directory as the argument; all files
2332 from the directory will be loaded as credentials.
2334 * A new D-Bus property ControlGroupId is now exposed on service units,
2335 that encapsulates the service's numeric cgroup ID that newer kernels
2336 assign to each cgroup.
2338 * PID 1 gained support for configuring the "pre-timeout" of watchdog
2339 devices and the associated governor, via the new
2340 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
2341 options in /etc/systemd/system.conf.
2343 * systemctl's --timestamp= option gained a new choice "unix", to show
2344 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
2346 * A new "taint" flag named "old-kernel" is introduced which is set when
2347 the kernel systemd runs on is older then the current baseline version
2348 (see above). The flag is shown in "systemctl status" output.
2350 * Two additional taint flags "short-uid-range" and "short-gid-range"
2351 have been added as well, which are set when systemd notices it is run
2352 within a userns namespace that does not define the full 0…65535 UID
2355 * A new "unmerged-usr" taint flag has been added that is set whenever
2356 running on systems where /bin/ + /sbin/ are *not* symlinks to their
2357 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
2360 * Generators invoked by PID 1 will now have a couple of useful
2361 environment variables set describing the execution context a
2362 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
2363 system service manager, or from the per-user service
2364 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
2365 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
2366 systemd considers the current boot to be a "first"
2367 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
2368 detected and which type of hypervisor/container
2369 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
2370 kernel is built for.
2372 * PID 1 will now automatically pick up system credentials from qemu's
2373 fw_cfg interface, thus allowing passing arbitrary data into VM
2374 systems similar to how this is already supported for passing them
2375 into `systemd-nspawn` containers. Credentials may now also be passed
2376 in via the new kernel command line option `systemd.set_credential=`
2377 (note that kernel command line options are world-readable during
2378 runtime, and only useful for credentials that require no
2379 confidentiality). The credentials that can be passed to unified
2380 kernels that use the `systemd-stub` UEFI stub are now similarly
2381 picked up automatically. Automatic importing of system credentials
2382 this way can be turned off via the new
2383 `systemd.import_credentials=no` kernel command line option.
2385 * LoadCredential= will now automatically look for credentials in the
2386 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
2387 the argument is not an absolute path. Similarly,
2388 LoadCredentialEncrypted= will check the same directories plus
2389 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
2390 /usr/lib/credstore.encrypted/. The idea is to use those directories
2391 as the system-wide location for credentials that services should pick
2394 * System and service credentials are described in great detail in a new
2397 https://systemd.io/CREDENTIALS
2399 Changes in systemd-journald:
2401 * The journal JSON export format has been added to listed of stable
2402 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
2404 * journalctl --list-boots now supports JSON output and the --reverse option.
2406 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
2407 updated, BUILDING_IMAGES is new:
2409 https://systemd.io/JOURNAL_EXPORT_FORMATS
2410 https://systemd.io/BUILDING_IMAGES
2414 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
2415 calculators, etc.), the other AV production devices (DJ tables,
2416 keypads, etc.) that should accessible to the seat owner user by
2419 * udevadm trigger gained a new --prioritized-subsystem= option to
2420 process certain subsystems (and all their parent devices) earlier.
2422 systemd-udev-trigger.service now uses this new option to trigger
2423 block and TPM devices first, hopefully making the boot a bit faster.
2425 * udevadm trigger now implements --type=all, --initialized-match,
2426 --initialized-nomatch to trigger both subsystems and devices, only
2427 already-initialized devices, and only devices which haven't been
2428 initialized yet, respectively.
2430 * udevadm gained a new "wait" command for safely waiting for a specific
2431 device to show up in the udev device database. This is useful in
2432 scripts that asynchronously allocate a block device (e.g. through
2433 repartitioning, or allocating a loopback device or similar) and need
2434 to synchronize on the creation to complete.
2436 * udevadm gained a new "lock" command for locking one or more block
2437 devices while formatting it or writing a partition table to it. It is
2438 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
2439 usable in scripts dealing with block devices.
2441 * udevadm info will show a couple of additional device fields in its
2442 output, and will not apply a limited set of coloring to line types.
2444 * udevadm info --tree will now show a tree of objects (i.e. devices and
2445 suchlike) in the /sys/ hierarchy.
2447 * Block devices will now get a new set of device symlinks in
2448 /dev/disk/by-diskseq/<nr>, which may be used to reference block
2449 device nodes via the kernel's "diskseq" value. Note that this does
2450 not guarantee that opening a device by a symlink like this will
2451 guarantee that the opened device actually matches the specified
2452 diskseq value. To be safe against races, the actual diskseq value of
2453 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
2454 the one in the symlink path.
2456 * .link files gained support for setting MDI/MID-X on a link.
2458 * .link files gained support for [Match] Firmware= setting to match on
2459 the device firmware description string. By mistake, it was previously
2460 only supported in .network files.
2462 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
2463 and [SR-IOV] section to configure SR-IOV virtual functions.
2465 Changes in systemd-networkd:
2467 * The default scope for unicast routes configured through [Route]
2468 section is changed to "link", to make the behavior consistent with
2469 "ip route" command. The manual configuration of [Route] Scope= is
2472 * A new unit systemd-networkd-wait-online@<interface>.service has been
2473 added that can be used to wait for a specific network interface to be
2476 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
2477 that configures the eponymous kernel attribute on the bridge.
2479 * .netdev files now can be used to create virtual WLAN devices, and
2480 configure various settings on them, via the [WLAN] section.
2482 * .link/.network files gained support for [Match] Kind= setting to match
2483 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
2485 This value is also shown by 'networkctl status'.
2487 * The Local= setting in .netdev files for various virtual network
2488 devices gained support for specifying, in addition to the network
2489 address, the name of a local interface which must have the specified
2492 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
2493 files, to configure tunnels in external mode (a.k.a. collect metadata
2496 * [Network] L2TP= setting was removed. Please use interface specifier in
2497 Local= setting in .netdev files of corresponding L2TP interface.
2499 * New [DHCPServer] BootServerName=, BootServerAddress=, and
2500 BootFilename= settings can be used to configure the server address,
2501 server name, and file name sent in the DHCP packet (e.g. to configure
2504 Changes in systemd-resolved:
2506 * systemd-resolved is started earlier (in sysinit.target), so it
2507 available earlier and will also be started in the initrd if installed
2510 Changes in disk encryption:
2512 * systemd-cryptenroll can now control whether to require the user to
2513 enter a PIN when using TPM-based unlocking of a volume via the new
2514 --tpm2-with-pin= option.
2516 Option tpm2-pin= can be used in /etc/crypttab.
2518 * When unlocking devices via TPM, TPM2 parameter encryption is now
2519 used, to ensure that communication between CPU and discrete TPM chips
2520 cannot be eavesdropped to acquire disk encryption keys.
2522 * A new switch --fido2-credential-algorithm= has been added to
2523 systemd-cryptenroll allowing selection of the credential algorithm to
2524 use when binding encryption to FIDO2 tokens.
2526 Changes in systemd-hostnamed:
2528 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
2529 to override the values gleaned from the hwdb.
2531 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
2532 /sys/class/dmi/id) to override the chassis that is reported by
2535 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
2536 for reading the hardware serial number, as reportd by DMI. It also
2537 exposes a new method D-Bus property FirmwareVersion that encode the
2538 firmware version of the system.
2540 Changes in other components:
2542 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
2543 handling with the values that were configured during systemd build
2544 (if /etc/locale.conf has not been created through some other
2545 mechanism). This means that /etc/locale.conf should always have
2546 reasonable contents and we avoid a potential mismatch in defaults.
2548 * The userdbctl tool will now show UID range information as part of the
2549 list of known users.
2551 * A new build-time configuration setting default-user-shell= can be
2552 used to set the default shell for user records and nspawn shell
2553 invocations (instead of the default /bin/bash).
2555 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
2556 information dynamically at runtime via IPC.
2558 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
2559 whether a functioning TPM2 infrastructure is available, i.e. if
2560 firmware, kernel driver and systemd all have TPM2 support enabled and
2563 * The systemd-creds tool gained support for generating encrypted
2564 credentials that are using an empty encryption key. While this
2565 provides no integrity nor confidentiality it's useful to implement
2566 codeflows that work the same on TPM-ful and TPM2-less systems. The
2567 service manager will only accept credentials "encrypted" that way if
2568 a TPM2 device cannot be detected, to ensure that credentials
2569 "encrypted" like that cannot be used to trick TPM2 systems.
2571 * When deciding whether to colorize output, all systemd programs now
2572 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
2575 * Meson's new install_tag feature is now in use for several components,
2576 allowing to build and install select binaries only: pam, nss, devel
2577 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
2578 $ meson build systemd-boot
2579 $ meson install --tags systemd-boot --no-rebuild
2580 https://mesonbuild.com/Installing.html#installation-tags
2582 * A new build configuration option has been added, to allow selecting the
2583 default compression algorithm used by systemd-journald and systemd-coredump.
2584 This allows to build-in support for decompressing all supported formats,
2585 but choose a specific one for compression. E.g.:
2586 $ meson -Ddefault-compression=xz
2588 Experimental features:
2590 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
2591 loader.conf that implements booting Microsoft Windows from the
2592 sd-boot in a way that first reboots the system, to reset the TPM
2593 PCRs. This improves compatibility with BitLocker's TPM use, as the
2594 PCRs will only record the Windows boot process, and not sd-boot
2595 itself, thus retaining the PCR measurements not involving sd-boot.
2596 Note that this feature is experimental for now, and is likely going
2597 to be generalized and renamed in a future release, without retaining
2598 compatibility with the current implementation.
2600 * A new systemd-sysupdate component has been added that automatically
2601 discovers, downloads, and installs A/B-style updates for the host
2602 installation itself, or container images, portable service images,
2603 and other assets. See the new systemd-sysupdate man page for updates.
2605 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
2606 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
2607 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
2608 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
2609 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
2610 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
2611 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
2612 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
2613 David, David Bond, Davide Cavalca, David Tardon, davijosw,
2614 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
2615 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
2616 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
2617 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
2618 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
2619 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
2620 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
2621 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
2622 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
2623 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
2624 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
2625 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
2626 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
2627 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
2628 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
2629 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
2630 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
2631 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
2632 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
2633 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
2634 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
2635 Simon Ellmann, Sonali Srivastava, Stefan Seering,
2636 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
2637 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
2638 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
2639 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
2640 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
2641 Zbigniew Jędrzejewski-Szmek, наб
2643 — Edinburgh, 2022-05-21
2647 * Support for encrypted and authenticated credentials has been added.
2648 This extends the credential logic introduced with v247 to support
2649 non-interactive symmetric encryption and authentication, based on a
2650 key that is stored on the /var/ file system or in the TPM2 chip (if
2651 available), or the combination of both (by default if a TPM2 chip
2652 exists the combination is used, otherwise the /var/ key only). The
2653 credentials are automatically decrypted at the moment a service is
2654 started, and are made accessible to the service itself in unencrypted
2655 form. A new tool 'systemd-creds' encrypts credentials for this
2656 purpose, and two new service file settings LoadCredentialEncrypted=
2657 and SetCredentialEncrypted= configure such credentials.
2659 This feature is useful to store sensitive material such as SSL
2660 certificates, passwords and similar securely at rest and only decrypt
2661 them when needed, and in a way that is tied to the local OS
2662 installation or hardware.
2664 * systemd-gpt-auto-generator can now automatically set up discoverable
2665 LUKS2 encrypted swap partitions.
2667 * The GPT Discoverable Partitions Specification has been substantially
2668 extended with support for root and /usr/ partitions for the majority
2669 of architectures systemd supports. This includes platforms that do
2670 not natively support UEFI, because even though GPT is specified under
2671 UEFI umbrella, it is useful on other systems too. Specifically,
2672 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
2673 Portable Services use the concept without requiring UEFI.
2675 * The GPT Discoverable Partitions Specifications has been extended with
2676 a new set of partitions that may carry PKCS#7 signatures for Verity
2677 partitions, encoded in a simple JSON format. This implements a simple
2678 mechanism for building disk images that are fully authenticated and
2679 can be tested against a set of cryptographic certificates. This is
2680 now implemented for the various systemd tools that can operate with
2681 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
2682 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
2683 The PKCS#7 signatures are passed to the kernel (where they are
2684 checked against certificates from the kernel keyring), or can be
2685 verified against certificates provided in userspace (via a simple
2686 drop-in file mechanism).
2688 * systemd-dissect's inspection logic will now report for which uses a
2689 disk image is intended. Specifically, it will display whether an
2690 image is suitable for booting on UEFI or in a container (using
2691 systemd-nspawn's --image= switch), whether it can be used as portable
2692 service, or attached as system extension.
2694 * The system-extension.d/ drop-in files now support a new field
2695 SYSEXT_SCOPE= that may encode which purpose a system extension image
2696 is for: one of "initrd", "system" or "portable". This is useful to
2697 make images more self-descriptive, and to ensure system extensions
2698 cannot be attached in the wrong contexts.
2700 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
2701 be used in portable service images to indicate which unit prefixes
2704 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
2705 now is able to decode images for non-native architectures as well.
2706 This allows systemd-nspawn to boot images of non-native architectures
2707 if the corresponding user mode emulator is installed and
2708 systemd-binfmtd is running.
2710 * systemd-logind gained new settings HandlePowerKeyLongPress=,
2711 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
2712 HandleHibernateKeyLongPress= which may be used to configure actions
2713 when the relevant keys are pressed for more than 5s. This is useful
2714 on devices that only have hardware for a subset of these keys. By
2715 default, if the reboot key is pressed long the poweroff operation is
2716 now triggered, and when the suspend key is pressed long the hibernate
2717 operation is triggered. Long pressing the other two keys currently
2718 does not trigger any operation by default.
2720 * When showing unit status updates on the console during boot and
2721 shutdown, and a service is slow to start so that the cylon animation
2722 is shown, the most recent sd_notify() STATUS= text is now shown as
2723 well. Services may use this to make the boot/shutdown output easier
2724 to understand, and to indicate what precisely a service that is slow
2725 to start or stop is waiting for. In particular, the per-user service
2726 manager instance now reports what it is doing and which service it is
2727 waiting for this way to the system service manager.
2729 * The service manager will now re-execute on reception of the
2730 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
2731 only when running as PID 1. There was no signal to request this when
2732 running as per-user service manager, i.e. as any other PID than 1.
2733 SIGRTMIN+25 works for both system and user managers.
2735 * The hardware watchdog logic in PID 1 gained support for operating
2736 with the default timeout configured in the hardware, instead of
2737 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
2738 request this behavior.
2740 * A new kernel command line option systemd.watchdog_sec= is now
2741 understood which may be used to override the hardware watchdog
2742 time-out for the boot.
2744 * A new setting DefaultOOMScoreAdjust= is now supported in
2745 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
2746 to set the default process OOM score adjustment value for processes
2747 started by the service manager. For per-user service managers this
2748 now defaults to 100, but for per-system service managers is left as
2749 is. This means that by default now services forked off the user
2750 service manager are more likely to be killed by the OOM killer than
2751 system services or the managers themselves.
2753 * A new per-service setting RestrictFileSystems= as been added that
2754 restricts the file systems a service has access to by their type.
2755 This is based on the new BPF LSM of the Linux kernel. It provides an
2756 effective way to make certain API file systems unavailable to
2757 services (and thus minimizing attack surface). A new command
2758 "systemd-analyze filesystems" has been added that lists all known
2759 file system types (and how they are grouped together under useful
2762 * Services now support a new setting RestrictNetworkInterfaces= for
2763 restricting access to specific network interfaces.
2765 * Service unit files gained new settings StartupAllowedCPUs= and
2766 StartupAllowedMemoryNodes=. These are similar to their counterparts
2767 without the "Startup" prefix and apply during the boot process
2768 only. This is useful to improve boot-time behavior of the system and
2769 assign resources differently during boot than during regular
2770 runtime. This is similar to the preexisting StartupCPUWeight=
2773 * Related to this: the various StartupXYZ= settings
2774 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
2775 during shutdown. The settings not prefixed with "Startup" hence apply
2776 during regular runtime, and those that are prefixed like that apply
2777 during boot and shutdown.
2779 * A new per-unit set of conditions/asserts
2780 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
2781 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
2782 pressure is above the configured threshold, using the kernel PSI
2783 feature. For more details see systemd.unit(5) and
2784 https://docs.kernel.org/accounting/psi.html
2786 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
2787 ProtectKernelLogs=yes can now be used.
2789 * The default maximum numbers of inodes have been raised from 64k to 1M
2790 for /dev/, and from 400k to 1M for /tmp/.
2792 * The per-user service manager learnt support for communicating with
2793 systemd-oomd to acquire OOM kill information.
2795 * A new service setting ExecSearchPath= has been added that allows
2796 changing the search path for executables for services. It affects
2797 where we look for the binaries specified in ExecStart= and similar,
2798 and the specified directories are also added the $PATH environment
2799 variable passed to invoked processes.
2801 * A new setting RuntimeRandomizedExtraSec= has been added for service
2802 and scope units that allows extending the runtime time-out as
2803 configured by RuntimeMaxSec= with a randomized amount.
2805 * The syntax of the service unit settings RuntimeDirectory=,
2806 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
2807 if the specified value is now suffixed with a colon, followed by
2808 another filename, the latter will be created as symbolic link to the
2809 specified directory. This allows creating these service directories
2810 together with alias symlinks to make them available under multiple
2813 * Service unit files gained two new settings TTYRows=/TTYColumns= for
2814 configuring rows/columns of the TTY device passed to
2815 stdin/stdout/stderr of the service. This is useful to propagate TTY
2816 dimensions to a virtual machine.
2818 * A new service unit file setting ExitType= has been added that
2819 specifies when to assume a service has exited. By default systemd
2820 only watches the main process of a service. By setting
2821 ExitType=cgroup it can be told to wait for the last process in a
2824 * Automount unit files gained a new setting ExtraOptions= that can be
2825 used to configure additional mount options to pass to the kernel when
2826 mounting the autofs instance.
2828 * "Urlification" (generation of ESC sequences that generate clickable
2829 hyperlinks in modern terminals) may now be turned off altogether
2832 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
2833 settings that default to 200 and 2 s respectively. The ratelimit
2834 ensures that a path unit cannot cause PID1 to busy-loop when it is
2835 trying to trigger a service that is skipped because of a Condition*=
2836 not being satisfied. This matches the configuration and behaviour of
2839 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
2840 as a plug-in for cryptsetup. This means the plain cryptsetup command
2841 may now be used to unlock volumes set up this way.
2843 * The TPM2 logic in cryptsetup will now automatically detect systems
2844 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
2845 updates the SHA1 banks. In such a case PCR policies will be
2846 automatically bound to the latter, not the former. This makes the PCR
2847 policies reliable, but of course do not provide the same level of
2848 trust as SHA256 banks.
2850 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
2851 RSA primary keys in addition to ECC, improving compatibility with
2852 TPM2 chips that do not support ECC. RSA keys are much slower to use
2853 than ECC, and hence are only used if ECC is not available.
2855 * /etc/crypttab gained support for a new token-timeout= setting for
2856 encrypted volumes that allows configuration of the maximum time to
2857 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
2858 the logic will query the user for a regular passphrase/recovery key
2861 * Support for activating dm-integrity volumes at boot via a new file
2862 /etc/integritytab and the tool systemd-integritysetup have been
2863 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
2864 with dm-integrity instead of dm-crypt/dm-verity.
2866 * The systemd-veritysetup-generator now understands a new usrhash=
2867 kernel command line option for specifying the Verity root hash for
2868 the partition backing the /usr/ file system. A matching set of
2869 systemd.verity_usr_* kernel command line options has been added as
2870 well. These all work similar to the corresponding options for the
2873 * The sd-device API gained a new API call sd_device_get_diskseq() to
2874 return the DISKSEQ property of a device structure. The "disk
2875 sequence" concept is a new feature recently introduced to the Linux
2876 kernel that allows detecting reuse cycles of block devices, i.e. can
2877 be used to recognize when loopback block devices are reused for a
2878 different purpose or CD-ROM drives get their media changed.
2880 * A new unit systemd-boot-update.service has been added. If enabled
2881 (the default) and the sd-boot loader is detected to be installed, it
2882 is automatically updated to the newest version when out of date. This
2883 is useful to ensure the boot loader remains up-to-date, and updates
2884 automatically propagate from the OS tree in /usr/.
2886 * sd-boot will now build with SBAT by default in order to facilitate
2887 working with recent versions of Shim that require it to be present.
2889 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
2890 This is used to robustly generate boot entry titles for Windows.
2892 * A new generic target unit factory-reset.target has been added. It is
2893 hooked into systemd-logind similar in fashion to
2894 reboot/poweroff/suspend/hibernate, and is supposed to be used to
2895 initiate a factory reset operation. What precisely this operation
2896 entails is up for the implementer to decide, the primary goal of the
2897 new unit is provide a framework where to plug in the implementation
2898 and how to trigger it.
2900 * A new meson build-time option 'clock-valid-range-usec-max' has been
2901 added which takes a time in µs and defaults to 15 years. If the RTC
2902 time is noticed to be more than the specified time ahead of the
2903 built-in epoch of systemd (which by default is the release timestamp
2904 of systemd) it is assumed that the RTC is not working correctly, and
2905 the RTC is reset to the epoch. (It already is reset to the epoch when
2906 noticed to be before it.) This should increase the chance that time
2907 doesn't accidentally jump too far ahead due to faulty hardware or
2910 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
2911 which may be used to automatically save the current system time to
2912 disk in regular intervals. This is useful to maintain a roughly
2913 monotonic clock even without RTC hardware and with some robustness
2914 against abnormal system shutdown.
2916 * systemd-analyze verify gained support for a pair of new --image= +
2917 --root= switches for verifying units below a specific root
2918 directory/image instead of on the host.
2920 * systemd-analyze verify gained support for verifying unit files under
2921 an explicitly specified unit name, independently of what the filename
2924 * systemd-analyze verify gained a new switch --recursive-errors= which
2925 controls whether to only fail on errors found in the specified units
2926 or recursively any dependent units.
2928 * systemd-analyze security now supports a new --offline mode for
2929 analyzing unit files stored on disk instead of loaded units. It may
2930 be combined with --root=/--image to analyze unit files under a root
2931 directory or disk image. It also learnt a new --threshold= parameter
2932 for specifying an exposure level threshold: if the exposure level
2933 exceeds the specified value the call will fail. It also gained a new
2934 --security-policy= switch for configuring security policies to
2935 enforce on the units. A policy is a JSON file that lists which tests
2936 shall be weighted how much to determine the overall exposure
2937 level. Altogether these new features are useful for fully automatic
2938 analysis and enforcement of security policies on unit files.
2940 * systemd-analyze security gain a new --json= switch for JSON output.
2942 * systemd-analyze learnt a new --quiet switch for reducing
2943 non-essential output. It's honored by the "dot", "syscall-filter",
2944 "filesystems" commands.
2946 * systemd-analyze security gained a --profile= option that can be used
2947 to take into account a portable profile when analyzing portable
2948 services, since a lot of the security-related settings are enabled
2951 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
2952 files, binaries and executables and prints metadata information,
2953 including the build-id and other info described on:
2954 https://systemd.io/COREDUMP_PACKAGE_METADATA/
2956 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
2957 section, for automatically propagating DNS settings from other
2960 * The static lease DHCP server logic in systemd-networkd may now serve
2961 IP addresses outside of the configured IP pool range for the server.
2963 * CAN support in systemd-networkd gained four new settings Loopback=,
2964 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
2965 control modes. It gained a number of further settings for tweaking
2968 * The [CAN] section in .network file gained new TimeQuantaNSec=,
2969 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
2970 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
2971 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
2972 DataSyncJumpWidth= settings to control bit-timing processed by the
2975 * DHCPv4 client support in systemd-networkd learnt a new Label= option
2976 for configuring the address label to apply to configure IPv4
2979 * The [IPv6AcceptRA] section of .network files gained support for a new
2980 UseMTU= setting that may be used to control whether to apply the
2981 announced MTU settings to the local interface.
2983 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
2984 setting to control whether the DHCPv4 client request and process the
2987 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
2988 [DHCPPrefixDelegation], as now the prefix delegation is also supported
2989 with DHCPv4 protocol by enabling the Use6RD= setting.
2991 * The [DHCPPrefixDelegation] section in .network file gained a new
2992 setting UplinkInterface= to specify the upstream interface.
2994 * The [DHCPv6] section in .network file gained a new setting
2995 UseDelegatedPrefix= to control whether the delegated prefixes will be
2996 propagated to the downstream interfaces.
2998 * The [IPv6AcceptRA] section of .network files now understands two new
2999 settings UseGateway=/UseRoutePrefix= for explicitly configuring
3000 whether to use the relevant fields from the IPv6 Router Advertisement
3003 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
3004 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
3005 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
3006 [IPv6AcceptRA] section to control when the DHCPv6 client is started
3007 and how the delegated prefixes are handled by the DHCPv6 client.
3009 * The IPv6Token= section in the [Network] section is deprecated, and
3010 the [IPv6AcceptRA] section gained the Token= setting for its
3011 replacement. The [IPv6Prefix] section also gained the Token= setting.
3012 The Token= setting gained 'eui64' mode to explicitly configure an
3013 address with the EUI64 algorithm based on the interface MAC address.
3014 The 'prefixstable' mode can now optionally take a secret key. The
3015 Token= setting in the [DHCPPrefixDelegation] section now supports all
3016 algorithms supported by the same settings in the other sections.
3018 * The [RoutingPolicyRule] section of .network file gained a new
3019 SuppressInterfaceGroup= setting.
3021 * The IgnoreCarrierLoss= setting in the [Network] section of .network
3022 files now allows a duration to be specified, controlling how long to
3023 wait before reacting to carrier loss.
3025 * The [DHCPServer] section of .network file gained a new Router=
3026 setting to specify the router address.
3028 * The [CAKE] section of .network files gained various new settings
3029 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
3030 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
3031 and UseRawPacketSize= for configuring CAKE.
3033 * systemd-networkd now ships with new default .network files:
3034 80-container-vb.network which matches host-side network bridge device
3035 created by systemd-nspawn's --network-bridge or --network-zone
3036 switch, and 80-6rd-tunnel.network which matches automatically created
3037 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
3039 * systemd-networkd's handling of Endpoint= resolution for WireGuard
3040 interfaces has been improved.
3042 * systemd-networkd will now automatically configure routes to addresses
3043 specified in AllowedIPs=. This feature can be controlled via
3044 RouteTable= and RouteMetric= settings in [WireGuard] or
3045 [WireGuardPeer] sections.
3047 * systemd-networkd will now once again automatically generate persistent
3048 MAC addresses for batadv and bridge interfaces. Users can disable this
3049 by using MACAddress=none in .netdev files.
3051 * systemd-networkd and systemd-udevd now support IP over InfiniBand
3052 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
3053 systemd.netdev files gained the [IPoIB] section.
3055 * systemd-networkd and systemd-udevd now support net.ifname-policy=
3056 option on the kernel command-line. This is implemented through the
3057 systemd-network-generator service that automatically generates
3058 appropriate .link, .network, and .netdev files.
3060 * The various systemd-udevd "ethtool" buffer settings now understand
3061 the special value "max" to configure the buffers to the maximum the
3064 * systemd-udevd's .link files may now configure a large variety of
3065 NIC coalescing settings, plus more hardware offload settings.
3067 * .link files gained a new WakeOnLanPassword= setting in the [Link]
3068 section that allows to specify a WoL "SecureOn" password on hardware
3071 * systemd-nspawn's --setenv= switch now supports an additional syntax:
3072 if only a variable name is specified (i.e. without being suffixed by
3073 a '=' character and a value) the current value of the environment
3074 variable is propagated to the container. e.g. --setenv=FOO will
3075 lookup the current value of $FOO in the environment, and pass it down
3076 to the container. Similar behavior has been added to homectl's,
3077 machinectl's and systemd-run's --setenv= switch.
3079 * systemd-nspawn gained a new switch --suppress-sync= which may be used
3080 to optionally suppress the effect of the sync()/fsync()/fdatasync()
3081 system calls for the container payload. This is useful for build
3082 system environments where safety against abnormal system shutdown is
3083 not essential as all build artifacts can be regenerated any time, but
3084 the performance win is beneficial.
3086 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
3087 same value that PID 1 uses for most forked off processes.
3089 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
3090 uidmap/nouidmap options as last parameter. If "uidmap" is used the
3091 bind mounts are created with UID mapping taking place that ensures
3092 the host's file ownerships are mapped 1:1 to container file
3093 ownerships, even if user namespacing is used. This way
3094 files/directories bound into containers will no longer show up as
3095 owned by the nobody user as they typically did if no special care was
3096 taken to shift them manually.
3098 * When discovering Windows installations sd-boot will now attempt to
3099 show the Windows version.
3101 * The color scheme to use in sd-boot may now be configured at
3104 * sd-boot gained the ability to change screen resolution during
3105 boot-time, by hitting the "r" key. This will cycle through available
3106 resolutions and save the last selection.
3108 * sd-boot learnt a new hotkey "f". When pressed the system will enter
3109 firmware setup. This is useful in environments where it is difficult
3110 to hit the right keys early enough to enter the firmware, and works
3111 on any firmware regardless which key it natively uses.
3113 * sd-boot gained support for automatically booting into the menu item
3114 selected on the last boot (using the "@saved" identifier for menu
3117 * sd-boot gained support for automatically loading all EFI drivers
3118 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
3119 Partition (ESP). These drivers are loaded before the menu entries are
3120 loaded. This is useful e.g. to load additional file system drivers
3121 for the XBOOTLDR partition.
3123 * systemd-boot will now paint the input cursor on its own instead of
3124 relying on the firmware to do so, increasing compatibility with broken
3125 firmware that doesn't make the cursor reasonably visible.
3127 * sd-boot now embeds a .osrel PE section like we expect from Boot
3128 Loader Specification Type #2 Unified Kernels. This means sd-boot
3129 itself may be used in place of a Type #2 Unified Kernel. This is
3130 useful for debugging purposes as it allows chain-loading one a
3131 (development) sd-boot instance from another.
3133 * sd-boot now supports a new "devicetree" field in Boot Loader
3134 Specification Type #1 entries: if configured the specified device
3135 tree file is installed before the kernel is invoked. This is useful
3136 for installing/applying new devicetree files without updating the
3139 * Similarly, sd-stub now can read devicetree data from a PE section
3140 ".dtb" and apply it before invoking the kernel.
3142 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
3143 gained the ability to pick up credentials and sysext files, wrap them
3144 in a cpio archive, and pass as an additional initrd to the invoked
3145 Linux kernel, in effect placing those files in the /.extra/ directory
3146 of the initrd environment. This is useful to implement trusted initrd
3147 environments which are fully authenticated but still can be extended
3148 (via sysexts) and parameterized (via encrypted/authenticated
3149 credentials, see above).
3151 Credentials can be located next to the kernel image file (credentials
3152 specific to a single boot entry), or in one of the shared directories
3153 (credentials applicable to multiple boot entries).
3155 * sd-stub now comes with a full man page, that explains its feature set
3156 and how to combine a kernel image, an initrd and the stub to build a
3157 complete EFI unified kernel image, implementing Boot Loader
3158 Specification Type #2.
3160 * sd-stub may now provide the initrd to the executed kernel via the
3161 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
3162 non-x86 architectures.
3164 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
3165 may be used to set the boot menu time-out of the boot loader (for all
3166 or just the subsequent boot).
3168 * bootctl and kernel-install will now read variables
3169 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
3170 /etc/kernel/install.conf. When set, it specifies the layout to use
3171 for installation directories on the boot partition, so that tools
3172 don't need to guess it based on the already-existing directories. The
3173 only value that is defined natively is "bls", corresponding to the
3175 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
3176 kernel-install that implement a different layout can declare other
3177 values for this variable.
3179 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
3180 assumption that if the user installed sd-boot to the ESP, they intend
3181 to use the entry layout understood by sd-boot. It'll also write
3182 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
3183 (and it wasn't specified in the config file yet). Similarly,
3184 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
3185 wasn't specified in the config file yet). Effectively, those changes
3186 mean that the machine-id used for boot loader entry installation is
3187 "frozen" upon first use and becomes independent of the actual
3190 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
3191 images created for distribution ("golden images") are built with no
3192 machine-id, so that a unique machine-id can be created on the first
3193 boot. But those images may contain boot loader entries with the
3194 machine-id used during build included in paths. Using a "frozen"
3195 value allows unambiguously identifying entries that match the
3196 specific installation, while still permitting parallel installations
3199 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
3200 kernel-install to guess the installation layout. This fixes the
3201 problem where a (possibly empty) directory in the boot partition is
3202 created from a different layout causing kernel-install plugins to
3203 assume the wrong layout. A particular example of how this may happen
3204 is the grub2 package in Fedora which includes directories under /boot
3205 directly in its file list. Various other packages pull in grub2 as a
3206 dependency, so it may be installed even if unused, breaking
3207 installations that use the bls layout.
3209 * bootctl and systemd-bless-boot can now be linked statically.
3211 * systemd-sysext now optionally doesn't insist on extension-release.d/
3212 files being placed in the image under the image's file name. If the
3213 file system xattr user.extension-release.strict is set on the
3214 extension release file, it is accepted regardless of its name. This
3215 relaxes security restrictions a bit, as system extension may be
3216 attached under a wrong name this way.
3218 * udevadm's test-builtin command learnt a new --action= switch for
3219 testing the built-in with the specified action (in place of the
3222 * udevadm info gained new switches --property=/--value for showing only
3223 specific udev properties/values instead of all.
3225 * A new hwdb database has been added that contains matches for various
3226 types of signal analyzers (protocol analyzers, logic analyzers,
3227 oscilloscopes, multimeters, bench power supplies, etc.) that should
3228 be accessible to regular users.
3230 * A new hwdb database entry has been added that carries information
3231 about types of cameras (regular or infrared), and in which direction
3232 they point (front or back).
3234 * A new rule to allow console users access to rfkill by default has been
3237 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
3238 now also owned by the system group "sgx".
3240 * A new build-time meson option "extra-net-naming-schemes=" has been
3241 added to define additional naming schemes for udev's network
3242 interface naming logic. This is useful for enterprise distributions
3243 and similar which want to pin the schemes of certain distribution
3244 releases under a specific name and previously had to patch the
3245 sources to introduce new named schemes.
3247 * The predictable naming logic for network interfaces has been extended
3248 to generate stable names from Xen netfront device information.
3250 * hostnamed's chassis property can now be sourced from chassis-type
3251 field encoded in devicetree (in addition to the existing DMI
3254 * systemd-cgls now optionally displays cgroup IDs and extended
3255 attributes for each cgroup. (Controllable via the new --xattr= +
3256 --cgroup-id= switches.)
3258 * coredumpctl gained a new --all switch for operating on all
3259 Journal files instead of just the local ones.
3261 * systemd-coredump will now use libdw/libelf via dlopen() rather than
3262 directly linking, allowing users to easily opt-out of backtrace/metadata
3263 analysis of core files, and reduce image sizes when this is not needed.
3265 * systemd-coredump will now analyze core files with libdw/libelf in a
3266 forked, sandboxed process.
3268 * systemd-homed will now try to unmount an activate home area in
3269 regular intervals once the user logged out fully. Previously this was
3270 attempted exactly once but if the home directory was busy for some
3271 reason it was not tried again.
3273 * systemd-homed's LUKS2 home area backend will now create a BSD file
3274 system lock on the image file while the home area is active
3275 (i.e. mounted). If a home area is found to be locked, logins are
3276 politely refused. This should improve behavior when using home areas
3277 images that are accessible via the network from multiple clients, and
3278 reduce the chance of accidental file system corruption in that case.
3280 * Optionally, systemd-homed will now drop the kernel buffer cache once
3281 a user has fully logged out, configurable via the new --drop-caches=
3284 * systemd-homed now makes use of UID mapped mounts for the home areas.
3285 If the kernel and used file system support it, files are now
3286 internally owned by the "nobody" user (i.e. the user typically used
3287 for indicating "this ownership is not mapped"), and dynamically
3288 mapped to the UID used locally on the system via the UID mapping
3289 mount logic of recent kernels. This makes migrating home areas
3290 between different systems cheaper because recursively chown()ing file
3291 system trees is no longer necessary.
3293 * systemd-homed's CIFS backend now optionally supports CIFS service
3294 names with a directory suffix, in order to place home directories in
3295 a subdirectory of a CIFS share, instead of the top-level directory.
3297 * systemd-homed's CIFS backend gained support for specifying additional
3298 mount options in the JSON user record (cifsExtraMountOptions field,
3299 and --cifs-extra-mount-options= homectl switch). This is for example
3300 useful for configuring mount options such as "noserverino" that some
3301 SMB3 services require (use that to run a homed home directory from a
3302 FritzBox SMB3 share this way).
3304 * systemd-homed will now default to btrfs' zstd compression for home
3305 areas. This is inspired by Fedora's recent decision to switch to zstd
3308 * Additional mount options to use when mounting the file system of
3309 LUKS2 volumes in systemd-homed has been added. Via the
3310 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
3311 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
3312 systemd-homed or via the luksExtraMountOptions user record JSON
3313 property. (Exposed via homectl --luks-extra-mount-options)
3315 * homectl's resize command now takes the special size specifications
3316 "min" and "max" to shrink/grow the home area to the minimum/maximum
3317 size possible, taking disk usage/space constraints and file system
3318 limitations into account. Resizing is now generally graceful: the
3319 logic will try to get as close to the specified size as possible, but
3320 not consider it a failure if the request couldn't be fulfilled
3323 * systemd-homed gained the ability to automatically shrink home areas
3324 on logout to their minimal size and grow them again on next
3325 login. This ensures that while inactive, a home area only takes up
3326 the minimal space necessary, but once activated, it provides
3327 sufficient space for the user's needs. This behavior is only
3328 supported if btrfs is used as file system inside the home area
3329 (because only for btrfs online growing/shrinking is implemented in
3330 the kernel). This behavior is now enabled by default, but may be
3331 controlled via the new --auto-resize-mode= setting of homectl.
3333 * systemd-homed gained support for automatically re-balancing free disk
3334 space among active home areas, in case the LUKS2 backends are used,
3335 and no explicit disk size was requested. This way disk space is
3336 automatically managed and home areas resized in regular intervals and
3337 manual resizing when disk space becomes scarce should not be
3338 necessary anymore. This behavior is only supported if btrfs is used
3339 within the home areas (as only then online shrinking and growing is
3340 supported), and may be configured via the new rebalanceWeight JSON
3341 user record field (as exposed via the new --rebalance-weight= homectl
3342 setting). Re-balancing is mostly automatic, but can also be requested
3343 explicitly via "homectl rebalance", which is synchronous, and thus
3344 may be used to wait until the rebalance run is complete.
3346 * userdbctl gained a --json= switch for configured the JSON formatting
3347 to use when outputting user or group records.
3349 * userdbctl gained a new --multiplexer= switch for explicitly
3350 configuring whether to use the systemd-userdbd server side user
3351 record resolution logic.
3353 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
3354 for chaining up another command to execute after completing the
3355 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
3356 configuration of a single command to invoke, this maybe used to
3357 invoke multiple: first userdbctl's own implementation, and then any
3358 other also configured in the command line.
3360 * The sd-event API gained a new function sd_event_add_inotify_fd() that
3361 is similar to sd_event_add_inotify() but accepts a file descriptor
3362 instead of a path in the file system for referencing the inode to
3365 * The sd-event API gained a new function
3366 sd_event_source_set_ratelimit_expire_callback() that may be used to
3367 define a callback function that is called whenever an event source
3368 leaves the rate limiting phase.
3370 * New documentation has been added explaining which steps are necessary
3371 to port systemd to a new architecture:
3373 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
3375 * The x-systemd.makefs option in /etc/fstab now explicitly supports
3376 ext2, ext3, and f2fs file systems.
3378 * Mount units and units generated from /etc/fstab entries with 'noauto'
3379 are now ordered the same as other units. Effectively, they will be
3380 started earlier (if something actually pulled them in) and stopped
3381 later, similarly to normal mount units that are part of
3382 fs-local.target. This change should be invisible to users, but
3383 should prevent those units from being stopped too early during
3386 * The systemd-getty-generator now honors a new kernel command line
3387 argument systemd.getty_auto= and a new environment variable
3388 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
3389 example useful to turn off gettys inside of containers or similar
3392 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
3393 (in addition to 127.0.0.53, as before). If DNS requests are sent to
3394 this address they are propagated in "bypass" mode only, i.e. are
3395 almost not processed locally, but mostly forwarded as-is to the
3396 current upstream DNS servers. This provides a stable DNS server
3397 address that proxies all requests dynamically to the right upstream
3398 DNS servers even if these dynamically change. This stub does not do
3399 mDNS/LLMNR resolution. However, it will translate look-ups to
3400 DNS-over-TLS if necessary. This new stub is particularly useful in
3401 container/VM environments, or for tethering setups: use DNAT to
3402 redirect traffic to any IP address to this stub.
3404 * systemd-importd now honors new environment variables
3405 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
3406 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
3407 generation, btrfs quota setup and disk synchronization.
3409 * systemd-importd and systemd-resolved can now be optionally built with
3410 OpenSSL instead of libgcrypt.
3412 * systemd-repart no longer requires OpenSSL.
3414 * systemd-sysusers will no longer create the redundant 'nobody' group
3415 by default, as the 'nobody' user is already created with an
3416 appropriate primary group.
3418 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
3420 * systemctl show-environment gained support for --output=json.
3422 * pam_systemd will now first try to use the X11 abstract socket, and
3423 fallback to the socket file in /tmp/.X11-unix/ only if that does not
3426 * systemd-journald will no longer go back to volatile storage
3427 regardless of configuration when its unit is restarted.
3429 * Initial support for the LoongArch architecture has been added (system
3430 call lists, GPT partition table UUIDs, etc).
3432 * systemd-journald's own logging messages are now also logged to the
3433 journal itself when systemd-journald logs to /dev/kmsg.
3435 * systemd-journald now re-enables COW for archived journal files on
3436 filesystems that support COW. One benefit of this change is that
3437 archived journal files will now get compressed on btrfs filesystems
3438 that have compression enabled.
3440 * systemd-journald now deduplicates fields in a single log message
3441 before adding it to the journal. In archived journal files, it will
3442 also punch holes for unused parts and truncate the file as
3443 appropriate, leading to reductions in disk usage.
3445 * journalctl --verify was extended with more informative error
3448 * More of sd-journal's functions are now resistant against journal file
3451 * The shutdown command learnt a new option --show, to display the
3454 * A LICENSES/ directory is now included in the git tree. It contains a
3455 README.md file that explains the licenses used by source files in
3456 this repository. It also contains the text of all applicable
3457 licenses as they appear on spdx.org.
3459 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
3460 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
3461 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
3462 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
3463 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
3464 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
3465 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
3466 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
3467 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
3468 Christian Brauner, Christian Göttsche, Christian Wehrli,
3469 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
3470 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
3471 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
3472 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
3473 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
3474 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
3475 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
3476 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
3477 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
3478 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
3479 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
3480 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
3481 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
3482 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
3483 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
3484 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
3485 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
3486 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
3487 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
3488 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
3489 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
3490 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
3491 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
3492 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
3493 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
3494 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
3495 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
3496 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
3497 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
3498 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
3499 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
3500 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
3501 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
3502 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
3503 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
3504 Дамјан Георгиевски, наб
3506 — Warsaw, 2021-12-23
3510 * When operating on disk images via the --image= switch of various
3511 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
3512 no 'root=' parameter on the kernel command line, and multiple
3513 suitable root or /usr/ partitions exist in the image, then a simple
3514 comparison inspired by strverscmp() is done on the GPT partition
3515 label, and the newest partition is picked. This permits a simple and
3516 generic whole-file-system A/B update logic where new operating system
3517 versions are dropped into partitions whose label is then updated with
3518 a matching version identifier.
3520 * systemd-sysusers now supports querying the passwords to set for the
3521 users it creates via the "credentials" logic introduced in v247: the
3522 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
3523 credentials are consulted for the password to use (either in UNIX
3524 hashed form, or literally). By default these credentials are inherited
3525 down from PID1 (which in turn imports it from a container manager if
3526 there is one). This permits easy configuration of user passwords
3527 during first boot. Example:
3529 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
3531 Note that systemd-sysusers operates in purely additive mode: it
3532 executes no operation if the declared users already exist, and hence
3533 doesn't set any passwords as effect of the command line above if the
3534 specified root user exists already in the image. (Note that
3535 --volatile=yes ensures it doesn't, though.)
3537 * systemd-firstboot now also supports querying various system
3538 parameters via the credential subsystems. Thus, as above this may be
3539 used to initialize important system parameters on first boot of
3540 previously unprovisioned images (i.e. images with a mostly empty
3543 * PID 1 may now show both the unit name and the unit description
3544 strings in its status output during boot. This may be configured with
3545 StatusUnitFormat=combined in system.conf or
3546 systemd.status-unit-format=combined on the kernel command line.
3548 * The systemd-machine-id-setup tool now supports a --image= switch for
3549 provisioning a machine ID file into an OS disk image, similar to how
3550 --root= operates on an OS file tree. This matches the existing switch
3551 of the same name for systemd-tmpfiles, systemd-firstboot, and
3552 systemd-sysusers tools.
3554 * Similarly, systemd-repart gained support for the --image= switch too.
3555 In combination with the existing --size= option, this makes the tool
3556 particularly useful for easily growing disk images in a single
3557 invocation, following the declarative rules included in the image
3560 * systemd-repart's partition configuration files gained support for a
3561 new switch MakeDirectories= which may be used to create arbitrary
3562 directories inside file systems that are created, before registering
3563 them in the partition table. This is useful in particular for root
3564 partitions to create mount point directories for other partitions
3565 included in the image. For example, a disk image that contains a
3566 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
3567 create /home/ and /var/ as empty directories in the root file system
3568 on its creation, so that the resulting image can be mounted
3569 immediately, even in read-only mode.
3571 * systemd-repart's CopyBlocks= setting gained support for the special
3572 value "auto". If used, a suitable matching partition on the booted OS
3573 is found as source to copy blocks from. This is useful when
3574 implementing replicating installers, that are booted from one medium
3575 and then stream their own root partition onto the target medium.
3577 * systemd-repart's partition configuration files gained support for a
3578 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
3579 GPT partition flags for the created partitions: this is useful for
3580 marking newly created partitions as read-only, or as not being
3581 subject for automatic mounting from creation on.
3583 * The /etc/os-release file has been extended with two new (optional)
3584 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
3585 information for OS images that are updated comprehensively and
3586 atomically as one image. Two new specifiers %M, %A now resolve to
3587 these two fields in the various configuration options that resolve
3590 * portablectl gained a new switch --extension= for enabling portable
3591 service images with extensions that follow the extension image
3592 concept introduced with v248, and thus allows layering multiple
3593 images when setting up the root filesystem of the service.
3595 * systemd-coredump will now extract ELF build-id information from
3596 processes dumping core and include it in the coredump report.
3597 Moreover, it will look for ELF .note.package sections with
3598 distribution packaging meta-information about the crashing process.
3599 This is useful to directly embed the rpm or deb (or any other)
3600 package name and version in ELF files, making it easy to match
3601 coredump reports with the specific package for which the software was
3602 compiled. This is particularly useful on environments with ELF files
3603 from multiple vendors, different distributions and versions, as is
3604 common today in our containerized and sand-boxed world. For further
3607 https://systemd.io/COREDUMP_PACKAGE_METADATA
3609 * A new udev hardware database has been added for FireWire devices
3612 * The "net_id" built-in of udev has been updated with three
3613 backwards-incompatible changes:
3615 - PCI hotplug slot names on s390 systems are now parsed as
3616 hexadecimal numbers. They were incorrectly parsed as decimal
3617 previously, or ignored if the name was not a valid decimal
3620 - PCI onboard indices up to 65535 are allowed. Previously, numbers
3621 above 16383 were rejected. This primarily impacts s390 systems,
3622 where values up to 65535 are used.
3624 - Invalid characters in interface names are replaced with "_".
3626 The new version of the net naming scheme is "v249". The previous
3627 scheme can be selected via the "net.naming-scheme=v247" kernel
3628 command line parameter.
3630 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
3631 NULL bus object, for which they will return false. Or in other words,
3632 an unallocated bus connection is neither ready nor open.
3634 * The sd-device API acquired a new API function
3635 sd_device_get_usec_initialized() that returns the monotonic time when
3636 the udev device first appeared in the database.
3638 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
3639 sd_device_get_trigger_uuid(). The former is similar to
3640 sd_device_trigger() but returns a randomly generated UUID that is
3641 associated with the synthetic uevent generated by the call. This UUID
3642 may be read from the sd_device object a monitor eventually receives,
3643 via the sd_device_get_trigger_uuid(). This interface requires kernel
3644 4.13 or above to work, and allows tracking a synthetic uevent through
3645 the entire device management stack. The "udevadm trigger --settle"
3646 logic has been updated to make use of this concept if available to
3647 wait precisely for the uevents it generates. "udevadm trigger" also
3648 gained a new parameter --uuid that prints the UUID for each generated
3651 * sd-device also gained new APIs sd_device_new_from_ifname() and
3652 sd_device_new_from_ifindex() for allocating an sd-device object for
3653 the specified network interface. The former accepts an interface name
3654 (either a primary or an alternative name), the latter an interface
3657 * The native Journal protocol has been documented. Clients may talk
3658 this as alternative to the classic BSD syslog protocol for locally
3659 delivering log records to the Journal. The protocol has been stable
3660 for a long time and in fact been implemented already in a variety
3661 of alternative client libraries. This documentation makes the support
3664 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
3666 * A new BPFProgram= setting has been added to service files. It may be
3667 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
3668 file, or a bind mount or symlink to one. This may be used to upload
3669 and manage BPF programs externally and then hook arbitrary systemd
3672 * The "home.arpa" domain that has been officially declared as the
3673 choice for domain for local home networks per RFC 8375 has been added
3674 to the default NTA list of resolved, since DNSSEC is generally not
3675 available on private domains.
3677 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
3679 * A new ManageForeignRoutingPolicyRules= setting has been added to
3680 .network files which may be used to exclude foreign-created routing
3681 policy rules from systemd-networkd management.
3683 * systemd-network-wait-online gained two new switches -4 and -6 that
3684 may be used to tweak whether to wait for only IPv4 or only IPv6
3687 * .network files gained a new RequiredFamilyForOnline= setting to
3688 fine-tune whether to require an IPv4 or IPv6 address in order to
3689 consider an interface "online".
3691 * networkctl will now show an over-all "online" state in the per-link
3694 * In .network files a new OutgoingInterface= setting has been added to
3695 specify the output interface in bridge FDB setups.
3697 * In .network files the Multipath group ID may now be configured for
3698 [NextHop] entries, via the new Group= setting.
3700 * The DHCP server logic configured in .network files gained a new
3701 setting RelayTarget= that turns the server into a DHCP server relay.
3702 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
3703 to further tweak the DHCP relay behaviour.
3705 * The DHCP server logic also gained a new ServerAddress= setting in
3706 .network files that explicitly specifies the server IP address to
3707 use. If not specified, the address is determined automatically, as
3710 * The DHCP server logic in systemd-networkd gained support for static
3711 DHCP leases, configurable via the [DHCPServerStaticLease]
3712 section. This allows explicitly mapping specific MAC addresses to
3713 fixed IP addresses and vice versa.
3715 * The RestrictAddressFamilies= setting in service files now supports a
3716 new special value "none". If specified sockets of all address
3717 families will be made unavailable to services configured that way.
3719 * systemd-fstab-generator and systemd-repart have been updated to
3720 support booting from disks that carry only a /usr/ partition but no
3721 root partition yet, and where systemd-repart can add it in on the
3722 first boot. This is useful for implementing systems that ship with a
3723 single /usr/ file system, and whose root file system shall be set up
3724 and formatted on a LUKS-encrypted volume whose key is generated
3725 locally (and possibly enrolled in the TPM) during the first boot.
3727 * The [Address] section of .network files now accepts a new
3728 RouteMetric= setting that configures the routing metric to use for
3729 the prefix route created as effect of the address configuration.
3730 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
3731 gained matching settings for their prefix routes. (The option of the
3732 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
3733 it conceptually belongs there; the old option is still understood for
3736 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
3739 * A new udev property ID_NET_DHCP_BROADCAST on network interface
3740 devices is now honoured by systemd-networkd, controlling whether to
3741 issue DHCP offers via broadcasting. This is used to ensure that s390
3742 layer 3 network interfaces work out-of-the-box with systemd-networkd.
3744 * nss-myhostname and systemd-resolved will now synthesize address
3745 records for a new special hostname "_outbound". The name will always
3746 resolve to the local IP addresses most likely used for outbound
3747 connections towards the default routes. On multi-homed hosts this is
3748 useful to have a stable handle referring to "the" local IP address
3749 that matters most, to the point where this is defined.
3751 * The Discoverable Partition Specification has been updated with a new
3752 GPT partition flag "grow-file-system" defined for its partition
3753 types. Whenever partitions with this flag set are automatically
3754 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
3755 of systemd-nspawn or other tools; and as opposed to explicit mounting
3756 via /etc/fstab), the file system within the partition is
3757 automatically grown to the full size of the partition. If the file
3758 system size already matches the partition size this flag has no
3759 effect. Previously, this functionality has been available via the
3760 explicit x-systemd.growfs mount option, and this new flag extends
3761 this to automatically discovered mounts. A new GrowFileSystem=
3762 setting has been added to systemd-repart drop-in files that allows
3763 configuring this partition flag. This new flag defaults to on for
3764 partitions automatically created by systemd-repart, except if they
3765 are marked read-only. See the specification for further details:
3767 https://systemd.io/DISCOVERABLE_PARTITIONS
3769 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
3770 section. If enabled (which is the default), and an NTP server address
3771 is acquired through a DHCP lease on this interface an explicit route
3772 to this address is created on this interface to ensure that NTP
3773 traffic to the NTP server acquired on an interface is also routed
3774 through that interface. The pre-existing RoutesToDNS= setting that
3775 implements the same for DNS servers is now enabled by default.
3777 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
3778 been added that may be used to restrict the network interfaces
3779 sockets created by the service may be bound to. This is implemented
3782 * A new ConditionFirmware= setting has been added to unit files to
3783 conditionalize on certain firmware features. At the moment it may
3784 check whether running on a UEFI system, a device.tree system, or if
3785 the system is compatible with some specified device-tree feature.
3787 * A new ConditionOSRelease= setting has been added to unit files to
3788 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
3789 operators may be used to check if some field has some specific value
3790 or do an alphanumerical comparison. Equality comparisons are useful
3791 for fields like ID, but relative comparisons for fields like
3792 VERSION_ID or IMAGE_VERSION.
3794 * hostnamed gained a new Describe() D-Bus method that returns a JSON
3795 serialization of the host data it exposes. This is exposed via
3796 "hostnamectl --json=" to acquire a host identity description in JSON.
3797 It's our intention to add a similar features to most services and
3798 objects systemd manages, in order to simplify integration with
3799 program code that can consume JSON.
3801 * Similarly, networkd gained a Describe() method on its Manager and
3802 Link bus objects. This is exposed via "networkctl --json=".
3804 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
3805 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
3806 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
3807 that is used both to get the value (when no argument is given), and
3808 to set the value (when an argument is specified). The old names
3809 continue to be supported for compatibility.
3811 * systemd-detect-virt and ConditionVirtualization= are now able to
3812 correctly identify Amazon EC2 environments.
3814 * The LogLevelMax= setting of unit files now applies not only to log
3815 messages generated *by* the service, but also to log messages
3816 generated *about* the service by PID 1. To suppress logs concerning a
3817 specific service comprehensively, set this option to a high log
3820 * bootctl gained support for a new --make-machine-id-directory= switch
3821 that allows precise control on whether to create the top-level
3822 per-machine directory in the boot partition that typically contains
3823 Type 1 boot loader entries.
3825 * During build SBAT data to include in the systemd-boot EFI PE binaries
3826 may be specified now.
3828 * /etc/crypttab learnt a new option "headless". If specified any
3829 requests to query the user interactively for passwords or PINs will
3830 be skipped. This is useful on systems that are headless, i.e. where
3831 an interactive user is generally not present.
3833 * /etc/crypttab also learnt a new option "password-echo=" that allows
3834 configuring whether the encryption password prompt shall echo the
3835 typed password and if so, do so literally or via asterisks. (The
3836 default is the same behaviour as before: provide echo feedback via
3839 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
3840 systemd-homed has been updated to allow explicit configuration of the
3841 "user presence" and "user verification" checks, as well as whether a
3842 PIN is required for authentication, via the new switches
3843 --fido2-with-user-presence=, --fido2-with-user-verification=,
3844 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
3845 features are available, and may be enabled or disabled depends on the
3848 * systemd-nspawn's --private-user= switch now accepts the special value
3849 "identity" which configures a user namespacing environment with an
3850 identity mapping of 65535 UIDs. This means the container UID 0 is
3851 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
3852 this doesn't appear to be useful, however it does reduce the attack
3853 surface a bit, since the resulting container will possess process
3854 capabilities only within its namespace and not on the host.
3856 * systemd-nspawn's --private-user-chown switch has been replaced by a
3857 more generic --private-user-ownership= switch that accepts one of
3858 three values: "chown" is equivalent to the old --private-user-chown,
3859 and "off" is equivalent to the absence of the old switch. The value
3860 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
3861 of files and directories of the underlying image to the chosen UID
3862 range for the container. "auto" is equivalent to "map" if UID mapping
3863 mount are supported, otherwise it is equivalent to "chown". The short
3864 -U switch systemd-nspawn now implies --private-user-ownership=auto
3865 instead of the old --private-user-chown. Effectively this means: if
3866 the backing file system supports UID mapping mounts the feature is
3867 now used by default if -U is used. Generally, it's a good idea to use
3868 UID mapping mounts instead of recursive chown()ing, since it allows
3869 running containers off immutable images (since no modifications of
3870 the images need to take place), and share images between multiple
3871 instances. Moreover, the recursive chown()ing operation is slow and
3872 can be avoided. Conceptually it's also a good thing if transient UID
3873 range uses do not leak into persistent file ownership anymore. TLDR:
3874 finally, the last major drawback of user namespacing has been
3875 removed, and -U should always be used (unless you use btrfs, where
3876 UID mapped mounts do not exist; or your container actually needs
3877 privileges on the host).
3879 * nss-systemd now synthesizes user and group shadow records in addition
3880 to the main user and group records. Thus, hashed passwords managed by
3881 systemd-homed are now accessible via the shadow database.
3883 * The userdb logic (and thus nss-systemd, and so on) now read
3884 additional user/group definitions in JSON format from the drop-in
3885 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
3886 /usr/lib/userdb/. This is a simple and powerful mechanism for making
3887 additional users available to the system, with full integration into
3888 NSS including the shadow databases. Since the full JSON user/group
3889 record format is supported this may also be used to define users with
3890 resource management settings and other runtime settings that
3891 pam_systemd and systemd-logind enforce at login.
3893 * The userdbctl tool gained two new switches --with-dropin= and
3894 --with-varlink= which can be used to fine-tune the sources used for
3895 user database lookups.
3897 * systemd-nspawn gained a new switch --bind-user= for binding a host
3898 user account into the container. This does three things: the user's
3899 home directory is bind mounted from the host into the container,
3900 below the /run/userdb/home/ hierarchy. A free UID is picked in the
3901 container, and a user namespacing UID mapping to the host user's UID
3902 installed. And finally, a minimal JSON user and group record (along
3903 with its hashed password) is dropped into /run/host/userdb/. These
3904 records are picked up automatically by the userdb drop-in logic
3905 describe above, and allow the user to login with the same password as
3906 on the host. Effectively this means: if host and container run new
3907 enough systemd versions making a host user available to the container
3908 is trivially simple.
3910 * systemd-journal-gatewayd now supports the switches --user, --system,
3911 --merge, --file= that are equivalent to the same switches of
3912 journalctl, and permit exposing only the specified subset of the
3915 * The OnFailure= dependency between units is now augmented with a
3916 implicit reverse dependency OnFailureOf= (this new dependency cannot
3917 be configured directly it's only created as effect of an OnFailure=
3918 dependency in the reverse order — it's visible in "systemctl show"
3919 however). Similar, Slice= now has an reverse dependency SliceOf=,
3920 that is also not configurable directly, but useful to determine all
3921 units that are members of a slice.
3923 * A pair of new dependency types between units PropagatesStopTo= +
3924 StopPropagatedFrom= has been added, that allows propagation of unit
3925 stop events between two units. It operates similar to the existing
3926 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
3928 * A new dependency type OnSuccess= has been added (plus the reverse
3929 dependency OnSuccessOf=, which cannot be configured directly, but
3930 exists only as effect of the reverse OnSuccess=). It is similar to
3931 OnFailure=, but triggers in the opposite case: when a service exits
3932 cleanly. This allows "chaining up" of services where one or more
3933 services are started once another service has successfully completed.
3935 * A new dependency type Upholds= has been added (plus the reverse
3936 dependency UpheldBy=, which cannot be configured directly, but exists
3937 only as effect of Upholds=). This dependency type is a stronger form
3938 of Wants=: if a unit has an UpHolds= dependency on some other unit
3939 and the former is active then the latter is started whenever it is
3940 found inactive (and no job is queued for it). This is an alternative
3941 to Restart= inside service units, but less configurable, and the
3942 request to uphold a unit is not encoded in the unit itself but in
3943 another unit that intends to uphold it.
3945 * The systemd-ask-password tool now also supports reading passwords
3946 from the credentials subsystem, via the new --credential= switch.
3948 * The systemd-ask-password tool learnt a new switch --emoji= which may
3949 be used to explicit control whether the lock and key emoji (🔐) is
3950 shown in the password prompt on suitable TTYs.
3952 * The --echo switch of systemd-ask-password now optionally takes a
3953 parameter that controls character echo. It may either show asterisks
3954 (default, as before), turn echo off entirely, or echo the typed
3955 characters literally.
3957 * The systemd-ask-password tool also gained a new -n switch for
3958 suppressing output of a trailing newline character when writing the
3959 acquired password to standard output, similar to /bin/echo's -n
3962 * New documentation has been added that describes the organization of
3963 the systemd source code tree:
3965 https://systemd.io/ARCHITECTURE
3967 * Units using ConditionNeedsUpdate= will no longer be activated in
3970 * It is now possible to list a template unit in the WantedBy= or
3971 RequiredBy= settings of the [Install] section of another template
3972 unit, which will be instantiated using the same instance name.
3974 * A new MemoryAvailable property is available for units. If the unit,
3975 or the slices it is part of, have a memory limit set via MemoryMax=/
3976 MemoryHigh=, MemoryAvailable will indicate how much more memory the
3977 unit can claim before hitting the limits.
3979 * systemd-coredump will now try to stay below the cgroup memory limit
3980 placed on itself or one of the slices it runs under, if the storage
3981 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
3982 since files written on such filesystems count toward the cgroup memory
3983 limit. If there is not enough available memory in such cases to store
3984 the core file uncompressed, systemd-coredump will skip to compressed
3985 storage directly (if enabled) and it will avoid analyzing the core file
3986 to print backtrace and metadata in the journal.
3988 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
3989 of a path matches the configured expectations, and remove it if not.
3991 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
3992 specify which of the several available filesystem timestamps (access
3993 time, birth time, change time, modification time) to look at when
3994 deciding whether a path has aged enough to be cleaned.
3996 * A new IPv6StableSecretAddress= setting has been added to .network
3997 files, which takes an IPv6 address to use as secret for IPv6 address
4000 * The [DHCPServer] logic in .network files gained support for a new
4001 UplinkInterface= setting that permits configuration of the uplink
4002 interface name to propagate DHCP lease information from.
4004 * The WakeOnLan= setting in .link files now accepts a list of flags
4005 instead of a single one, to configure multiple wake-on-LAN policies.
4007 * User-space defined tracepoints (USDT) have been added to udev at
4008 strategic locations. This is useful for tracing udev behaviour and
4009 performance with bpftrace and similar tools.
4011 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
4012 setting a network timeout time.
4014 * If a system service is running in a new mount namespace (RootDirectory=
4015 and friends), all file systems will be mounted with MS_NOSUID by
4016 default, unless the system is running with SELinux enabled.
4018 * When enumerating time zones the timedatectl tool will now consult the
4019 'tzdata.zi' file shipped by the IANA time zone database package, in
4020 addition to 'zone1970.tab', as before. This makes sure time zone
4021 aliases are now correctly supported. Some distributions so far did
4022 not install this additional file, most do however. If you
4023 distribution does not install it yet, it might make sense to change
4026 * Intel HID rfkill event is no longer masked, since it's the only
4027 source of rfkill event on newer HP laptops. To have both backward and
4028 forward compatibility, userspace daemon needs to debounce duplicated
4029 events in a short time window.
4031 Contributions from: Aakash Singh, adrian5, Albert Brox,
4032 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
4033 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
4034 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
4035 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
4036 Christian Hesse, Daniel Schaefer, Dan Streetman,
4037 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
4038 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
4039 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
4040 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
4041 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
4042 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
4043 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
4044 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
4045 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
4046 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
4047 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
4048 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
4049 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
4050 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
4051 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
4052 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
4053 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
4054 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
4055 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
4056 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
4057 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
4058 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
4059 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
4060 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
4061 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
4063 — Edinburgh, 2021-07-07
4067 * A concept of system extension images is introduced. Such images may
4068 be used to extend the /usr/ and /opt/ directory hierarchies at
4069 runtime with additional files (even if the file system is read-only).
4070 When a system extension image is activated, its /usr/ and /opt/
4071 hierarchies and os-release information are combined via overlayfs
4072 with the file system hierarchy of the host OS.
4074 A new systemd-sysext tool can be used to merge, unmerge, list, and
4075 refresh system extension hierarchies. See
4076 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
4078 The systemd-sysext.service automatically merges installed system
4079 extensions during boot (before basic.target, but not in very early
4080 boot, since various file systems have to be mounted first).
4082 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
4083 supported system extension level.
4085 * A new ExtensionImages= unit setting can be used to apply the same
4086 system extension image concept from systemd-sysext to the namespaced
4087 file hierarchy of specific services, following the same rules and
4090 * Support for a new special "root=tmpfs" kernel command-line option has
4091 been added. When specified, a tmpfs is mounted on /, and mount.usr=
4092 should be used to point to the operating system implementation.
4094 * A new configuration file /etc/veritytab may be used to configure
4095 dm-verity integrity protection for block devices. Each line is in the
4096 format "volume-name data-device hash-device roothash options",
4097 similar to /etc/crypttab.
4099 * A new kernel command-line option systemd.verity.root_options= may be
4100 used to configure dm-verity behaviour for the root device.
4102 * The key file specified in /etc/crypttab (the third field) may now
4103 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
4104 acquired by connecting to that socket and reading from it. This
4105 allows the implementation of a service to provide key information
4106 dynamically, at the moment when it is needed.
4108 * When the hostname is set explicitly to "localhost", systemd-hostnamed
4109 will respect this. Previously such a setting would be mostly silently
4110 ignored. The goal is to honour configuration as specified by the
4113 * The fallback hostname that will be used by the system manager and
4114 systemd-hostnamed can now be configured in two new ways: by setting
4115 DEFAULT_HOSTNAME= in os-release(5), or by setting
4116 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
4117 also be configured during compilation. The environment variable is
4118 intended for testing and local overrides, the os-release(5) field is
4119 intended to allow customization by different variants of a
4120 distribution that share the same compiled packages.
4122 * The environment block of the manager itself may be configured through
4123 a new ManagerEnvironment= setting in system.conf or user.conf. This
4124 complements existing ways to set the environment block (the kernel
4125 command line for the system manager, the inherited environment and
4126 user@.service unit file settings for the user manager).
4128 * systemd-hostnamed now exports the default hostname and the source of
4129 the configured hostname ("static", "transient", or "default") as
4132 * systemd-hostnamed now exports the "HardwareVendor" and
4133 "HardwareModel" D-Bus properties, which are supposed to contain a
4134 pair of cleaned up, human readable strings describing the system's
4135 vendor and model. It's typically sourced from the firmware's DMI
4136 tables, but may be augmented from a new hwdb database. hostnamectl
4137 shows this in the status output.
4139 * Support has been added to systemd-cryptsetup for extracting the
4140 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
4141 metadata header. This allows the information how to open the
4142 encrypted device to be embedded directly in the device and obviates
4143 the need for configuration in an external file.
4145 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
4146 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
4147 pre-existing support for PKCS#11 security tokens).
4149 * systemd-repart may enroll encrypted partitions using TPM2
4150 hardware. This may be useful for example to create an encrypted /var
4151 partition bound to the machine on first boot.
4153 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
4154 and PKCS#11 security tokens to LUKS volumes, list and destroy
4157 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
4159 It also supports enrolling "recovery keys" and regular passphrases.
4161 * The libfido2 dependency is now based on dlopen(), so that the library
4162 is used at runtime when installed, but is not a hard runtime
4165 * systemd-cryptsetup gained support for two new options in
4166 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
4167 request synchronous processing of encryption/decryption IO.
4169 * The manager may be configured at compile time to use the fexecve()
4170 instead of the execve() system call when spawning processes. Using
4171 fexecve() closes a window between checking the security context of an
4172 executable and spawning it, but unfortunately the kernel displays
4173 stale information in the process' "comm" field, which impacts ps
4176 * The configuration option -Dcompat-gateway-hostname has been dropped.
4177 "_gateway" is now the only supported name.
4179 * The ConditionSecurity=tpm2 unit file setting may be used to check if
4180 the system has at least one TPM2 (tpmrm class) device.
4182 * A new ConditionCPUFeature= has been added that may be used to
4183 conditionalize units based on CPU features. For example,
4184 ConditionCPUFeature=rdrand will condition a unit so that it is only
4185 run when the system CPU supports the RDRAND opcode.
4187 * The existing ConditionControlGroupController= setting has been
4188 extended with two new values "v1" and "v2". "v2" means that the
4189 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
4190 hierarchy or the hybrid hierarchy are used.
4192 * A new PrivateIPC= setting on a unit file allows executed processes to
4193 be moved into a private IPC namespace, with separate System V IPC
4194 identifiers and POSIX message queues.
4196 A new IPCNamespacePath= allows the unit to be joined to an existing
4199 * The tables of system calls in seccomp filters are now automatically
4200 generated from kernel lists exported on
4201 https://fedora.juszkiewicz.com.pl/syscalls.html.
4203 The following architectures should now have complete lists:
4204 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
4205 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
4207 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
4208 on /run/ if it is not already a mount point. A writable /run/ has
4209 always been a requirement for a functioning system, but this was not
4210 guaranteed when using a read-only image.
4212 Users can always specify BindPaths= or InaccessiblePaths= as
4213 overrides, and they will take precedence. If the host's root mount
4214 point is used, there is no change in behaviour.
4216 * New bind mounts and file system image mounts may be injected into the
4217 mount namespace of a service (without restarting it). This is exposed
4218 respectively as 'systemctl bind <unit> <path>…' and
4219 'systemctl mount-image <unit> <image>…'.
4221 * The StandardOutput= and StandardError= settings can now specify files
4222 to be truncated for output (as "truncate:<path>").
4224 * The ExecPaths= and NoExecPaths= settings may be used to specify
4225 noexec for parts of the file system.
4227 * sd-bus has a new function sd_bus_open_user_machine() to open a
4228 connection to the session bus of a specific user in a local container
4229 or on the local host. This is exposed in the existing -M switch to
4230 systemctl and similar tools:
4232 systemctl --user -M lennart@foobar start foo
4234 This will connect to the user bus of a user "lennart" in container
4235 "foobar". If no container name is specified, the specified user on
4236 the host itself is connected to
4238 systemctl --user -M lennart@ start quux
4240 * sd-bus also gained a convenience function sd_bus_message_send() to
4241 simplify invocations of sd_bus_send(), taking only a single
4242 parameter: the message to send.
4244 * sd-event allows rate limits to be set on event sources, for dealing
4245 with high-priority event sources that might starve out others. See
4246 the new man page sd_event_source_set_ratelimit(3) for details.
4248 * systemd.link files gained a [Link] Promiscuous= switch, which allows
4249 the device to be raised in promiscuous mode.
4251 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
4252 number of TX and RX queues to be configured.
4254 New [Link] TransmitQueueLength= setting allows the size of the TX
4255 queue to be configured.
4257 New [Link] GenericSegmentOffloadMaxBytes= and
4258 GenericSegmentOffloadMaxSegments= allow capping the packet size and
4259 the number of segments accepted in Generic Segment Offload.
4261 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
4262 wireless routing protocol that operates on ISO/OSI Layer 2 only and
4263 uses ethernet frames to route/bridge packets. This encompasses a new
4264 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
4265 new settings in .netdev files, and a new BatmanAdvanced= setting in
4268 * systemd.network files gained a [Network] RouteTable= configuration
4269 switch to select the routing policy table.
4271 systemd.network files gained a [RoutingPolicyRule] Type=
4272 configuration switch (one of "blackhole, "unreachable", "prohibit").
4274 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
4275 RouteAllowList= settings to ignore/accept route advertisements from
4276 routers matching specified prefixes. The DenyList= setting has been
4277 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
4280 systemd.network files gained a [DHCPv6] UseAddress= setting to
4281 optionally ignore the address provided in the lease.
4283 systemd.network files gained a [DHCPv6PrefixDelegation]
4284 ManageTemporaryAddress= switch.
4286 systemd.network files gained a new ActivationPolicy= setting which
4287 allows configuring how the UP state of an interface shall be managed,
4288 i.e. whether the interface is always upped, always downed, or may be
4289 upped/downed by the user using "ip link set dev".
4291 * The default for the Broadcast= setting in .network files has slightly
4292 changed: the broadcast address will not be configured for wireguard
4295 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
4296 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
4297 configuration options for VLAN packet handling.
4299 * udev rules may now set log_level= option. This allows debug logs to
4300 be enabled for select events, e.g. just for a specific subsystem or
4301 even a single device.
4303 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
4304 DATA_PREPARED_ID properties for block devices with ISO9660 file
4307 * udev now exports decoded DMI information about installed memory slots
4308 as device properties under the /sys/class/dmi/id/ pseudo device.
4310 * /dev/ is not mounted noexec anymore. This didn't provide any
4311 significant security benefits and would conflict with the executable
4312 mappings used with /dev/sgx device nodes. The previous behaviour can
4313 be restored for individual services with NoExecPaths=/dev (or by allow-
4314 listing and excluding /dev from ExecPaths=).
4316 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
4317 and /dev/vhost-net are owned by the kvm group.
4319 * The hardware database has been extended with a list of fingerprint
4320 readers that correctly support USB auto-suspend using data from
4323 * systemd-resolved can now answer DNSSEC questions through the stub
4324 resolver interface in a way that allows local clients to do DNSSEC
4325 validation themselves. For a question with DO+CD set, it'll proxy the
4326 DNS query and respond with a mostly unmodified packet received from
4327 the upstream server.
4329 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
4330 resolved.conf. If true the service will provide caching even for DNS
4331 lookups made to an upstream DNS server on the 127.0.0.1/::1
4332 addresses. By default (and when the option is false) systemd-resolved
4333 will not cache such lookups, in order to avoid duplicate local
4334 caching, under the assumption the local upstream server caches
4337 * systemd-resolved now implements RFC5001 NSID in its local DNS
4338 stub. This may be used by local clients to determine whether they are
4339 talking to the DNS resolver stub or a different DNS server.
4341 * When resolving host names and other records resolvectl will now
4342 report where the data was acquired from (i.e. the local cache, the
4343 network, locally synthesized, …) and whether the network traffic it
4344 effected was encrypted or not. Moreover the tool acquired a number of
4345 new options --cache=, --synthesize=, --network=, --zone=,
4346 --trust-anchor=, --validate= that take booleans and may be used to
4347 tweak a lookup, i.e. whether it may be answered from cached
4348 information, locally synthesized information, information acquired
4349 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
4350 anchor, and whether DNSSEC validation shall be executed for the
4353 * systemd-nspawn gained a new --ambient-capability= setting
4354 (AmbientCapability= in .nspawn files) to configure ambient
4355 capabilities passed to the container payload.
4357 * systemd-nspawn gained the ability to configure the firewall using the
4358 nftables subsystem (in addition to the existing iptables
4359 support). Similarly, systemd-networkd's IPMasquerade= option now
4360 supports nftables as back-end, too. In both cases NAT on IPv6 is now
4361 supported too, in addition to IPv4 (the iptables back-end still is
4364 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
4365 retains its meaning, but has been deprecated. Please switch to either
4366 "ivp4" or "both" (if covering IPv6 is desired).
4368 * systemd-importd will now download .verity and .roothash.p7s files
4369 along with the machine image (as exposed via machinectl pull-raw).
4371 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
4372 setting to configure the time a unit's cgroup needs to exceed memory
4373 pressure limits before action will be taken, and a new
4374 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
4377 systemd-oomd is now considered fully supported (the usual
4378 backwards-compatibility promises apply). Swap is not required for
4379 operation, but it is still recommended.
4381 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
4382 configures the retry delay when trying to contact servers.
4384 * systemd-stdio-bridge gained --system/--user options to connect to the
4385 system bus (previous default) or the user session bus.
4387 * systemd-localed may now call locale-gen to generate missing locales
4388 on-demand (UTF-8-only). This improves integration with Debian-based
4389 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
4391 * systemctl --check-inhibitors=true may now be used to obey inhibitors
4392 even when invoked non-interactively. The old --ignore-inhibitors
4393 switch is now deprecated and replaced by --check-inhibitors=false.
4395 * systemctl import-environment will now emit a warning when called
4396 without any arguments (i.e. to import the full environment block of
4397 the called program). This command will usually be invoked from a
4398 shell, which means that it'll inherit a bunch of variables which are
4399 specific to that shell, and usually to the TTY the shell is connected
4400 to, and don't have any meaning in the global context of the system or
4401 user service manager. Instead, only specific variables should be
4402 imported into the manager environment block.
4404 Similarly, programs which update the manager environment block by
4405 directly calling the D-Bus API of the manager, should also push
4406 specific variables, and not the full inherited environment.
4408 * systemctl's status output now shows unit state with a more careful
4409 choice of Unicode characters: units in maintenance show a "○" symbol
4410 instead of the usual "●", failed units show "×", and services being
4413 * coredumpctl gained a --debugger-arguments= switch to pass arguments
4414 to the debugger. It also gained support for showing coredump info in
4415 a simple JSON format.
4417 * systemctl/loginctl/machinectl's --signal= option now accept a special
4418 value "list", which may be used to show a brief table with known
4419 process signals and their numbers.
4421 * networkctl now shows the link activation policy in status.
4423 * Various tools gained --pager/--no-pager/--json= switches to
4424 enable/disable the pager and provide JSON output.
4426 * Various tools now accept two new values for the SYSTEMD_COLORS
4427 environment variable: "16" and "256", to configure how many terminal
4428 colors are used in output.
4430 * less 568 or newer is now required for the auto-paging logic of the
4431 various tools. Hyperlink ANSI sequences in terminal output are now
4432 used even if a pager is used, and older versions of less are not able
4433 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
4434 disable this output again.
4436 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
4437 builds, non-merged-usr builds) are now officially deprecated. A
4438 warning is emitted during build. Support is slated to be removed in
4439 about a year (when the Debian Bookworm release development starts).
4441 * Systems with the legacy cgroup v1 hierarchy are now marked as
4442 "tainted", to make it clearer that using the legacy hierarchy is not
4445 * systemd-localed will now refuse to configure a keymap which is not
4446 installed in the file system. This is intended as a bug fix, but
4447 could break cases where systemd-localed was used to configure the
4448 keymap in advanced of it being installed. It is necessary to install
4449 the keymap file first.
4451 * The main git development branch has been renamed to 'main'.
4453 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
4454 for partitions, as in the vast majority of cases they contain none
4455 and are used internally by the bootloader (eg: uboot).
4457 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
4458 spawned processes to the PID of the process itself. This may be used
4459 by programs for detecting whether they were forked off by the service
4460 manager itself or are a process forked off further down the tree.
4462 * The sd-device API gained four new calls: sd_device_get_action() to
4463 determine the uevent add/remove/change/… action the device object has
4464 been seen for, sd_device_get_seqno() to determine the uevent sequence
4465 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
4466 object from stat(2) data of a device node, and sd_device_trigger() to
4467 write to the 'uevent' attribute of a device.
4469 * For most tools the --no-legend= switch has been replaced by
4470 --legend=no and --legend=yes, to force whether tables are shown with
4473 * Units acquired a new property "Markers" that takes a list of zero,
4474 one or two of the following strings: "needs-reload" and
4475 "needs-restart". These markers may be set via "systemctl
4476 set-property". Once a marker is set, "systemctl reload-or-restart
4477 --marked" may be invoked to execute the operation the units are
4478 marked for. This is useful for package managers that want to mark
4479 units for restart/reload while updating, but effect the actual
4480 operations at a later step at once.
4482 * The sd_bus_message_read_strv() API call of sd-bus may now also be
4483 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
4486 * bootctl will now report whether the UEFI firmware used a TPM2 device
4487 and measured the boot process into it.
4489 * systemd-tmpfiles learnt support for a new environment variable
4490 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
4491 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
4492 even if the root fs of the system is not itself a btrfs volume.
4494 * systemd-detect-virt/ConditionVirtualization= will now explicitly
4495 detect Docker/Podman environments where possible. Moreover, they
4496 should be able to generically detect any container manager as long as
4497 it assigns the container a cgroup.
4499 * portablectl gained a new "reattach" verb for detaching/reattaching a
4500 portable service image, useful for updating images on-the-fly.
4502 * Intel SGX enclave device nodes (which expose a security feature of
4503 newer Intel CPUs) will now be owned by a new system group "sgx".
4505 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
4506 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
4507 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
4508 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
4509 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
4510 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
4511 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
4512 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
4513 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
4514 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
4515 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
4516 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
4517 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
4518 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
4519 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
4520 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
4521 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
4522 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
4523 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
4524 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
4525 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
4526 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
4527 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
4528 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
4529 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
4530 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
4531 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
4532 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
4533 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
4534 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
4535 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
4536 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
4537 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
4538 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
4539 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
4540 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
4541 Zmicer Turok, Дамјан Георгиевски
4543 — Berlin, 2021-03-30
4547 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
4548 "bind" and "unbind" to the Linux device model. When this kernel
4549 change was made, systemd-udevd was only minimally updated to handle
4550 and propagate these new event types. The introduction of these new
4551 uevents (which are typically generated for USB devices and devices
4552 needing a firmware upload before being functional) resulted in a
4553 number of issues which we so far didn't address. We hoped the kernel
4554 maintainers would themselves address these issues in some form, but
4555 that did not happen. To handle them properly, many (if not most) udev
4556 rules files shipped in various packages need updating, and so do many
4557 programs that monitor or enumerate devices with libudev or sd-device,
4558 or otherwise process uevents. Please note that this incompatibility
4559 is not fault of systemd or udev, but caused by an incompatible kernel
4560 change that happened back in Linux 4.14, but is becoming more and
4561 more visible as the new uevents are generated by more kernel drivers.
4563 To minimize issues resulting from this kernel change (but not avoid
4564 them entirely) starting with systemd-udevd 247 the udev "tags"
4565 concept (which is a concept for marking and filtering devices during
4566 enumeration and monitoring) has been reworked: udev tags are now
4567 "sticky", meaning that once a tag is assigned to a device it will not
4568 be removed from the device again until the device itself is removed
4569 (i.e. unplugged). This makes sure that any application monitoring
4570 devices that match a specific tag is guaranteed to both see uevents
4571 where the device starts being relevant, and those where it stops
4572 being relevant (the latter now regularly happening due to the new
4573 "unbind" uevent type). The udev tags concept is hence now a concept
4574 tied to a *device* instead of a device *event* — unlike for example
4575 udev properties whose lifecycle (as before) is generally tied to a
4576 device event, meaning that the previously determined properties are
4577 forgotten whenever a new uevent is processed.
4579 With the newly redefined udev tags concept, sometimes it's necessary
4580 to determine which tags are the ones applied by the most recent
4581 uevent/database update, in order to discern them from those
4582 originating from earlier uevents/database updates of the same
4583 device. To accommodate for this a new automatic property CURRENT_TAGS
4584 has been added that works similar to the existing TAGS property but
4585 only lists tags set by the most recent uevent/database
4586 update. Similarly, the libudev/sd-device API has been updated with
4587 new functions to enumerate these 'current' tags, in addition to the
4588 existing APIs that now enumerate the 'sticky' ones.
4590 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
4591 essential that all udev rules files and applications are updated to
4592 handle the new events. Specifically:
4594 • All rule files that currently use a header guard similar to
4595 ACTION!="add|change",GOTO="xyz_end" should be updated to use
4596 ACTION=="remove",GOTO="xyz_end" instead, so that the
4597 properties/tags they add are also applied whenever "bind" (or
4598 "unbind") is seen. (This is most important for all physical device
4599 types — those for which "bind" and "unbind" are currently
4600 generated, for all other device types this change is still
4601 recommended but not as important — but certainly prepares for
4602 future kernel uevent type additions).
4604 • Similarly, all code monitoring devices that contains an 'if' branch
4605 discerning the "add" + "change" uevent actions from all other
4606 uevents actions (i.e. considering devices only relevant after "add"
4607 or "change", and irrelevant on all other events) should be reworked
4608 to instead negatively check for "remove" only (i.e. considering
4609 devices relevant after all event types, except for "remove", which
4610 invalidates the device). Note that this also means that devices
4611 should be considered relevant on "unbind", even though conceptually
4612 this — in some form — invalidates the device. Since the precise
4613 effect of "unbind" is not generically defined, devices should be
4614 considered relevant even after "unbind", however I/O errors
4615 accessing the device should then be handled gracefully.
4617 • Any code that uses device tags for deciding whether a device is
4618 relevant or not most likely needs to be updated to use the new
4619 udev_device_has_current_tag() API (or sd_device_has_current_tag()
4620 in case sd-device is used), to check whether the tag is set at the
4621 moment an uevent is seen (as opposed to the existing
4622 udev_device_has_tag() API which checks if the tag ever existed on
4623 the device, following the API concept redefinition explained
4626 We are very sorry for this breakage and the requirement to update
4627 packages using these interfaces. We'd again like to underline that
4628 this is not caused by systemd/udev changes, but result of a kernel
4631 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
4632 packages have not retriggered devices once the udev package (or any
4633 auxiliary package installing additional udev rules) is updated. We
4634 intend to work with major distributions to change this, so that
4635 "udevadm trigger -a change" is issued on such upgrades, ensuring that
4636 the updated ruleset is applied to the devices already discovered, so
4637 that (asynchronously) after the upgrade completed the udev database
4638 is consistent with the updated rule set. This means udev rules must
4639 be ready to be retriggered with a "change" action any time, and
4640 result in correct and complete udev database entries. While the
4641 majority of udev rule files known to us currently get this right,
4642 some don't. Specifically, there are udev rules files included in
4643 various packages that only set udev properties on the "add" action,
4644 but do not handle the "change" action. If a device matching those
4645 rules is retriggered with the "change" action (as is intended here)
4646 it would suddenly lose the relevant properties. This always has been
4647 problematic, but as soon as all udev devices are triggered on relevant
4648 package upgrades this will become particularly so. It is strongly
4649 recommended to fix offending rules so that they can handle a "change"
4650 action at any time, and acquire all necessary udev properties even
4651 then. Or in other words: the header guard mentioned above
4652 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
4653 this, as it makes sure rules are rerun on "change" correctly, and
4654 accumulate the correct and complete set of udev properties. udev rule
4655 definitions that cannot handle "change" events being triggered at
4656 arbitrary times should be considered buggy.
4658 * The MountAPIVFS= service file setting now defaults to on if
4659 RootImage= and RootDirectory= are used, which means that with those
4660 two settings /proc/, /sys/ and /dev/ are automatically properly set
4661 up for services. Previous behaviour may be restored by explicitly
4662 setting MountAPIVFS=off.
4664 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
4665 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
4666 latter it takes precedence over the former, similar to how most of
4667 systemd's own configuration is handled. Given that PAM stack
4668 definitions are primarily put together by OS vendors/distributions
4669 (though possibly overridden by users), this systemd release moves its
4670 own PAM stack configuration for the "systemd-user" PAM service (i.e.
4671 for the PAM session invoked by the per-user user@.service instance)
4672 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
4673 packages' vendor versions of their PAM stack definitions from
4674 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
4675 desired the location to which systemd installs its PAM stack
4676 configuration may be changed via the -Dpamconfdir Meson option.
4678 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
4679 libpwquality and libcryptsetup have been changed to be based on
4680 dlopen(): instead of regular dynamic library dependencies declared in
4681 the binary ELF headers, these libraries are now loaded on demand
4682 only, if they are available. If the libraries cannot be found the
4683 relevant operations will fail gracefully, or a suitable fallback
4684 logic is chosen. This is supposed to be useful for general purpose
4685 distributions, as it allows minimizing the list of dependencies the
4686 systemd packages pull in, permitting building of more minimal OS
4687 images, while still making use of these "weak" dependencies should
4688 they be installed. Since many package managers automatically
4689 synthesize package dependencies from ELF shared library dependencies,
4690 some additional manual packaging work has to be done now to replace
4691 those (slightly downgraded from "required" to "recommended" or
4692 whatever is conceptually suitable for the package manager). Note that
4693 this change does not alter build-time behaviour: as before the
4694 build-time dependencies have to be installed during build, even if
4695 they now are optional during runtime.
4697 * sd-event.h gained a new call sd_event_add_time_relative() for
4698 installing timers relative to the current time. This is mostly a
4699 convenience wrapper around the pre-existing sd_event_add_time() call
4700 which installs absolute timers.
4702 * sd-event event sources may now be placed in a new "exit-on-failure"
4703 mode, which may be controlled via the new
4704 sd_event_source_get_exit_on_failure() and
4705 sd_event_source_set_exit_on_failure() functions. If enabled, any
4706 failure returned by the event source handler functions will result in
4707 exiting the event loop (unlike the default behaviour of just
4708 disabling the event source but continuing with the event loop). This
4709 feature is useful to set for all event sources that define "primary"
4710 program behaviour (where failure should be fatal) in contrast to
4711 "auxiliary" behaviour (where failure should remain local).
4713 * Most event source types sd-event supports now accept a NULL handler
4714 function, in which case the event loop is exited once the event
4715 source is to be dispatched, using the userdata pointer — converted to
4716 a signed integer — as exit code of the event loop. Previously this
4717 was supported for IO and signal event sources already. Exit event
4718 sources still do not support this (simply because it makes little
4719 sense there, as the event loop is already exiting when they are
4722 * A new per-unit setting RootImageOptions= has been added which allows
4723 tweaking the mount options for any file system mounted as effect of
4724 the RootImage= setting.
4726 * Another new per-unit setting MountImages= has been added, that allows
4727 mounting additional disk images into the file system tree accessible
4730 * Timer units gained a new FixedRandomDelay= boolean setting. If
4731 enabled, the random delay configured with RandomizedDelaySec= is
4732 selected in a way that is stable on a given system (though still
4733 different for different units).
4735 * Socket units gained a new setting Timestamping= that takes "us", "ns"
4736 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
4739 * systemd-repart now generates JSON output when requested with the new
4742 * systemd-machined's OpenMachineShell() bus call will now pass
4743 additional policy metadata data fields to the PolicyKit
4744 authentication request.
4746 * systemd-tmpfiles gained a new -E switch, which is equivalent to
4747 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
4748 --exclude=/sys. It's particularly useful in combination with --root=,
4749 when operating on OS trees that do not have any of these four runtime
4750 directories mounted, as this means no files below these subtrees are
4751 created or modified, since those mount points should probably remain
4754 * systemd-tmpfiles gained a new --image= switch which is like --root=,
4755 but takes a disk image instead of a directory as argument. The
4756 specified disk image is mounted inside a temporary mount namespace
4757 and the tmpfiles.d/ drop-ins stored in the image are executed and
4758 applied to the image. systemd-sysusers similarly gained a new
4759 --image= switch, that allows the sysusers.d/ drop-ins stored in the
4760 image to be applied onto the image.
4762 * Similarly, the journalctl command also gained an --image= switch,
4763 which is a quick one-step solution to look at the log data included
4766 * journalctl's --output=cat option (which outputs the log content
4767 without any metadata, just the pure text messages) will now make use
4768 of terminal colors when run on a suitable terminal, similarly to the
4771 * JSON group records now support a "description" string that may be
4772 used to add a human-readable textual description to such groups. This
4773 is supposed to match the user's GECOS field which traditionally
4774 didn't have a counterpart for group records.
4776 * The "systemd-dissect" tool that may be used to inspect OS disk images
4777 and that was previously installed to /usr/lib/systemd/ has now been
4778 moved to /usr/bin/, reflecting its updated status of an officially
4779 supported tool with a stable interface. It gained support for a new
4780 --mkdir switch which when combined with --mount has the effect of
4781 creating the directory to mount the image to if it is missing
4782 first. It also gained two new commands --copy-from and --copy-to for
4783 copying files and directories in and out of an OS image without the
4784 need to manually mount it. It also acquired support for a new option
4785 --json= to generate JSON output when inspecting an OS image.
4787 * The cgroup2 file system is now mounted with the
4788 "memory_recursiveprot" mount option, supported since kernel 5.7. This
4789 means that the MemoryLow= and MemoryMin= unit file settings now apply
4790 recursively to whole subtrees.
4792 * systemd-homed now defaults to using the btrfs file system — if
4793 available — when creating home directories in LUKS volumes. This may
4794 be changed with the DefaultFileSystemType= setting in homed.conf.
4795 It's now the default file system in various major distributions and
4796 has the major benefit for homed that it can be grown and shrunk while
4797 mounted, unlike the other contenders ext4 and xfs, which can both be
4798 grown online, but not shrunk (in fact xfs is the technically most
4799 limited option here, as it cannot be shrunk at all).
4801 * JSON user records managed by systemd-homed gained support for
4802 "recovery keys". These are basically secondary passphrases that can
4803 unlock user accounts/home directories. They are computer-generated
4804 rather than user-chosen, and typically have greater entropy.
4805 homectl's --recovery-key= option may be used to add a recovery key to
4806 a user account. The generated recovery key is displayed as a QR code,
4807 so that it can be scanned to be kept in a safe place. This feature is
4808 particularly useful in combination with systemd-homed's support for
4809 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
4810 security tokens are lost. Recovery keys may be entered wherever the
4811 system asks for a password.
4813 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
4814 home directory which indicates that a home directory has not been
4815 deactivated cleanly when offline. This flag is useful to identify
4816 home directories for which the offline discard logic did not run when
4817 offlining, and where it would be a good idea to log in again to catch
4820 * systemctl gained a new parameter --timestamp= which may be used to
4821 change the style in which timestamps are output, i.e. whether to show
4822 them in local timezone or UTC, or whether to show µs granularity.
4824 * Alibaba's "pouch" container manager is now detected by
4825 systemd-detect-virt, ConditionVirtualization= and similar
4826 constructs. Similar, they now also recognize IBM PowerVM machine
4829 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
4830 place to use for propagating external mounts into the
4831 container. Similarly /run/host/notify is now used as the socket path
4832 for container payloads to communicate with the container manager
4833 using sd_notify(). The container manager now uses the
4834 /run/host/inaccessible/ directory to place "inaccessible" file nodes
4835 of all relevant types which may be used by the container payload as
4836 bind mount source to over-mount inodes to make them inaccessible.
4837 /run/host/container-manager will now be initialized with the same
4838 string as the $container environment variable passed to the
4839 container's PID 1. /run/host/container-uuid will be initialized with
4840 the same string as $container_uuid. This means the /run/host/
4841 hierarchy is now the primary way to make host resources available to
4842 the container. The Container Interface documents these new files and
4845 https://systemd.io/CONTAINER_INTERFACE
4847 * Support for the "ConditionNull=" unit file condition has been
4848 deprecated and undocumented for 6 years. systemd started to warn
4849 about its use 1.5 years ago. It has now been removed entirely.
4851 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
4852 a sd_bus_error struct and a list of error names, and checks if the
4853 error matches one of these names. It's a convenience wrapper that is
4854 useful in cases where multiple errors shall be handled the same way.
4856 * A new system call filter list "@known" has been added, that contains
4857 all system calls known at the time systemd was built.
4859 * Behaviour of system call filter allow lists has changed slightly:
4860 system calls that are contained in @known will result in EPERM by
4861 default, while those not contained in it result in ENOSYS. This
4862 should improve compatibility because known system calls will thus be
4863 communicated as prohibited, while unknown (and thus newer ones) will
4864 be communicated as not implemented, which hopefully has the greatest
4865 chance of triggering the right fallback code paths in client
4868 * "systemd-analyze syscall-filter" will now show two separate sections
4869 at the bottom of the output: system calls known during systemd build
4870 time but not included in any of the filter groups shown above, and
4871 system calls defined on the local kernel but known during systemd
4874 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
4875 systemd-nspawn all system call filter violations will be logged by
4876 the kernel (audit). This is useful for tracking down system calls
4877 invoked by container payloads that are prohibited by the container's
4878 system call filter policy.
4880 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
4881 systemd-nspawn (and other programs that use seccomp) all seccomp
4882 filtering is turned off.
4884 * Two new unit file settings ProtectProc= and ProcSubset= have been
4885 added that expose the hidepid= and subset= mount options of procfs.
4886 All processes of the unit will only see processes in /proc that are
4887 are owned by the unit's user. This is an important new sandboxing
4888 option that is recommended to be set on all system services. All
4889 long-running system services that are included in systemd itself set
4890 this option now. This option is only supported on kernel 5.8 and
4891 above, since the hidepid= option supported on older kernels was not a
4892 per-mount option but actually applied to the whole PID namespace.
4894 * Socket units gained a new boolean setting FlushPending=. If enabled
4895 all pending socket data/connections are flushed whenever the socket
4896 unit enters the "listening" state, i.e. after the associated service
4899 * The unit file setting NUMAMask= gained a new "all" value: when used,
4900 all existing NUMA nodes are added to the NUMA mask.
4902 * A new "credentials" logic has been added to system services. This is
4903 a simple mechanism to pass privileged data to services in a safe and
4904 secure way. It's supposed to be used to pass per-service secret data
4905 such as passwords or cryptographic keys but also associated less
4906 private information such as user names, certificates, and similar to
4907 system services. Each credential is identified by a short user-chosen
4908 name and may contain arbitrary binary data. Two new unit file
4909 settings have been added: SetCredential= and LoadCredential=. The
4910 former allows setting a credential to a literal string, the latter
4911 sets a credential to the contents of a file (or data read from a
4912 user-chosen AF_UNIX stream socket). Credentials are passed to the
4913 service via a special credentials directory, one file for each
4914 credential. The path to the credentials directory is passed in a new
4915 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
4916 are passed in the file system they may be easily referenced in
4917 ExecStart= command lines too, thus no explicit support for the
4918 credentials logic in daemons is required (though ideally daemons
4919 would look for the bits they need in $CREDENTIALS_DIRECTORY
4920 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
4921 backed by unswappable memory if privileges allow it, immutable if
4922 privileges allow it, is accessible only to the service's UID, and is
4923 automatically destroyed when the service stops.
4925 * systemd-nspawn supports the same credentials logic. It can both
4926 consume credentials passed to it via the aforementioned
4927 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
4928 to its payload. The service manager/PID 1 has been updated to match
4929 this: it can also accept credentials from the container manager that
4930 invokes it (in fact: any process that invokes it), and passes them on
4931 to its services. Thus, credentials can be propagated recursively down
4932 the tree: from a system's service manager to a systemd-nspawn
4933 service, to the service manager that runs as container payload and to
4934 the service it runs below. Credentials may also be added on the
4935 systemd-nspawn command line, using new --set-credential= and
4936 --load-credential= command line switches that match the
4937 aforementioned service settings.
4939 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
4940 the partition drop-ins which may be used to format/LUKS
4941 encrypt/populate any created partitions. The partitions are
4942 encrypted/formatted/populated before they are registered in the
4943 partition table, so that they appear atomically: either the
4944 partitions do not exist yet or they exist fully encrypted, formatted,
4945 and populated — there is no time window where they are
4946 "half-initialized". Thus the system is robust to abrupt shutdown: if
4947 the tool is terminated half-way during its operations on next boot it
4948 will start from the beginning.
4950 * systemd-repart's --size= operation gained a new "auto" value. If
4951 specified, and operating on a loopback file it is automatically sized
4952 to the minimal size the size constraints permit. This is useful to
4953 use "systemd-repart" as an image builder for minimally sized images.
4955 * systemd-resolved now gained a third IPC interface for requesting name
4956 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
4957 interface is now supported. The nss-resolve NSS module has been
4958 modified to use this new interface instead of D-Bus. Using Varlink
4959 has a major benefit over D-Bus: it works without a broker service,
4960 and thus already during earliest boot, before the dbus daemon has
4961 been started. This means name resolution via systemd-resolved now
4962 works at the same time systemd-networkd operates: from earliest boot
4963 on, including in the initrd.
4965 * systemd-resolved gained support for a new DNSStubListenerExtra=
4966 configuration file setting which may be used to specify additional IP
4967 addresses the built-in DNS stub shall listen on, in addition to the
4968 main one on 127.0.0.53:53.
4970 * Name lookups issued via systemd-resolved's D-Bus and Varlink
4971 interfaces (and thus also via glibc NSS if nss-resolve is used) will
4972 now honour a trailing dot in the hostname: if specified the search
4973 path logic is turned off. Thus "resolvectl query foo." is now
4974 equivalent to "resolvectl query --search=off foo.".
4976 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
4977 exposes how /etc/resolv.conf is currently managed: by resolved (and
4978 in which mode if so) or another subsystem. "resolvctl" will display
4979 this property in its status output.
4981 * The resolv.conf snippets systemd-resolved provides will now set "."
4982 as the search domain if no other search domain is known. This turns
4983 off the derivation of an implicit search domain by nss-dns for the
4984 hostname, when the hostname is set to an FQDN. This change is done to
4985 make nss-dns using resolv.conf provided by systemd-resolved behave
4986 more similarly to nss-resolve.
4988 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
4989 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
4990 "birth" time (btime) of a file in addition to the atime, mtime, and
4993 * systemd-analyze gained a new verb "capability" that lists all known
4994 capabilities by the systemd build and by the kernel.
4996 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
4997 advance the system clock to it at boot if it is noticed to be before
4998 that time. Previously, PID 1 would only advance the time to an epoch
4999 time that is set during build-time. With this new file OS builders
5000 can change this epoch timestamp on individual OS images without
5001 having to rebuild systemd.
5003 * systemd-logind will now listen to the KEY_RESTART key from the Linux
5004 input layer and reboot the system if it is pressed, similarly to how
5005 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
5006 was originally defined in the Multimedia context (to restart playback
5007 of a song or film), but is now primarily used in various embedded
5008 devices for "Reboot" buttons. Accordingly, systemd-logind will now
5009 honour it as such. This may configured in more detail via the new
5010 HandleRebootKey= and RebootKeyIgnoreInhibited=.
5012 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
5013 copying OS trees, for example in "systemd-nspawn --ephemeral",
5014 "systemd-nspawn --template=", "machinectl clone" and similar. This is
5015 useful when operating with OSTree images, which use hardlinks heavily
5016 throughout, and where such copies previously resulting in "exploding"
5019 * systemd-nspawn's --console= setting gained support for a new
5020 "autopipe" value, which is identical to "interactive" when invoked on
5021 a TTY, and "pipe" otherwise.
5023 * systemd-networkd's .network files gained support for explicitly
5024 configuring the multicast membership entries of bridge devices in the
5025 [BridgeMDB] section. It also gained support for the PIE queuing
5026 discipline in the [FlowQueuePIE] sections.
5028 * systemd-networkd's .netdev files may now be used to create "BareUDP"
5029 tunnels, configured in the new [BareUDP] setting.
5031 * systemd-networkd's Gateway= setting in .network files now accepts the
5032 special values "_dhcp4" and "_ipv6ra" to configure additional,
5033 locally defined, explicit routes to the gateway acquired via DHCP or
5034 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
5035 but still accepted for backwards compatibility.
5037 * systemd-networkd's [IPv6PrefixDelegation] section and
5038 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
5039 IPv6SendRA= (the old names are still accepted for backwards
5042 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
5043 boolean setting in [Network] section. If enabled, the delegated prefix
5044 gained by another link will be configured, and an address within the
5045 prefix will be assigned.
5047 * systemd-networkd's .network files gained the Announce= boolean setting
5048 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
5049 prefix will be announced through IPv6 router advertisement (IPv6 RA).
5050 The setting is enabled by default.
5052 * VXLAN tunnels may now be marked as independent of any underlying
5053 network interface via the new Independent= boolean setting.
5055 * systemctl gained support for two new verbs: "service-log-level" and
5056 "service-log-target" may be used on services that implement the
5057 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
5058 adjust the log level and target. All of systemd's long-running
5059 services support this now, but ideally all system services would
5060 implement this interface to make the system more uniformly
5063 * The SystemCallErrorNumber= unit file setting now accepts the new
5064 "kill" and "log" actions, in addition to arbitrary error number
5065 specifications as before. If "kill" the processes are killed on the
5066 event, if "log" the offending system call is audit logged.
5068 * A new SystemCallLog= unit file setting has been added that accepts a
5069 list of system calls that shall be logged about (audit).
5071 * The OS image dissection logic (as used by RootImage= in unit files or
5072 systemd-nspawn's --image= switch) has gained support for identifying
5073 and mounting explicit /usr/ partitions, which are now defined in the
5074 discoverable partition specification. This should be useful for
5075 environments where the root file system is
5076 generated/formatted/populated dynamically on first boot and combined
5077 with an immutable /usr/ tree that is supplied by the vendor.
5079 * In the final phase of shutdown, within the systemd-shutdown binary
5080 we'll now try to detach MD devices (i.e software RAID) in addition to
5081 loopback block devices and DM devices as before. This is supposed to
5082 be a safety net only, in order to increase robustness if things go
5083 wrong. Storage subsystems are expected to properly detach their
5084 storage volumes during regular shutdown already (or in case of
5085 storage backing the root file system: in the initrd hook we return to
5088 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
5089 will now log the thread ID in their log output. This is useful when
5090 working with heavily threaded programs.
5092 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
5093 not use the RDRAND CPU instruction. This is useful in environments
5094 such as replay debuggers where non-deterministic behaviour is not
5097 * The autopaging logic in systemd's various tools (such as systemctl)
5098 has been updated to turn on "secure" mode in "less"
5099 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
5100 detected. This disables invoking external programs from the pager,
5101 via the pipe logic. This behaviour may be overridden via the new
5102 $SYSTEMD_PAGERSECURE environment variable.
5104 * Units which have resource limits (.service, .mount, .swap, .slice,
5105 .socket, and .slice) gained new configuration settings
5106 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
5107 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
5108 limits and optional action taken by systemd-oomd.
5110 * A new service systemd-oomd has been added. It monitors resource
5111 contention for selected parts of the unit hierarchy using the PSI
5112 information reported by the kernel, and kills processes when memory
5113 or swap pressure is above configured limits. This service is only
5114 enabled by default in developer mode (see below) and should be
5115 considered a preview in this release. Behaviour details and option
5116 names are subject to change without the usual backwards-compatibility
5119 * A new helper oomctl has been added to introspect systemd-oomd state.
5120 It is only enabled by default in developer mode and should be
5121 considered a preview without the usual backwards-compatibility
5124 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
5125 enabled, systemd reads the system UID boundaries from /etc/login.defs
5126 at runtime, instead of using the built-in values selected during
5127 build. This is an option to improve compatibility for upgrades from
5128 old systems. It's strongly recommended not to make use of this
5129 functionality on new systems (or even enable it during build), as it
5130 makes something runtime-configurable that is mostly an implementation
5131 detail of the OS, and permits avoidable differences in deployments
5132 that create all kinds of problems in the long run.
5134 * New meson option '-Dmode=developer|release' has been added. When
5135 'developer', additional checks and features are enabled that are
5136 relevant during upstream development, e.g. verification that
5137 semi-automatically-generated documentation has been properly updated
5138 following API changes. Those checks are considered hints for
5139 developers and are not actionable in downstream builds. In addition,
5140 extra features that are not ready for general consumption may be
5141 enabled in developer mode. It is thus recommended to set
5142 '-Dmode=release' in end-user and distro builds.
5144 * systemd-cryptsetup gained support for processing detached LUKS
5145 headers specified on the kernel command line via the header=
5146 parameter of the luks.options= kernel command line option. The same
5147 device/path syntax as for key files is supported for header files
5150 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
5151 slot index data for devices that are connected through a PCI bridge
5152 where the _SUN index is associated with the bridge instead of the
5153 network device itself. Previously this would create ambiguous device
5154 naming if multiple network interfaces were connected to the same PCI
5155 bridge. Since this is a naming scheme incompatibility on systems that
5156 possess hardware like this it has been introduced as new naming
5157 scheme "v247". The previous scheme can be selected via the
5158 "net.naming-scheme=v245" kernel command line parameter.
5160 * ConditionFirstBoot= semantics have been modified to be safe towards
5161 abnormal system power-off during first boot. Specifically, the
5162 "systemd-machine-id-commit.service" service now acts as boot
5163 milestone indicating when the first boot process is sufficiently
5164 complete in order to not consider the next following boot also a
5165 first boot. If the system is reset before this unit is reached the
5166 first time, the next boot will still be considered a first boot; once
5167 it has been reached, no further boots will be considered a first
5168 boot. The "first-boot-complete.target" unit now acts as official hook
5169 point to order against this. If a service shall be run on every boot
5170 until the first boot fully succeeds it may thus be ordered before
5171 this target unit (and pull it in) and carry ConditionFirstBoot=
5174 * bootctl's set-default and set-oneshot commands now accept the three
5175 special strings "@default", "@oneshot", "@current" in place of a boot
5176 entry id. These strings are resolved to the current default and
5177 oneshot boot loader entry, as well as the currently booted one. Thus
5178 a command "bootctl set-default @current" may be used to make the
5179 currently boot menu item the new default for all subsequent boots.
5181 * "systemctl edit" has been updated to show the original effective unit
5182 contents in commented form in the text editor.
5184 * Units in user mode are now segregated into three new slices:
5185 session.slice (units that form the core of graphical session),
5186 app.slice ("normal" user applications), and background.slice
5187 (low-priority tasks). Unless otherwise configured, user units are
5188 placed in app.slice. The plan is to add resource limits and
5189 protections for the different slices in the future.
5191 * New GPT partition types for RISCV32/64 for the root and /usr
5192 partitions, and their associated Verity partitions have been defined,
5193 and are now understood by systemd-gpt-auto-generator, and the OS
5194 image dissection logic.
5196 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
5197 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
5198 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
5199 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
5200 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
5201 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
5202 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
5203 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
5204 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
5205 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
5206 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
5207 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
5208 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
5209 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
5210 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
5211 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
5212 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
5213 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
5214 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
5215 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
5216 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
5217 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
5218 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
5219 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
5220 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
5221 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
5222 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
5223 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
5224 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
5225 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
5226 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
5227 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
5228 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
5230 – Warsaw, 2020-11-26
5234 * The service manager gained basic support for cgroup v2 freezer. Units
5235 can now be suspended or resumed either using new systemctl verbs,
5236 freeze and thaw respectively, or via D-Bus.
5238 * PID 1 may now automatically load pre-compiled AppArmor policies from
5239 /etc/apparmor/earlypolicy during early boot.
5241 * The CPUAffinity= setting in service unit files now supports a new
5242 special value "numa" that causes the CPU affinity masked to be set
5243 based on the NUMA mask.
5245 * systemd will now log about all left-over processes remaining in a
5246 unit when the unit is stopped. It will now warn about services using
5247 KillMode=none, as this is generally an unsafe thing to make use of.
5249 * Two new unit file settings
5250 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
5251 added. They may be used to check whether a specific file system path
5252 resides on a block device that is encrypted on the block level
5253 (i.e. using dm-crypt/LUKS).
5255 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
5256 has been added that may be used for simple environment checks. This
5257 is particularly useful when passing in environment variables from a
5258 container manager (or from PAM in case of the systemd --user
5261 * .service unit files now accept a new setting CoredumpFilter= which
5262 allows configuration of the memory sections coredumps of the
5263 service's processes shall include.
5265 * .mount units gained a new ReadWriteOnly= boolean option. If set
5266 it will not be attempted to mount a file system read-only if mounting
5267 in read-write mode doesn't succeed. An option x-systemd.rw-only is
5268 available in /etc/fstab to control the same.
5270 * .socket units gained a new boolean setting PassPacketInfo=. If
5271 enabled, the kernel will attach additional per-packet metadata to all
5272 packets read from the socket, as an ancillary message. This controls
5273 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
5274 depending on socket type.
5276 * .service units gained a new setting RootHash= which may be used to
5277 specify the root hash for verity enabled disk images which are
5278 specified in RootImage=. RootVerity= may be used to specify a path to
5279 the Verity data matching a RootImage= file system. (The latter is
5280 only useful for images that do not contain the Verity data embedded
5281 into the same image that carries a GPT partition table following the
5282 Discoverable Partition Specification). Similarly, systemd-nspawn
5283 gained a new switch --verity-data= that takes a path to a file with
5284 the verity data of the disk image supplied in --image=, if the image
5285 doesn't contain the verity data itself.
5287 * .service units gained a new setting RootHashSignature= which takes
5288 either a base64 encoded PKCS#7 signature of the root hash specified
5289 with RootHash=, or a path to a file to read the signature from. This
5290 allows validation of the root hash against public keys available in
5291 the kernel keyring, and is only supported on recent kernels
5292 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
5293 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
5294 this mechanism has also been added to systemd-veritysetup.
5296 * .service unit files gained two new options
5297 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
5298 tune behaviour if a start or stop timeout is hit, i.e. whether to
5299 terminate the service with SIGTERM, SIGABRT or SIGKILL.
5301 * Most options in systemd that accept hexadecimal values prefixed with
5302 0x in additional to the usual decimal notation now also support octal
5303 notation when the 0o prefix is used and binary notation if the 0b
5306 * Various command line parameters and configuration file settings that
5307 configure key or certificate files now optionally take paths to
5308 AF_UNIX sockets in the file system. If configured that way a stream
5309 connection is made to the socket and the required data read from
5310 it. This is a simple and natural extension to the existing regular
5311 file logic, and permits other software to provide keys or
5312 certificates via simple IPC services, for example when unencrypted
5313 storage on disk is not desired. Specifically, systemd-networkd's
5314 Wireguard and MACSEC key file settings as well as
5315 systemd-journal-gatewayd's and systemd-journal-remote's PEM
5316 key/certificate parameters support this now.
5318 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
5319 configuration files that support specifier expansion learnt six new
5320 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
5321 resolve to the various ID fields from /etc/os-release, %l resolves to
5322 the "short" hostname of the system, i.e. the hostname configured in
5323 the kernel truncated at the first dot.
5325 * Support for the .include syntax in unit files has been removed. The
5326 concept has been obsolete for 6 years and we started warning about
5327 its pending removal 2 years ago (also see NEWS file below). It's
5330 * StandardError= and StandardOutput= in unit files no longer support
5331 the "syslog" and "syslog-console" switches. They were long removed
5332 from the documentation, but will now result in warnings when used,
5333 and be converted to "journal" and "journal+console" automatically.
5335 * If the service setting User= is set to the "nobody" user, a warning
5336 message is now written to the logs (but the value is nonetheless
5337 accepted). Setting User=nobody is unsafe, since the primary purpose
5338 of the "nobody" user is to own all files whose owner cannot be mapped
5339 locally. It's in particular used by the NFS subsystem and in user
5340 namespacing. By running a service under this user's UID it might get
5341 read and even write access to all these otherwise unmappable files,
5342 which is quite likely a major security problem.
5344 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
5345 and others) now have a size and inode limits applied (50% of RAM for
5346 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
5347 that the implicit kernel default is 50% too, so there is no change
5348 in the size limit for /tmp and /dev/shm.
5350 * nss-mymachines lost support for resolution of users and groups, and
5351 now only does resolution of hostnames. This functionality is now
5352 provided by nss-systemd. Thus, the 'mymachines' entry should be
5353 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
5354 (and 'systemd' added if it is not already there).
5356 * A new kernel command line option systemd.hostname= has been added
5357 that allows controlling the hostname that is initialized early during
5360 * A kernel command line option "udev.blockdev_read_only" has been
5361 added. If specified all hardware block devices that show up are
5362 immediately marked as read-only by udev. This option is useful for
5363 making sure that a specific boot under no circumstances modifies data
5364 on disk. Use "blockdev --setrw" to undo the effect of this, per
5367 * A new boolean kernel command line option systemd.swap= has been
5368 added, which may be used to turn off automatic activation of swap
5369 devices listed in /etc/fstab.
5371 * New kernel command line options systemd.condition-needs-update= and
5372 systemd.condition-first-boot= have been added, which override the
5373 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
5376 * A new kernel command line option systemd.clock-usec= has been added
5377 that allows setting the system clock to the specified time in µs
5378 since Jan 1st, 1970 early during boot. This is in particular useful
5379 in order to make test cases more reliable.
5381 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
5382 systemd-coredump to save core files for suid processes. When saving
5383 the core file, systemd-coredump will use the effective uid and gid of
5384 the process that faulted.
5386 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
5387 now automatically set to "Y" at boot, in order to enable pstore
5388 generation for collection with systemd-pstore.
5390 * We provide a set of udev rules to enable auto-suspend on PCI and USB
5391 devices that were tested to correctly support it. Previously, this
5392 was distributed as a set of udev rules, but has now been replaced by
5393 by a set of hwdb entries (and a much shorter udev rule to take action
5394 if the device modalias matches one of the new hwdb entries).
5396 As before, entries are periodically imported from the database
5397 maintained by the ChromiumOS project. If you have a device that
5398 supports auto-suspend correctly and where it should be enabled by
5399 default, please submit a patch that adds it to the database (see
5400 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
5402 * systemd-udevd gained the new configuration option timeout_signal= as well
5403 as a corresponding kernel command line option udev.timeout_signal=.
5404 The option can be used to configure the UNIX signal that the main
5405 daemon sends to the worker processes on timeout. Setting the signal
5406 to SIGABRT is useful for debugging.
5408 * .link files managed by systemd-udevd gained options RxFlowControl=,
5409 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
5410 order to configure various flow control parameters. They also gained
5411 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
5412 frame ring buffer sizes.
5414 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
5415 enabled systemd-networkd manages all routes configured by other tools.
5417 * .network files managed by systemd-networkd gained a new section
5418 [SR-IOV], in order to configure SR-IOV capable network devices.
5420 * systemd-networkd's [IPv6Prefix] section in .network files gained a
5421 new boolean setting Assign=. If enabled an address from the prefix is
5422 automatically assigned to the interface.
5424 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
5425 controls delegated prefixes assigned by DHCPv6 client. The section
5426 has three settings: SubnetID=, Assign=, and Token=. The setting
5427 SubnetID= allows explicit configuration of the preferred subnet that
5428 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
5429 Assign= is enabled (which is the default) an address from any acquired
5430 delegated prefix is automatically chosen and assigned to the
5431 interface. The setting Token= specifies an optional address generation
5434 * systemd-networkd's [Network] section gained a new setting
5435 IPv4AcceptLocal=. If enabled the interface accepts packets with local
5438 * systemd-networkd gained support for configuring the HTB queuing
5439 discipline in the [HierarchyTokenBucket] and
5440 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
5441 be configured in the [PFIFO] section, "GRED" in
5442 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
5443 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
5444 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
5445 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
5446 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
5447 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
5449 * systemd-networkd gained support for a new Termination= setting in the
5450 [CAN] section for configuring the termination resistor. It also
5451 gained a new ListenOnly= setting for controlling whether to only
5452 listen on CAN interfaces, without interfering with traffic otherwise
5453 (which is useful for debugging/monitoring CAN network
5454 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
5455 been added to configure various CAN-FD aspects.
5457 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
5458 When enabled, DHCPv6 will be attempted right-away without requiring an
5459 Router Advertisement packet suggesting it first (i.e. without the 'M'
5460 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
5461 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
5462 the RA packets suggest it.
5464 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
5465 which may be used to turn off use of the gateway information provided
5466 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
5467 used to configure how to process leases that lack a lifetime option.
5469 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
5470 setting SendVendorOption= allowing configuration of additional vendor
5471 options to send in the DHCP requests/responses. The [DHCPv6] section
5472 gained a new SendOption= setting for sending arbitrary DHCP
5473 options. RequestOptions= has been added to request arbitrary options
5474 from the server. UserClass= has been added to set the DHCP user class
5477 * systemd-networkd's [DHCPServer] section gained a new set of options
5478 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
5479 information about these three protocols in the DHCP lease. It also
5480 gained support for including "MUD" URLs ("Manufacturer Usage
5481 Description"). Support for "MUD" URLs was also added to the LLDP
5482 stack, configurable in the [LLDP] section in .network files.
5484 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
5485 mode. Also, the sections now support a new setting SourceMACAddress=.
5487 * systemd-networkd's .netdev files now support a new setting
5488 VLANProtocol= in the [Bridge] section that allows configuration of
5489 the VLAN protocol to use.
5491 * systemd-networkd supports a new Group= setting in the [Link] section
5492 of the .network files, to control the link group.
5494 * systemd-networkd's [Network] section gained a new
5495 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
5496 link local address is generated.
5498 * A new default .network file is now shipped that matches TUN/TAP
5499 devices that begin with "vt-" in their name. Such interfaces will
5500 have IP routing onto the host links set up automatically. This is
5501 supposed to be used by VM managers to trivially acquire a network
5502 interface which is fully set up for host communication, simply by
5503 carefully picking an interface name to use.
5505 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
5506 which sets the route priority for routes specified by the DHCP server.
5508 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
5509 which configures the vendor class information sent to DHCP server.
5511 * The BlackList= settings in .network files' [DHCPv4] and
5512 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
5513 are still understood to provide compatibility.
5515 * networkctl gained the new "forcerenew" command for forcing all DHCP
5516 server clients to renew their lease. The interface "status" output
5517 will now show numerous additional fields of information about an
5518 interface. There are new "up" and "down" commands to bring specific
5519 interfaces up or down.
5521 * systemd-resolved's DNS= configuration option now optionally accepts a
5522 port number (after ":") and a host name (after "#"). When the host
5523 name is specified, the DNS-over-TLS certificate is validated to match
5524 the specified hostname. Additionally, in case of IPv6 addresses, an
5525 interface may be specified (after "%").
5527 * systemd-resolved may be configured to forward single-label DNS names.
5528 This is not standard-conformant, but may make sense in setups where
5529 public DNS servers are not used.
5531 * systemd-resolved's DNS-over-TLS support gained SNI validation.
5533 * systemd-nspawn's --resolv-conf= switch gained a number of new
5534 supported values. Specifically, options starting with "replace-" are
5535 like those prefixed "copy-" but replace any existing resolv.conf
5536 file. And options ending in "-uplink" and "-stub" can now be used to
5537 propagate other flavours of resolv.conf into the container (as
5538 defined by systemd-resolved).
5540 * The various programs included in systemd can now optionally output
5541 their log messages on stderr prefixed with a timestamp, controlled by
5542 the $SYSTEMD_LOG_TIME environment variable.
5544 * systemctl gained a new "-P" switch that is a shortcut for "--value
5547 * "systemctl list-units" and "systemctl list-machines" no longer hide
5548 their first output column with --no-legend. To hide the first column,
5551 * "systemctl reboot" takes the option "--reboot-argument=".
5552 The optional positional argument to "systemctl reboot" is now
5553 being deprecated in favor of this option.
5555 * systemd-run gained a new switch --slice-inherit. If specified the
5556 unit it generates is placed in the same slice as the systemd-run
5559 * systemd-journald gained support for zstd compression of large fields
5560 in journal files. The hash tables in journal files have been hardened
5561 against hash collisions. This is an incompatible change and means
5562 that journal files created with new systemd versions are not readable
5563 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
5564 environment variable for systemd-journald.service is set to 0 this
5565 new hardening functionality may be turned off, so that generated
5566 journal files remain compatible with older journalctl
5569 * journalctl will now include a clickable link in the default output for
5570 each log message for which a URL with further documentation is
5571 known. This is only supported on terminal emulators that support
5572 clickable hyperlinks, and is turned off if a pager is used (since
5573 "less" still doesn't support hyperlinks,
5574 unfortunately). Documentation URLs may be included in log messages
5575 either by including a DOCUMENTATION= journal field in it, or by
5576 associating a journal message catalog entry with the log message's
5577 MESSAGE_ID, which then carries a "Documentation:" tag.
5579 * journald.conf gained a new boolean setting Audit= that may be used to
5580 control whether systemd-journald will enable audit during
5583 * when systemd-journald's log stream is broken up into multiple lines
5584 because the PID of the sender changed this is indicated in the
5585 generated log records via the _LINE_BREAK=pid-change field.
5587 * journalctl's "-o cat" output mode will now show one or more journal
5588 fields specified with --output-fields= instead of unconditionally
5589 MESSAGE=. This is useful to retrieve a very specific set of fields
5590 without any decoration.
5592 * The sd-journal.h API gained two new functions:
5593 sd_journal_enumerate_available_unique() and
5594 sd_journal_enumerate_available_data() that operate like their
5595 counterparts that lack the _available_ in the name, but skip items
5596 that cannot be read and processed by the local implementation
5597 (i.e. are compressed in an unsupported format or such),
5599 * coredumpctl gained a new --file= switch, matching the same one in
5600 journalctl: a specific journal file may be specified to read the
5603 * coredumps collected by systemd-coredump may now be compressed using
5606 * systemd-binfmt gained a new switch --unregister for unregistering all
5607 registered entries at once. This is now invoked automatically at
5608 shutdown, so that binary formats registered with the "F" flag will
5609 not block clean file system unmounting.
5611 * systemd-notify's --pid= switch gained new values: "parent", "self",
5612 "auto" for controlling which PID to send to the service manager: the
5613 systemd-notify process' PID, or the one of the process invoking it.
5615 * systemd-logind's Session bus object learnt a new method call
5616 SetType() for temporarily updating the session type of an already
5617 allocated session. This is useful for upgrading tty sessions to
5618 graphical ones once a compositor is invoked.
5620 * systemd-socket-proxy gained a new switch --exit-idle-time= for
5621 configuring an exit-on-idle time.
5623 * systemd-repart's --empty= setting gained a new value "create". If
5624 specified a new empty regular disk image file is created under the
5625 specified name. Its size may be specified with the new --size=
5626 option. The latter is also supported without the "create" mode, in
5627 order to grow existing disk image files to the specified size. These
5628 two new options are useful when creating or manipulating disk images
5629 instead of operating on actual block devices.
5631 * systemd-repart drop-ins now support a new UUID= setting to control
5632 the UUID to assign to a newly created partition.
5634 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
5637 * systemd-repart's Label= setting now support the usual, simple
5638 specifier expansion.
5640 * systemd-homed's LUKS backend gained the ability to discard empty file
5641 system blocks automatically when the user logs out. This is enabled
5642 by default to ensure that home directories take minimal space when
5643 logged out but get full size guarantees when logged in. This may be
5644 controlled with the new --luks-offline-discard= switch to homectl.
5646 * If systemd-homed detects that /home/ is encrypted as a whole it will
5647 now default to the directory or subvolume backends instead of the
5648 LUKS backend, in order to avoid double encryption. The default
5649 storage and file system may now be configured explicitly, too, via
5650 the new /etc/systemd/homed.conf configuration file.
5652 * systemd-homed now supports unlocking home directories with FIDO2
5653 security tokens that support the 'hmac-secret' extension, in addition
5654 to the existing support for PKCS#11 security token unlocking
5655 support. Note that many recent hardware security tokens support both
5656 interfaces. The FIDO2 support is accessible via homectl's
5657 --fido2-device= option.
5659 * homectl's --pkcs11-uri= setting now accepts two special parameters:
5660 if "auto" is specified and only one suitable PKCS#11 security token
5661 is plugged in, its URL is automatically determined and enrolled for
5662 unlocking the home directory. If "list" is specified a brief table of
5663 suitable PKCS#11 security tokens is shown. Similar, the new
5664 --fido2-device= option also supports these two special values, for
5665 automatically selecting and listing suitable FIDO2 devices.
5667 * The /etc/crypttab tmp option now optionally takes an argument
5668 selecting the file system to use. Moreover, the default is now
5669 changed from ext2 to ext4.
5671 * There's a new /etc/crypttab option "keyfile-erase". If specified the
5672 key file listed in the same line is removed after use, regardless if
5673 volume activation was successful or not. This is useful if the key
5674 file is only acquired transiently at runtime and shall be erased
5675 before the system continues to boot.
5677 * There's also a new /etc/crypttab option "try-empty-password". If
5678 specified, before asking the user for a password it is attempted to
5679 unlock the volume with an empty password. This is useful for
5680 installing encrypted images whose password shall be set on first boot
5681 instead of at installation time.
5683 * systemd-cryptsetup will now attempt to load the keys to unlock
5684 volumes with automatically from files in
5685 /etc/cryptsetup-keys.d/<volume>.key and
5686 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
5688 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
5689 /etc/crypttab, during boot.
5691 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
5692 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
5695 * A new generator systemd-xdg-autostart-generator has been added. It
5696 generates systemd unit files from XDG autostart .desktop files, and
5697 may be used to let the systemd user instance manage services that are
5698 started automatically as part of the desktop session.
5700 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
5701 to query and change the firmware's 'reboot into firmware' setup flag.
5703 * systemd-firstboot gained a new switch --kernel-command-line= that may
5704 be used to initialize the /etc/kernel/cmdline file of the image. It
5705 also gained a new switch --root-password-hashed= which is like
5706 --root-password= but accepts a pre-hashed UNIX password as
5707 argument. The new option --delete-root-password may be used to unset
5708 any password for the root user (dangerous!). The --root-shell= switch
5709 may be used to control the shell to use for the root account. A new
5710 --force option may be used to override any already set settings with
5711 the parameters specified on the command line (by default, the tool
5712 will not override what has already been set before, i.e. is purely
5715 * systemd-firstboot gained support for a new --image= switch, which is
5716 similar to --root= but accepts the path to a disk image file, on
5717 which it then operates.
5719 * A new sd-path.h API has been added to libsystemd. It provides a
5720 simple API for retrieving various search paths and primary
5721 directories for various resources.
5723 * A new call sd_notify_barrier() has been added to the sd-daemon.h
5724 API. The call will block until all previously sent sd_notify()
5725 messages have been processed by the service manager. This is useful
5726 to remove races caused by a process already having disappeared at the
5727 time a notification message is processed by the service manager,
5728 making correct attribution impossible. The systemd-notify tool will
5729 now make use of this call implicitly, but this can be turned off again
5730 via the new --no-block switch.
5732 * When sending a file descriptor (fd) to the service manager to keep
5733 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
5734 may be specified. If passed the service manager will refrain from
5735 poll()ing on the file descriptor. Traditionally (and when the
5736 parameter is not specified), the service manager will poll it for
5737 POLLHUP or POLLERR events, and immediately close the fds in that
5740 * The service manager (PID1) gained a new D-Bus method call
5741 SetShowStatus() which may be used to control whether it shall show
5742 boot-time status output on the console. This method has a similar
5743 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
5745 * The sd-bus API gained a number of convenience functions that take
5746 va_list arguments rather than "...". For example, there's now
5747 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
5748 it easier to build wrappers that accept variadic arguments and want
5749 to pass a ready va_list structure to sd-bus.
5751 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
5752 flag which alters how the userdata pointer to pass to the callbacks
5753 is determined. When the flag is set, the offset field is converted
5754 as-is into a pointer, without adding it to the object pointer the
5755 vtable is associated with.
5757 * sd-bus now exposes four new functions:
5758 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
5759 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
5760 validate strings to check if they qualify as various D-Bus concepts.
5762 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
5763 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
5764 that simplify adding argument names to D-Bus methods and signals.
5766 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
5768 * Various D-Bus APIs of systemd daemons now have man pages that
5769 document the methods, signals and properties.
5771 * The expectations on user/group name syntax are now documented in
5772 detail; documentation on how classic home directories may be
5773 converted into home directories managed by homed has been added;
5774 documentation regarding integration of homed/userdb functionality in
5775 desktops has been added:
5777 https://systemd.io/USER_NAMES
5778 https://systemd.io/CONVERTING_TO_HOMED
5779 https://systemd.io/USERDB_AND_DESKTOPS
5781 * Documentation for the on-disk Journal file format has been updated
5782 and has now moved to:
5784 https://systemd.io/JOURNAL_FILE_FORMAT
5786 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
5787 has been extended by a set of environment variables that expose
5788 select fields from the host's os-release file to the container
5789 payload. Similarly, host's os-release files can be mounted into the
5790 container underneath /run/host. Together, those mechanisms provide a
5791 standardized way to expose information about the host to the
5792 container payload. Both interfaces are implemented in systemd-nspawn.
5794 * All D-Bus services shipped in systemd now implement the generic
5795 LogControl1 D-Bus API which allows clients to change log level +
5796 target of the service during runtime.
5798 * Only relevant for developers: the mkosi.default symlink has been
5799 dropped from version control. Please create a symlink to one of the
5800 distribution-specific defaults in .mkosi/ based on your preference.
5802 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
5803 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
5804 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
5805 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
5806 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
5807 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
5808 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
5809 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
5810 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
5811 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
5812 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
5813 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
5814 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
5815 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
5816 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
5817 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
5818 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
5819 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
5820 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
5821 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
5822 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
5823 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
5824 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
5825 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
5826 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
5827 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
5828 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
5829 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
5830 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
5831 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
5832 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
5833 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
5834 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
5835 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
5836 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
5837 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
5838 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
5839 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
5841 – Warsaw, 2020-07-30
5845 * A new tool "systemd-repart" has been added, that operates as an
5846 idempotent declarative repartitioner for GPT partition tables.
5847 Specifically, a set of partitions that must or may exist can be
5848 configured via drop-in files, and during every boot the partition
5849 table on disk is compared with these files, creating missing
5850 partitions or growing existing ones based on configurable relative
5851 and absolute size constraints. The tool is strictly incremental,
5852 i.e. does not delete, shrink or move partitions, but only adds and
5853 grows them. The primary use-case is OS images that ship in minimized
5854 form, that on first boot are grown to the size of the underlying
5855 block device or augmented with additional partitions. For example,
5856 the root partition could be extended to cover the whole disk, or a
5857 swap or /home partitions could be added on first boot. It can also be
5858 used for systems that use an A/B update scheme but ship images with
5859 just the A partition, with B added on first boot. The tool is
5860 primarily intended to be run in the initrd, shortly before
5861 transitioning into the host OS, but can also be run after the
5862 transition took place. It automatically discovers the disk backing
5863 the root file system, and should hence not require any additional
5864 configuration besides the partition definition drop-ins. If no
5865 configuration drop-ins are present, no action is taken.
5867 * A new component "userdb" has been added, along with a small daemon
5868 "systemd-userdbd.service" and a client tool "userdbctl". The framework
5869 allows defining rich user and group records in a JSON format,
5870 extending on the classic "struct passwd" and "struct group"
5871 structures. Various components in systemd have been updated to
5872 process records in this format, including systemd-logind and
5873 pam-systemd. The user records are intended to be extensible, and
5874 allow setting various resource management, security and runtime
5875 parameters that shall be applied to processes and sessions of the
5876 user as they log in. This facility is intended to allow associating
5877 such metadata directly with user/group records so that they can be
5878 produced, extended and consumed in unified form. We hope that
5879 eventually frameworks such as sssd will generate records this way, so
5880 that for the first time resource management and various other
5881 per-user settings can be configured in LDAP directories and then
5882 provided to systemd (specifically to systemd-logind and pam-system)
5883 to apply on login. For further details see:
5885 https://systemd.io/USER_RECORD
5886 https://systemd.io/GROUP_RECORD
5887 https://systemd.io/USER_GROUP_API
5889 * A small new service systemd-homed.service has been added, that may be
5890 used to securely manage home directories with built-in encryption.
5891 The complete user record data is unified with the home directory,
5892 thus making home directories naturally migratable. Its primary
5893 back-end is based on LUKS volumes, but fscrypt, plain directories,
5894 and other storage schemes are also supported. This solves a couple of
5895 problems we saw with traditional ways to manage home directories, in
5896 particular when it comes to encryption. For further discussion of
5897 this, see the video of Lennart's talk at AllSystemsGo! 2019:
5899 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
5901 For further details about the format and expectations on home
5902 directories this new daemon makes, see:
5904 https://systemd.io/HOME_DIRECTORY
5906 * systemd-journald is now multi-instantiable. In addition to the main
5907 instance systemd-journald.service there's now a template unit
5908 systemd-journald@.service, with each instance defining a new named
5909 log 'namespace' (whose name is specified via the instance part of the
5910 unit name). A new unit file setting LogNamespace= has been added,
5911 taking such a namespace name, that assigns services to the specified
5912 log namespaces. As each log namespace is serviced by its own
5913 independent journal daemon, this functionality may be used to improve
5914 performance and increase isolation of applications, at the price of
5915 losing global message ordering. Each instance of journald has a
5916 separate set of configuration files, with possibly different disk
5917 usage limitations and other settings.
5919 journalctl now takes a new option --namespace= to show logs from a
5920 specific log namespace. The sd-journal.h API gained
5921 sd_journal_open_namespace() for opening the log stream of a specific
5922 log namespace. systemd-journald also gained the ability to exit on
5923 idle, which is useful in the context of log namespaces, as this means
5924 log daemons for log namespaces can be activated automatically on
5925 demand and will stop automatically when no longer used, minimizing
5928 * When systemd-tmpfiles copies a file tree using the 'C' line type it
5929 will now label every copied file according to the SELinux database.
5931 * When systemd/PID 1 detects it is used in the initrd it will now boot
5932 into initrd.target rather than default.target by default. This should
5933 make it simpler to build initrds with systemd as for many cases the
5934 only difference between a host OS image and an initrd image now is
5935 the presence of the /etc/initrd-release file.
5937 * A new kernel command line option systemd.cpu_affinity= is now
5938 understood. It's equivalent to the CPUAffinity= option in
5939 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
5940 itself and the default for all other processes.
5942 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
5943 equivalent), the SELinux database is now reloaded, ensuring that
5944 sockets and other file system objects are generated taking the new
5945 database into account.
5947 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
5948 "quiet" has been changed to imply that instead of
5949 "systemd.show-status=auto". In this mode, only messages about errors
5950 and significant delays in boot are shown on the console.
5952 * The sd-event.h API gained native support for the new Linux "pidfd"
5953 concept. This permits watching processes using file descriptors
5954 instead of PID numbers, which fixes a number of races and makes
5955 process supervision more robust and efficient. All of systemd's
5956 components will now use pidfds if the kernel supports it for process
5957 watching, with the exception of PID 1 itself, unfortunately. We hope
5958 to move PID 1 to exclusively using pidfds too eventually, but this
5959 requires some more kernel work first. (Background: PID 1 watches
5960 processes using waitid() with the P_ALL flag, and that does not play
5961 together nicely with pidfds yet.)
5963 * Closely related to this, the sd-event.h API gained two new calls
5964 sd_event_source_send_child_signal() (for sending a signal to a
5965 watched process) and sd_event_source_get_child_process_own() (for
5966 marking a process so that it is killed automatically whenever the
5967 event source watching it is freed).
5969 * systemd-networkd gained support for configuring Token Bucket Filter
5970 (TBF) parameters in its qdisc configuration support. Similarly,
5971 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
5972 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
5974 * systemd-networkd gained support for Intermediate Functional Block
5975 (IFB) network devices.
5977 * systemd-networkd gained support for configuring multi-path IP routes,
5978 using the new MultiPathRoute= setting in the [Route] section.
5980 * systemd-networkd's DHCPv4 client has been updated to support a new
5981 SendDecline= option. If enabled, duplicate address detection is done
5982 after a DHCP offer is received from the server. If a conflict is
5983 detected, the address is declined. The DHCPv4 client also gained
5984 support for a new RouteMTUBytes= setting that allows to configure the
5985 MTU size to be used for routes generated from DHCPv4 leases.
5987 * The PrefixRoute= setting in systemd-networkd's [Address] section of
5988 .network files has been deprecated, and replaced by AddPrefixRoute=,
5989 with its sense inverted.
5991 * The Gateway= setting of [Route] sections of .network files gained
5992 support for a special new value "_dhcp". If set, the configured
5993 static route uses the gateway host configured via DHCP.
5995 * New User= and SuppressPrefixLength= settings have been implemented
5996 for the [RoutingPolicyRule] section of .network files to configure
5997 source routing based on UID ranges and prefix length, respectively.
5999 * The Type= match property of .link files has been generalized to
6000 always match the device type shown by 'networkctl status', even for
6001 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
6004 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
6005 D-Bus message object as "sensitive". Those objects are erased from
6006 memory when they are freed. This concept is intended to be used for
6007 messages that contain security sensitive data. A new flag
6008 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
6009 in sd-bus vtables, causing any incoming and outgoing messages of
6010 those methods to be implicitly marked as "sensitive".
6012 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
6013 contents of a message (or parts thereof) to standard output for
6016 * systemd-sysusers gained support for creating users with the primary
6017 group named differently than the user.
6019 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
6020 gained support for growing XFS partitions. Previously it supported
6021 only ext4 and btrfs partitions.
6023 * The support for /etc/crypttab gained a new x-initrd.attach option. If
6024 set, the specified encrypted volume is unlocked already in the
6025 initrd. This concept corresponds to the x-initrd.mount option in
6028 * systemd-cryptsetup gained native support for unlocking encrypted
6029 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
6030 encryption of volumes to YubiKeys. This is exposed in the new
6031 pkcs11-uri= option in /etc/crypttab.
6033 * The /etc/fstab support in systemd now supports two new mount options
6034 x-systemd.{required,wanted}-by=, for explicitly configuring the units
6035 that the specified mount shall be pulled in by, in place of
6036 the usual local-fs.target/remote-fs.target.
6038 * The https://systemd.io/ web site has been relaunched, directly
6039 populated with most of the documentation included in the systemd
6040 repository. systemd also acquired a new logo, thanks to Tobias
6043 * systemd-udevd gained support for managing "alternative" network
6044 interface names, as supported by new Linux kernels. For the first
6045 time this permits assigning multiple (and longer!) names to a network
6046 interface. systemd-udevd will now by default assign the names
6047 generated via all supported naming schemes to each interface. This
6048 may be further tweaked with .link files and the AlternativeName= and
6049 AlternativeNamesPolicy= settings. Other components of systemd have
6050 been updated to support the new alternative names wherever
6051 appropriate. For example, systemd-nspawn will now generate
6052 alternative interface names for the host-facing side of container
6053 veth links based on the full container name without truncation.
6055 * systemd-nspawn interface naming logic has been updated in another way
6056 too: if the main interface name (i.e. as opposed to new-style
6057 "alternative" names) based on the container name is truncated, a
6058 simple hashing scheme is used to give different interface names to
6059 multiple containers whose names all begin with the same prefix. Since
6060 this changes the primary interface names pointing to containers if
6061 truncation happens, the old scheme may still be requested by
6062 selecting an older naming scheme, via the net.naming-scheme= kernel
6063 command line option.
6065 * PrivateUsers= in service files now works in services run by the
6066 systemd --user per-user instance of the service manager.
6068 * A new per-service sandboxing option ProtectClock= has been added that
6069 locks down write access to the system clock. It takes away device
6070 node access to /dev/rtc as well as the system calls that set the
6071 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
6072 Note that this option does not affect access to auxiliary services
6073 that allow changing the clock, for example access to
6076 * The systemd-id128 tool gained a new "show" verb for listing or
6077 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
6078 GPT partition table types.
6080 * The Discoverable Partitions Specification has been updated to support
6081 /var and /var/tmp partition discovery. Support for this has been
6082 added to systemd-gpt-auto-generator. For details see:
6084 https://systemd.io/DISCOVERABLE_PARTITIONS
6086 * "systemctl list-unit-files" has been updated to show a new column
6087 with the suggested enablement state based on the vendor preset files
6088 for the respective units.
6090 * "systemctl" gained a new option "--with-dependencies". If specified
6091 commands such as "systemctl status" or "systemctl cat" will now show
6092 all specified units along with all units they depend on.
6094 * networkctl gained support for showing per-interface logs in its
6097 * systemd-networkd-wait-online gained support for specifying the maximum
6098 operational state to wait for, and to wait for interfaces to
6101 * The [Match] section of .link and .network files now supports a new
6102 option PermanentMACAddress= which may be used to check against the
6103 permanent MAC address of a network device even if a randomized MAC
6106 * The [TrafficControlQueueingDiscipline] section in .network files has
6107 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
6108 dropped from the individual setting names.
6110 * Any .link and .network files that have an empty [Match] section (this
6111 also includes empty and commented-out files) will now be
6112 rejected. systemd-udev and systemd-networkd started warning about
6113 such files in version 243.
6115 * systemd-logind will now validate access to the operation of changing
6116 the virtual terminal via a polkit action. By default, only users
6117 with at least one session on a local VT are granted permission.
6119 * When systemd sets up PAM sessions that invoked service processes
6120 shall run in, the pam_setcred() API is now invoked, thus permitting
6121 PAM modules to set additional credentials for the processes.
6123 * portablectl attach/detach verbs now accept --now and --enable options
6124 to combine attachment with enablement and invocation, or detachment
6125 with stopping and disablement.
6127 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
6128 fixed, which in turn exposed bugs in unit configuration of services
6129 which have Type=oneshot and should only run once, but do not have
6130 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
6131 service may be started again after exiting successfully, for example
6132 as a dependency in another transaction. Affected services included
6133 some internal systemd services (most notably
6134 systemd-vconsole-setup.service, which was updated to have
6135 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
6136 plymouth has been suitably updated or patched before upgrading to
6137 this systemd release. See
6138 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
6139 additional discussion.
6141 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
6142 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
6143 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
6144 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
6145 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
6146 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
6147 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
6148 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
6149 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
6150 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
6151 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
6152 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
6153 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
6154 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
6155 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
6156 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
6157 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
6158 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
6159 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
6160 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
6161 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
6162 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
6163 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
6164 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
6167 – Warsaw, 2020-03-06
6171 * Support for the cpuset cgroups v2 controller has been added.
6172 Processes may be restricted to specific CPUs using the new
6173 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
6174 AllowedMemoryNodes= setting.
6176 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
6177 now be configured using a new RestartKillSignal= setting. This
6178 allows units which signals to request termination to implement
6179 different behaviour when stopping in preparation for a restart.
6181 * "systemctl clean" may now be used also for socket, mount, and swap
6184 * systemd will also read configuration options from the EFI variable
6185 SystemdOptions. This may be used to configure systemd behaviour when
6186 modifying the kernel command line is inconvenient, but configuration
6187 on disk is read too late, for example for the options related to
6188 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
6189 set the EFI variable.
6191 * systemd will now disable printk ratelimits in early boot. This should
6192 allow us to capture more logs from the early boot phase where normal
6193 storage is not available and the kernel ring buffer is used for
6194 logging. Configuration on the kernel command line has higher priority
6195 and overrides the systemd setting.
6197 systemd programs which log to /dev/kmsg directly use internal
6198 ratelimits to prevent runaway logging. (Normally this is only used
6199 during early boot, so in practice this change has very little
6202 * Unit files now support top level dropin directories of the form
6203 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
6204 that affects all corresponding unit files.
6206 * systemctl gained support for 'stop --job-mode=triggering' which will
6207 stop the specified unit and any units which could trigger it.
6209 * Unit status display now includes units triggering and triggered by
6210 the unit being shown.
6212 * The RuntimeMaxSec= setting is now supported by scopes, not just
6213 .service units. This is particularly useful for PAM sessions which
6214 create a scope unit for the user login. systemd.runtime_max_sec=
6215 setting may used with the pam_systemd module to limit the duration
6216 of the PAM session, for example for time-limited logins.
6218 * A new @pkey system call group is now defined to make it easier to
6219 allow-list memory protection syscalls for containers and services
6220 which need to use them.
6222 * systemd-udevd: removed the 30s timeout for killing stale workers on
6223 exit. systemd-udevd now waits for workers to finish. The hard-coded
6224 exit timeout of 30s was too short for some large installations, where
6225 driver initialization could be prematurely interrupted during initrd
6226 processing if the root file system had been mounted and init was
6227 preparing to switch root. If udevd is run without systemd and workers
6228 are hanging while udevd receives an exit signal, udevd will now exit
6229 when udev.event_timeout is reached for the last hanging worker. With
6230 systemd, the exit timeout can additionally be configured using
6231 TimeoutStopSec= in systemd-udevd.service.
6233 * udev now provides a program (fido_id) that identifies FIDO CTAP1
6234 ("U2F")/CTAP2 security tokens based on the usage declared in their
6235 report and descriptor and outputs suitable environment variables.
6236 This replaces the externally maintained allow lists of all known
6237 security tokens that were used previously.
6239 * Automatically generated autosuspend udev rules for allow-listed
6240 devices have been imported from the Chromium OS project. This should
6241 improve power saving with many more devices.
6243 * udev gained a new "CONST{key}=value" setting that allows matching
6244 against system-wide constants without forking a helper binary.
6245 Currently "arch" and "virt" keys are supported.
6247 * udev now opens CDROMs in non-exclusive mode when querying their
6248 capabilities. This should fix issues where other programs trying to
6249 use the CDROM cannot gain access to it, but carries a risk of
6250 interfering with programs writing to the disk, if they did not open
6251 the device in exclusive mode as they should.
6253 * systemd-networkd does not create a default route for IPv4 link local
6254 addressing anymore. The creation of the route was unexpected and was
6255 breaking routing in various cases, but people who rely on it being
6256 created implicitly will need to adjust. Such a route may be requested
6257 with DefaultRouteOnDevice=yes.
6259 Similarly, systemd-networkd will not assign a link-local IPv6 address
6260 when IPv6 link-local routing is not enabled.
6262 * Receive and transmit buffers may now be configured on links with
6263 the new RxBufferSize= and TxBufferSize= settings.
6265 * systemd-networkd may now advertise additional IPv6 routes. A new
6266 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
6269 * systemd-networkd may now configure "next hop" routes using the
6270 [NextHop] section and Gateway= and Id= settings.
6272 * systemd-networkd will now retain DHCP config on restarts by default
6273 (but this may be overridden using the KeepConfiguration= setting).
6274 The default for SendRelease= has been changed to true.
6276 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
6277 received from the server.
6279 The client will use the received SIP server list if UseSIP=yes is
6282 The client may be configured to request specific options from the
6283 server using a new RequestOptions= setting.
6285 The client may be configured to send arbitrary options to the server
6286 using a new SendOption= setting.
6288 A new IPServiceType= setting has been added to configure the "IP
6289 service type" value used by the client.
6291 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
6292 request prefix hints in the DHCPv6 solicitation.
6294 * The DHCPv4 server may be configured to send arbitrary options using
6295 a new SendOption= setting.
6297 * The DHCPv4 server may now be configured to emit SIP server list using
6298 the new EmitSIP= and SIP= settings.
6300 * systemd-networkd and networkctl may now renew DHCP leases on demand.
6301 networkctl has a new 'networkctl renew' verb.
6303 * systemd-networkd may now reconfigure links on demand. networkctl
6304 gained two new verbs: "reload" will reload the configuration, and
6305 "reconfigure DEVICE…" will reconfigure one or more devices.
6307 * .network files may now match on SSID and BSSID of a wireless network,
6308 i.e. the access point name and hardware address using the new SSID=
6309 and BSSID= options. networkctl will display the current SSID and
6310 BSSID for wireless links.
6312 .network files may also match on the wireless network type using the
6313 new WLANInterfaceType= option.
6315 * systemd-networkd now includes default configuration that enables
6316 link-local addressing when connected to an ad-hoc wireless network.
6318 * systemd-networkd may configure the Traffic Control queueing
6319 disciplines in the kernel using the new
6320 [TrafficControlQueueingDiscipline] section and Parent=,
6321 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
6322 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
6323 NetworkEmulatorDuplicateRate= settings.
6325 * systemd-tmpfiles gained a new w+ setting to append to files.
6327 * systemd-analyze dump will now report when the memory configuration in
6328 the kernel does not match what systemd has configured (usually,
6329 because some external program has modified the kernel configuration
6332 * systemd-analyze gained a new --base-time= switch instructs the
6333 'calendar' verb to resolve times relative to that timestamp instead
6334 of the present time.
6336 * journalctl --update-catalog now produces deterministic output (making
6337 reproducible image builds easier).
6339 * A new devicetree-overlay setting is now documented in the Boot Loader
6342 * The default value of the WatchdogSec= setting used in systemd
6343 services (the ones bundled with the project itself) may be set at
6344 configuration time using the -Dservice-watchdog= setting. If set to
6345 empty, the watchdogs will be disabled.
6347 * systemd-resolved validates IP addresses in certificates now when GnuTLS
6350 * libcryptsetup >= 2.0.1 is now required.
6352 * A configuration option -Duser-path= may be used to override the $PATH
6353 used by the user service manager. The default is again to use the same
6354 path as the system manager.
6356 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
6357 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
6360 * Service units gained a new sandboxing option ProtectKernelLogs= which
6361 makes sure the program cannot get direct access to the kernel log
6362 buffer anymore, i.e. the syslog() system call (not to be confused
6363 with the API of the same name in libc, which is not affected), the
6364 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
6365 inaccessible to the service. It's recommended to enable this setting
6366 for all services that should not be able to read from or write to the
6367 kernel log buffer, which are probably almost all.
6369 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
6370 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
6371 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
6372 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
6373 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
6374 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
6375 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
6376 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
6377 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
6378 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
6379 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
6380 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
6381 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
6382 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
6383 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
6384 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
6385 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
6386 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
6387 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
6388 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
6389 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
6390 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
6391 Zach Smith, Zbigniew Jędrzejewski-Szmek
6393 – Warsaw, 2019-11-29
6397 * This release enables unprivileged programs (i.e. requiring neither
6398 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
6399 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
6400 kernel for the whole UNIX group range, i.e. all processes. This
6401 change should be reasonably safe, as the kernel support for it was
6402 specifically implemented to allow safe access to ICMP Echo for
6403 processes lacking any privileges. If this is not desirable, it can be
6404 disabled again by setting the parameter to "1 0".
6406 * Previously, filters defined with SystemCallFilter= would have the
6407 effect that any calling of an offending system call would terminate
6408 the calling thread. This behaviour never made much sense, since
6409 killing individual threads of unsuspecting processes is likely to
6410 create more problems than it solves. With this release the default
6411 action changed from killing the thread to killing the whole
6412 process. For this to work correctly both a kernel version (>= 4.14)
6413 and a libseccomp version (>= 2.4.0) supporting this new seccomp
6414 action is required. If an older kernel or libseccomp is used the old
6415 behaviour continues to be used. This change does not affect any
6416 services that have no system call filters defined, or that use
6417 SystemCallErrorNumber= (and thus see EPERM or another error instead
6418 of being killed when calling an offending system call). Note that
6419 systemd documentation always claimed that the whole process is
6420 killed. With this change behaviour is thus adjusted to match the
6423 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
6424 4194304 by default, i.e. the full 22bit range the kernel allows, up
6425 from the old 16-bit range. This should improve security and
6426 robustness, as PID collisions are made less likely (though certainly
6427 still possible). There are rumours this might create compatibility
6428 problems, though at this moment no practical ones are known to
6429 us. Downstream distributions are hence advised to undo this change in
6430 their builds if they are concerned about maximum compatibility, but
6431 for everybody else we recommend leaving the value bumped. Besides
6432 improving security and robustness this should also simplify things as
6433 the maximum number of allowed concurrent tasks was previously bounded
6434 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
6435 only a single knob is left ("kernel.threads-max"). There have been
6436 concerns that usability is affected by this change because larger PID
6437 numbers are harder to type, but we believe the change from 5 digits
6438 to 7 digits doesn't hamper usability.
6440 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
6441 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
6442 hierarchically set default memory protection values for a particular
6443 subtree of the unit hierarchy.
6445 * Memory protection directives can now take a value of zero, allowing
6446 explicit opting out of a default value propagated by an ancestor.
6448 * systemd now defaults to the "unified" cgroup hierarchy setup during
6449 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
6450 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
6451 change reflects the fact that cgroupsv2 support has matured
6452 substantially in both systemd and in the kernel, and is clearly the
6453 way forward. Downstream production distributions might want to
6454 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
6455 their builds as unfortunately the popular container managers have not
6456 caught up with the kernel API changes.
6458 * Man pages are not built by default anymore (html pages were already
6459 disabled by default), to make development builds quicker. When
6460 building systemd for a full installation with documentation, meson
6461 should be called with -Dman=true and/or -Dhtml=true as appropriate.
6462 The default was changed based on the assumption that quick one-off or
6463 repeated development builds are much more common than full optimized
6464 builds for installation, and people need to pass various other
6465 options to when doing "proper" builds anyway, so the gain from making
6466 development builds quicker is bigger than the one time disruption for
6469 Two scripts are created in the *build* directory to generate and
6470 preview man and html pages on demand, e.g.:
6472 build/man/man systemctl
6473 build/man/html systemd.index
6475 * libidn2 is used by default if both libidn2 and libidn are installed.
6476 Please use -Dlibidn=true if libidn is preferred.
6478 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
6479 big-endian machines. Before, bytes were written and read in native
6480 machine order as exposed by the native libc __cpu_mask interface.
6481 Now, little-endian order is always used (CPUs 0–7 are described by
6482 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
6483 This change fixes D-Bus calls that cross endianness boundary.
6485 The presentation format used for CPUAffinity= by "systemctl show" and
6486 "systemd-analyze dump" is changed to present CPU indices instead of
6487 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
6488 shown as CPUAffinity=03000000000000000000000000000… (on
6489 little-endian) or CPUAffinity=00000000000000300000000000000… (on
6490 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
6491 input format. The maximum integer that will be printed in the new
6492 format is 8191 (four digits), while the old format always used a very
6493 long number (with the length varying by architecture), so they can be
6494 unambiguously distinguished.
6496 * /usr/sbin/halt.local is no longer supported. Implementation in
6497 distributions was inconsistent and it seems this functionality was
6500 To replace this functionality, users should:
6501 - either define a new unit and make it a dependency of final.target
6502 (systemctl add-wants final.target my-halt-local.service)
6503 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
6504 and ensure that it accepts "halt", "poweroff", "reboot", and
6505 "kexec" as an argument, see the description in systemd-shutdown(8).
6507 * When a [Match] section in .link or .network file is empty (contains
6508 no match patterns), a warning will be emitted. Please add any "match
6509 all" pattern instead, e.g. OriginalName=* or Name=* in case all
6510 interfaces should really be matched.
6512 * A new setting NUMAPolicy= may be used to set process memory
6513 allocation policy. This setting can be specified in
6514 /etc/systemd/system.conf and hence will set the default policy for
6515 PID1. The default policy can be overridden on a per-service
6516 basis. The related setting NUMAMask= is used to specify NUMA node
6517 mask that should be associated with the selected policy.
6519 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
6520 generates when processes it manages are reaching their memory limits,
6521 and will place their units in a special state, and optionally kill or
6522 stop the whole unit.
6524 * The service manager will now expose bus properties for the IO
6525 resources used by units. This information is also shown in "systemctl
6526 status" now (for services that have IOAccounting=yes set). Moreover,
6527 the IO accounting data is included in the resource log message
6528 generated whenever a unit stops.
6530 * Units may now configure an explicit timeout to wait for when killed
6531 with SIGABRT, for example when a service watchdog is hit. Previously,
6532 the regular TimeoutStopSec= timeout was applied in this case too —
6533 now a separate timeout may be set using TimeoutAbortSec=.
6535 * Services may now send a special WATCHDOG=trigger message with
6536 sd_notify() to trigger an immediate "watchdog missed" event, and thus
6537 trigger service termination. This is useful both for testing watchdog
6538 handling, but also for defining error paths in services, that shall
6539 be handled the same way as watchdog events.
6541 * There are two new per-unit settings IPIngressFilterPath= and
6542 IPEgressFilterPath= which allow configuration of a BPF program
6543 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
6544 to apply to the IP packet ingress/egress path of all processes of a
6545 unit. This is useful to allow running systemd services with BPF
6546 programs set up externally.
6548 * systemctl gained a new "clean" verb for removing the state, cache,
6549 runtime or logs directories of a service while it is terminated. The
6550 new verb may also be used to remove the state maintained on disk for
6551 timer units that have Persistent= configured.
6553 * During the last phase of shutdown systemd will now automatically
6554 increase the log level configured in the "kernel.printk" sysctl so
6555 that any relevant loggable events happening during late shutdown are
6556 made visible. Previously, loggable events happening so late during
6557 shutdown were generally lost if the "kernel.printk" sysctl was set to
6558 high thresholds, as regular logging daemons are terminated at that
6559 time and thus nothing is written to disk.
6561 * If processes terminated during the last phase of shutdown do not exit
6562 quickly systemd will now show their names after a short time, to make
6563 debugging easier. After a longer timeout they are forcibly killed,
6566 * journalctl (and the other tools that display logs) will now highlight
6567 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
6568 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
6569 are now shown in blue color, to separate them visually from regular
6570 logs. References to configuration files are now turned into clickable
6571 links on terminals that support that.
6573 * systemd-journald will now stop logging to /var/log/journal during
6574 shutdown when /var/ is on a separate mount, so that it can be
6575 unmounted safely during shutdown.
6577 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
6579 * systemd-resolved "Cache=" configuration option in resolved.conf has
6580 been extended to also accept the 'no-negative' value. Previously,
6581 only a boolean option was allowed (yes/no), having yes as the
6582 default. If this option is set to 'no-negative', negative answers are
6583 not cached while the old cache heuristics are used positive answers.
6584 The default remains unchanged.
6586 * The predictable naming scheme for network devices now supports
6587 generating predictable names for "netdevsim" devices.
6589 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
6592 Those two changes form a new net.naming-policy-scheme= entry.
6593 Distributions which want to preserve naming stability may want to set
6594 the -Ddefault-net-naming-scheme= configuration option.
6596 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
6597 interfaces natively.
6599 * systemd-networkd's bridge FDB support now allows configuration of a
6600 destination address for each entry (Destination=), as well as the
6601 VXLAN VNI (VNI=), as well as an option to declare what an entry is
6602 associated with (AssociatedWith=).
6604 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
6605 option for configuring the maximum number of DHCP lease requests. It
6606 also learnt a new BlackList= option for deny-listing DHCP servers (a
6607 similar setting has also been added to the IPv6 RA client), as well
6608 as a SendRelease= option for configuring whether to send a DHCP
6609 RELEASE message when terminating.
6611 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
6612 separately in the [DHCPv4] and [DHCPv6] sections.
6614 * systemd-networkd's DHCP support will now optionally create an
6615 implicit host route to the DNS server specified in the DHCP lease, in
6616 addition to the routes listed explicitly in the lease. This should
6617 ensure that in multi-homed systems DNS traffic leaves the systems on
6618 the interface that acquired the DNS server information even if other
6619 routes such as default routes exist. This behaviour may be turned on
6620 with the new RoutesToDNS= option.
6622 * systemd-networkd's VXLAN support gained a new option
6623 GenericProtocolExtension= for enabling VXLAN Generic Protocol
6624 Extension support, as well as IPDoNotFragment= for setting the IP
6625 "Don't fragment" bit on outgoing packets. A similar option has been
6626 added to the GENEVE support.
6628 * In systemd-networkd's [Route] section you may now configure
6629 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
6630 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
6631 propagation. The Type= setting now supports local, broadcast,
6632 anycast, multicast, any, xresolve routes, too.
6634 * systemd-networkd's [Network] section learnt a new option
6635 DefaultRouteOnDevice= for automatically configuring a default route
6636 onto the network device.
6638 * systemd-networkd's bridging support gained two new options ProxyARP=
6639 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
6640 MulticastRouter= for configuring multicast routing behaviour. A new
6641 option MulticastIGMPVersion= may be used to change bridge's multicast
6642 Internet Group Management Protocol (IGMP) version.
6644 * systemd-networkd's FooOverUDP support gained the ability to configure
6645 local and peer IP addresses via Local= and Peer=. A new option
6646 PeerPort= may be used to configure the peer's IP port.
6648 * systemd-networkd's TUN support gained a new setting VnetHeader= for
6649 tweaking Generic Segment Offload support.
6651 * The address family for policy rules may be specified using the new
6652 Family= option in the [RoutingPolicyRule] section.
6654 * networkctl gained a new "delete" command for removing virtual network
6655 devices, as well as a new "--stats" switch for showing device
6658 * networkd.conf gained a new setting SpeedMeter= and
6659 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
6660 measured speed may be shown by 'networkctl status'.
6662 * "networkctl status" now displays MTU and queue lengths, and more
6663 detailed information about VXLAN and bridge devices.
6665 * systemd-networkd's .network and .link files gained a new Property=
6666 setting in the [Match] section, to match against devices with
6667 specific udev properties.
6669 * systemd-networkd's tunnel support gained a new option
6670 AssignToLoopback= for selecting whether to use the loopback device
6671 "lo" as underlying device.
6673 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
6674 been renamed to LinkLayerAddress=, and it now allows configuration of
6677 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
6678 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
6679 IPv6 configuration (static or DHCPv6) was found for a given
6680 interface. It will not touch the sysctl otherwise.
6682 * The order of entries is $PATH used by the user manager instance was
6683 changed to put bin/ entries before the corresponding sbin/ entries.
6684 It is recommended to not rely on this order, and only ever have one
6685 binary with a given name in the system paths under /usr.
6687 * A new tool systemd-network-generator has been added that may generate
6688 .network, .netdev and .link files from IP configuration specified on
6689 the kernel command line in the format used by Dracut.
6691 * The CriticalConnection= setting in .network files is now deprecated,
6692 and replaced by a new KeepConfiguration= setting which allows more
6693 detailed configuration of the IP configuration to keep in place.
6695 * systemd-analyze gained a few new verbs:
6697 - "systemd-analyze timestamp" parses and converts timestamps. This is
6698 similar to the existing "systemd-analyze calendar" command which
6699 does the same for recurring calendar events.
6701 - "systemd-analyze timespan" parses and converts timespans (i.e.
6702 durations as opposed to points in time).
6704 - "systemd-analyze condition" will parse and test ConditionXYZ=
6707 - "systemd-analyze exit-status" will parse and convert exit status
6708 codes to their names and back.
6710 - "systemd-analyze unit-files" will print a list of all unit
6711 file paths and unit aliases.
6713 * SuccessExitStatus=, RestartPreventExitStatus=, and
6714 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
6715 is equivalent to "65"). Those exit status name mappings may be
6716 displayed with the systemd-analyze exit-status verb describe above.
6718 * systemd-logind now exposes a per-session SetBrightness() bus call,
6719 which may be used to securely change the brightness of a kernel
6720 brightness device, if it belongs to the session's seat. By using this
6721 call unprivileged clients can make changes to "backlight" and "leds"
6722 devices securely with strict requirements on session membership.
6723 Desktop environments may use this to generically make brightness
6724 changes to such devices without shipping private SUID binaries or
6725 udev rules for that purpose.
6727 * "udevadm info" gained a --wait-for-initialization switch to wait for
6728 a device to be initialized.
6730 * systemd-hibernate-resume-generator will now look for resumeflags= on
6731 the kernel command line, which is similar to rootflags= and may be
6732 used to configure device timeout for the hibernation device.
6734 * sd-event learnt a new API call sd_event_source_disable_unref() for
6735 disabling and unref'ing an event source in a single function. A
6736 related call sd_event_source_disable_unrefp() has been added for use
6737 with gcc's cleanup extension.
6739 * The sd-id128.h public API gained a new definition
6740 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
6743 * "busctl introspect" gained a new switch --xml-interface for dumping
6744 XML introspection data unmodified.
6746 * PID 1 may now show the unit name instead of the unit description
6747 string in its status output during boot. This may be configured in
6748 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
6749 kernel command line option systemd.status_unit_format=.
6751 * PID 1 now understands a new option KExecWatchdogSec= in
6752 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
6753 Previously watchdog functionality was only available for regular
6754 reboots. The new setting defaults to off, because we don't know in
6755 the general case if the watchdog will be reset after kexec (some
6756 drivers do reset it, but not all), and the new userspace might not be
6757 configured to handle the watchdog.
6759 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
6760 RebootWatchdogSec= to more clearly communicate what it is about. The
6761 old name is still accepted for compatibility.
6763 * The systemd.debug_shell kernel command line option now optionally
6764 takes a tty name to spawn the debug shell on, which allows a
6765 different tty to be selected than the built-in default.
6767 * Service units gained a new ExecCondition= setting which will run
6768 before ExecStartPre= and either continue execution of the unit (for
6769 clean exit codes), stop execution without marking the unit failed
6770 (for exit codes 1 through 254), or stop execution and fail the unit
6771 (for exit code 255 or abnormal termination).
6773 * A new service systemd-pstore.service has been added that pulls data
6774 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
6777 * timedatectl gained new verbs for configuring per-interface NTP
6778 service configuration for systemd-timesyncd.
6780 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
6781 2019. (You can set non-UTF-8 locales though, if you know their name.)
6783 * If variable assignments in sysctl.d/ files are prefixed with "-" any
6784 failures to apply them are now ignored.
6786 * systemd-random-seed.service now optionally credits entropy when
6787 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
6788 true for the service to enable this behaviour, but please consult the
6789 documentation first, since this comes with a couple of caveats.
6791 * systemd-random-seed.service is now a synchronization point for full
6792 initialization of the kernel's entropy pool. Services that require
6793 /dev/urandom to be correctly initialized should be ordered after this
6796 * The systemd-boot boot loader has been updated to optionally maintain
6797 a random seed file in the EFI System Partition (ESP). During the boot
6798 phase, this random seed is read and updated with a new seed
6799 cryptographically derived from it. Another derived seed is passed to
6800 the OS. The latter seed is then credited to the kernel's entropy pool
6801 very early during userspace initialization (from PID 1). This allows
6802 systems to boot up with a fully initialized kernel entropy pool from
6803 earliest boot on, and thus entirely removes all entropy pool
6804 initialization delays from systems using systemd-boot. Special care
6805 is taken to ensure different seeds are derived on system images
6806 replicated to multiple systems. "bootctl status" will show whether
6807 a seed was received from the boot loader.
6809 * bootctl gained two new verbs:
6811 - "bootctl random-seed" will generate the file in ESP and an EFI
6812 variable to allow a random seed to be passed to the OS as described
6815 - "bootctl is-installed" checks whether systemd-boot is currently
6818 * bootctl will warn if it detects that boot entries are misconfigured
6819 (for example if the kernel image was removed without purging the
6822 * A new document has been added describing systemd's use and support
6823 for the kernel's entropy pool subsystem:
6825 https://systemd.io/RANDOM_SEEDS
6827 * When the system is hibernated the swap device to write the
6828 hibernation image to is now automatically picked from all available
6829 swap devices, preferring the swap device with the highest configured
6830 priority over all others, and picking the device with the most free
6831 space if there are multiple devices with the highest priority.
6833 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
6834 option that permits selecting the timeout how long to wait for a
6835 device with an encryption key before asking for the password.
6837 * IOWeight= has learnt to properly set the IO weight when using the
6838 BFQ scheduler officially found in kernels 5.0+.
6840 * A new mailing list has been created for reporting of security issues:
6841 systemd-security@redhat.com. For mode details, see
6842 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
6844 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
6845 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
6846 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
6847 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
6848 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
6849 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
6850 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
6851 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
6852 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
6853 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
6854 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
6855 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
6856 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
6857 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
6858 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
6859 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
6860 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
6861 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
6862 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
6863 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
6864 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
6865 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
6866 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
6867 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
6868 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
6869 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
6870 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
6871 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
6872 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
6873 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
6875 – Camerino, 2019-09-03
6879 * In .link files, MACAddressPolicy=persistent (the default) is changed
6880 to cover more devices. For devices like bridges, tun, tap, bond, and
6881 similar interfaces that do not have other identifying information,
6882 the interface name is used as the basis for persistent seed for MAC
6883 and IPv4LL addresses. The way that devices that were handled
6884 previously is not changed, and this change is about covering more
6885 devices then previously by the "persistent" policy.
6887 MACAddressPolicy=random may be used to force randomized MACs and
6888 IPv4LL addresses for a device if desired.
6890 Hint: the log output from udev (at debug level) was enhanced to
6891 clarify what policy is followed and which attributes are used.
6892 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
6893 may be used to view this.
6895 Hint: if a bridge interface is created without any slaves, and gains
6896 a slave later, then now the bridge does not inherit slave's MAC.
6897 To inherit slave's MAC, for example, create the following file:
6899 # /etc/systemd/network/98-bridge-inherit-mac.link
6904 MACAddressPolicy=none
6907 * The .device units generated by systemd-fstab-generator and other
6908 generators do not automatically pull in the corresponding .mount unit
6909 as a Wants= dependency. This means that simply plugging in the device
6910 will not cause the mount unit to be started automatically. But please
6911 note that the mount unit may be started for other reasons, in
6912 particular if it is part of local-fs.target, and any unit which
6913 (transitively) depends on local-fs.target is started.
6915 * networkctl list/status/lldp now accept globbing wildcards for network
6916 interface names to match against all existing interfaces.
6918 * The $PIDFILE environment variable is set to point the absolute path
6919 configured with PIDFile= for processes of that service.
6921 * The fallback DNS server list was augmented with Cloudflare public DNS
6922 servers. Use `-Ddns-servers=` to set a different fallback.
6924 * A new special target usb-gadget.target will be started automatically
6925 when a USB Device Controller is detected (which means that the system
6926 is a USB peripheral).
6928 * A new unit setting CPUQuotaPeriodSec= assigns the time period
6929 relatively to which the CPU time quota specified by CPUQuota= is
6932 * A new unit setting ProtectHostname= may be used to prevent services
6933 from modifying hostname information (even if they otherwise would
6934 have privileges to do so).
6936 * A new unit setting NetworkNamespacePath= may be used to specify a
6937 namespace for service or socket units through a path referring to a
6938 Linux network namespace pseudo-file.
6940 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
6941 have an effect on .socket units: when used the listening socket is
6942 created within the configured network namespace instead of the host
6945 * ExecStart= command lines in unit files may now be prefixed with ':'
6946 in which case environment variable substitution is
6947 disabled. (Supported for the other ExecXYZ= settings, too.)
6949 * .timer units gained two new boolean settings OnClockChange= and
6950 OnTimezoneChange= which may be used to also trigger a unit when the
6951 system clock is changed or the local timezone is
6952 modified. systemd-run has been updated to make these options easily
6953 accessible from the command line for transient timers.
6955 * Two new conditions for units have been added: ConditionMemory= may be
6956 used to conditionalize a unit based on installed system
6957 RAM. ConditionCPUs= may be used to conditionalize a unit based on
6958 installed CPU cores.
6960 * The @default system call filter group understood by SystemCallFilter=
6961 has been updated to include the new rseq() system call introduced in
6964 * A new time-set.target has been added that indicates that the system
6965 time has been set from a local source (possibly imprecise). The
6966 existing time-sync.target is stronger and indicates that the time has
6967 been synchronized with a precise external source. Services where
6968 approximate time is sufficient should use the new target.
6970 * "systemctl start" (and related commands) learnt a new
6971 --show-transaction option. If specified brief information about all
6972 jobs queued because of the requested operation is shown.
6974 * systemd-networkd recognizes a new operation state 'enslaved', used
6975 (instead of 'degraded' or 'carrier') for interfaces which form a
6976 bridge, bond, or similar, and an new 'degraded-carrier' operational
6977 state used for the bond or bridge master interface when one of the
6978 enslaved devices is not operational.
6980 * .network files learnt the new IgnoreCarrierLoss= option for leaving
6981 networks configured even if the carrier is lost.
6983 * The RequiredForOnline= setting in .network files may now specify a
6984 minimum operational state required for the interface to be considered
6985 "online" by systemd-networkd-wait-online. Related to this
6986 systemd-networkd-wait-online gained a new option --operational-state=
6987 to configure the same, and its --interface= option was updated to
6988 optionally also take an operational state specific for an interface.
6990 * systemd-networkd-wait-online gained a new setting --any for waiting
6991 for only one of the requested interfaces instead of all of them.
6993 * systemd-networkd now implements L2TP tunnels.
6995 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
6996 may be used to cause autonomous and onlink prefixes received in IPv6
6997 Router Advertisements to be ignored.
6999 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
7000 file settings may be used to tweak bridge behaviour.
7002 * The new TripleSampling= option in .network files may be used to
7003 configure CAN triple sampling.
7005 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
7006 used to point to private or preshared key for a WireGuard interface.
7008 * /etc/crypttab now supports the same-cpu-crypt and
7009 submit-from-crypt-cpus options to tweak encryption work scheduling
7012 * systemd-tmpfiles will now take a BSD file lock before operating on a
7013 contents of directory. This may be used to temporarily exclude
7014 directories from aging by taking the same lock (useful for example
7015 when extracting a tarball into /tmp or /var/tmp as a privileged user,
7016 which might create files with really old timestamps, which
7017 nevertheless should not be deleted). For further details, see:
7019 https://systemd.io/TEMPORARY_DIRECTORIES
7021 * systemd-tmpfiles' h line type gained support for the
7022 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
7023 controlling project quota inheritance.
7025 * sd-boot and bootctl now implement support for an Extended Boot Loader
7026 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
7027 addition to the ESP partition mounted to /efi or /boot/efi.
7028 Configuration file fragments, kernels, initrds and other EFI images
7029 to boot will be loaded from both the ESP and XBOOTLDR partitions.
7030 The XBOOTLDR partition was previously described by the Boot Loader
7031 Specification, but implementation was missing in sd-boot. Support for
7032 this concept allows using the sd-boot boot loader in more
7033 conservative scenarios where the boot loader itself is placed in the
7034 ESP but the kernels to boot (and their metadata) in a separate
7037 * A system may now be booted with systemd.volatile=overlay on the
7038 kernel command line, which causes the root file system to be set up
7039 an overlayfs mount combining the root-only root directory with a
7040 writable tmpfs. In this setup, the underlying root device is not
7041 modified, and any changes are lost at reboot.
7043 * Similar, systemd-nspawn can now boot containers with a volatile
7044 overlayfs root with the new --volatile=overlay switch.
7046 * systemd-nspawn can now consume OCI runtime bundles using a new
7047 --oci-bundle= option. This implementation is fully usable, with most
7048 features in the specification implemented, but since this a lot of
7049 new code and functionality, this feature should most likely not
7050 be used in production yet.
7052 * systemd-nspawn now supports various options described by the OCI
7053 runtime specification on the command-line and in .nspawn files:
7054 --inaccessible=/Inaccessible= may be used to mask parts of the file
7055 system tree, --console=/--pipe may be used to configure how standard
7056 input, output, and error are set up.
7058 * busctl learned the `emit` verb to generate D-Bus signals.
7060 * systemd-analyze cat-config may be used to gather and display
7061 configuration spread over multiple files, for example system and user
7062 presets, tmpfiles.d, sysusers.d, udev rules, etc.
7064 * systemd-analyze calendar now takes an optional new parameter
7065 --iterations= which may be used to show a maximum number of iterations
7066 the specified expression will elapse next.
7068 * The sd-bus C API gained support for naming method parameters in the
7071 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
7072 the reboot() system call expects.
7074 * journalctl learnt a new --cursor-file= option that points to a file
7075 from which a cursor should be loaded in the beginning and to which
7076 the updated cursor should be stored at the end.
7078 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
7079 detected by systemd-detect-virt (and may also be used in
7080 ConditionVirtualization=).
7082 * The behaviour of systemd-logind may now be modified with environment
7083 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
7084 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
7085 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
7086 skip the relevant operation completely (when set to false), or to
7087 create a flag file in /run/systemd (when set to true), instead of
7088 actually commencing the real operation when requested. The presence
7089 of /run/systemd/reboot-to-firmware-setup,
7090 /run/systemd/reboot-to-boot-loader-menu, and
7091 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
7092 boot loader implementations to replace some steps logind performs
7093 during reboot with their own operations.
7095 * systemctl can be used to request a reboot into the boot loader menu
7096 or a specific boot loader entry with the new --boot-load-menu= and
7097 --boot-loader-entry= options to a reboot command. (This requires a
7098 boot loader that supports this, for example sd-boot.)
7100 * kernel-install will no longer unconditionally create the output
7101 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
7102 snippets, but will do only if the machine-specific parent directory
7103 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
7104 to create this parent directory during sd-boot installation.
7106 This makes it easier to use kernel-install with plugins which support
7107 a different layout of the bootloader partitions (for example grub2).
7109 * During package installation (with `ninja install`), we would create
7110 symlinks for getty@tty1.service, systemd-networkd.service,
7111 systemd-networkd.socket, systemd-resolved.service,
7112 remote-cryptsetup.target, remote-fs.target,
7113 systemd-networkd-wait-online.service, and systemd-timesyncd.service
7114 in /etc, as if `systemctl enable` was called for those units, to make
7115 the system usable immediately after installation. Now this is not
7116 done anymore, and instead calling `systemctl preset-all` is
7117 recommended after the first installation of systemd.
7119 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
7120 is built on seccomp. When turned on creation of SUID/SGID files is
7123 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
7124 implied if DynamicUser= is turned on for a service. This hardens
7125 these services, so that they neither can benefit from nor create
7126 SUID/SGID executables. This is a minor compatibility breakage, given
7127 that when DynamicUser= was first introduced SUID/SGID behaviour was
7128 unaffected. However, the security benefit of these two options is
7129 substantial, and the setting is still relatively new, hence we opted
7130 to make it mandatory for services with dynamic users.
7132 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
7133 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
7134 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
7135 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
7136 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
7137 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
7138 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
7139 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
7140 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
7141 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
7142 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
7143 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
7144 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
7145 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
7146 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
7147 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
7148 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
7149 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
7151 — Warsaw, 2019-04-11
7155 * The default locale can now be configured at compile time. Otherwise,
7156 a suitable default will be selected automatically (one of C.UTF-8,
7157 en_US.UTF-8, and C).
7159 * The version string shown by systemd and other tools now includes the
7160 git commit hash when built from git. An override may be specified
7161 during compilation, which is intended to be used by distributions to
7162 include the package release information.
7164 * systemd-cat can now filter standard input and standard error streams
7165 for different syslog priorities using the new --stderr-priority=
7168 * systemd-journald and systemd-journal-remote reject entries which
7169 contain too many fields (CVE-2018-16865) and set limits on the
7170 process' command line length (CVE-2018-16864).
7172 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
7175 * A new network device NamePolicy "keep" is implemented for link files,
7176 and used by default in 99-default.link (the fallback configuration
7177 provided by systemd). With this policy, if the network device name
7178 was already set by userspace, the device will not be renamed again.
7179 This matches the naming scheme that was implemented before
7180 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
7181 is also enabled by default, even if not specified. Effectively, this
7182 means that if naming-scheme >= 240 is specified, network devices will
7183 be renamed according to the configuration, even if they have been
7184 renamed already, if "keep" is not specified as the naming policy in
7185 the .link file. The 99-default.link file provided by systemd includes
7186 "keep" for backwards compatibility, but it is recommended for user
7187 installed .link files to *not* include it.
7189 The "kernel" policy, which keeps kernel names declared to be
7190 "persistent", now works again as documented.
7192 * kernel-install script now optionally takes the paths to one or more
7193 initrd files, and passes them to all plugins.
7195 * The mincore() system call has been dropped from the @system-service
7196 system call filter group, as it is pretty exotic and may potentially
7197 used for side-channel attacks.
7199 * -fPIE is dropped from compiler and linker options. Please specify
7200 -Db_pie=true option to meson to build position-independent
7201 executables. Note that the meson option is supported since meson-0.49.
7203 * The fs.protected_regular and fs.protected_fifos sysctls, which were
7204 added in Linux 4.19 to make some data spoofing attacks harder, are
7205 now enabled by default. While this will hopefully improve the
7206 security of most installations, it is technically a backwards
7207 incompatible change; to disable these sysctls again, place the
7208 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
7210 fs.protected_regular = 0
7211 fs.protected_fifos = 0
7213 Note that the similar hardlink and symlink protection has been
7214 enabled since v199, and may be disabled likewise.
7216 * The files read from the EnvironmentFile= setting in unit files now
7217 parse backslashes inside quotes literally, matching the behaviour of
7220 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
7221 now automatically become NOPs when run in a chroot() environment.
7223 * The tmpfiles.d/ "C" line type will now copy directory trees not only
7224 when the destination is so far missing, but also if it already exists
7225 as a directory and is empty. This is useful to cater for systems
7226 where directory trees are put together from multiple separate mount
7227 points but otherwise empty.
7229 * A new function sd_bus_close_unref() (and the associated
7230 sd_bus_close_unrefp()) has been added to libsystemd, that combines
7231 sd_bus_close() and sd_bus_unref() in one.
7233 * udevadm control learnt a new option for --ping for testing whether a
7234 systemd-udevd instance is running and reacting.
7236 * udevadm trigger learnt a new option for --wait-daemon for waiting
7237 systemd-udevd daemon to be initialized.
7239 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
7240 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
7241 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
7242 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
7243 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
7244 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
7245 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
7246 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
7247 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
7248 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
7249 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
7250 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
7251 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
7252 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
7253 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
7254 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
7255 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
7257 — Berlin, 2019-02-14
7261 * NoNewPrivileges=yes has been set for all long-running services
7262 implemented by systemd. Previously, this was problematic due to
7263 SELinux (as this would also prohibit the transition from PID1's label
7264 to the service's label). This restriction has since been lifted, but
7265 an SELinux policy update is required.
7266 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
7268 * DynamicUser=yes is dropped from systemd-networkd.service,
7269 systemd-resolved.service and systemd-timesyncd.service, which was
7270 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
7271 and since v236 for systemd-timesyncd.service. The users and groups
7272 systemd-network, systemd-resolve and systemd-timesync are created
7273 by systemd-sysusers again. Distributors or system administrators
7274 may need to create these users and groups if they not exist (or need
7275 to re-enable DynamicUser= for those units) while upgrading systemd.
7276 Also, the clock file for systemd-timesyncd may need to move from
7277 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
7279 * When unit files are loaded from disk, previously systemd would
7280 sometimes (depending on the unit loading order) load units from the
7281 target path of symlinks in .wants/ or .requires/ directories of other
7282 units. This meant that unit could be loaded from different paths
7283 depending on whether the unit was requested explicitly or as a
7284 dependency of another unit, not honouring the priority of directories
7285 in search path. It also meant that it was possible to successfully
7286 load and start units which are not found in the unit search path, as
7287 long as they were requested as a dependency and linked to from
7288 .wants/ or .requires/. The target paths of those symlinks are not
7289 used for loading units anymore and the unit file must be found in
7292 * A new service type has been added: Type=exec. It's very similar to
7293 Type=simple but ensures the service manager will wait for both fork()
7294 and execve() of the main service binary to complete before proceeding
7295 with follow-up units. This is primarily useful so that the manager
7296 propagates any errors in the preparation phase of service execution
7297 back to the job that requested the unit to be started. For example,
7298 consider a service that has ExecStart= set to a file system binary
7299 that doesn't exist. With Type=simple starting the unit would be
7300 considered instantly successful, as only fork() has to complete
7301 successfully and the manager does not wait for execve(), and hence
7302 its failure is seen "too late". With the new Type=exec service type
7303 starting the unit will fail, as the manager will wait for the
7304 execve() and notice its failure, which is then propagated back to the
7307 NOTE: with the next release 241 of systemd we intend to change the
7308 systemd-run tool to default to Type=exec for transient services
7309 started by it. This should be mostly safe, but in specific corner
7310 cases might result in problems, as the systemd-run tool will then
7311 block on NSS calls (such as user name look-ups due to User=) done
7312 between the fork() and execve(), which under specific circumstances
7313 might cause problems. It is recommended to specify "-p Type=simple"
7314 explicitly in the few cases where this applies. For regular,
7315 non-transient services (i.e. those defined with unit files on disk)
7316 we will continue to default to Type=simple.
7318 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
7319 userspace processes is set to 1024 (soft) and 4096
7320 (hard). Previously, systemd passed this on unmodified to all
7321 processes it forked off. With this systemd release the hard limit
7322 systemd passes on is increased to 512K, overriding the kernel's
7323 defaults and substantially increasing the number of simultaneous file
7324 descriptors unprivileged userspace processes can allocate. Note that
7325 the soft limit remains at 1024 for compatibility reasons: the
7326 traditional UNIX select() call cannot deal with file descriptors >=
7327 1024 and increasing the soft limit globally might thus result in
7328 programs unexpectedly allocating a high file descriptor and thus
7329 failing abnormally when attempting to use it with select() (of
7330 course, programs shouldn't use select() anymore, and prefer
7331 poll()/epoll, but the call unfortunately remains undeservedly popular
7332 at this time). This change reflects the fact that file descriptor
7333 handling in the Linux kernel has been optimized in more recent
7334 kernels and allocating large numbers of them should be much cheaper
7335 both in memory and in performance than it used to be. Programs that
7336 want to take benefit of the increased limit have to "opt-in" into
7337 high file descriptors explicitly by raising their soft limit. Of
7338 course, when they do that they must acknowledge that they cannot use
7339 select() anymore (and neither can any shared library they use — or
7340 any shared library used by any shared library they use and so on).
7341 Which default hard limit is most appropriate is of course hard to
7342 decide. However, given reports that ~300K file descriptors are used
7343 in real-life applications we believe 512K is sufficiently high as new
7344 default for now. Note that there are also reports that using very
7345 high hard limits (e.g. 1G) is problematic: some software allocates
7346 large arrays with one element for each potential file descriptor
7347 (Java, …) — a high hard limit thus triggers excessively large memory
7348 allocations in these applications. Hopefully, the new default of 512K
7349 is a good middle ground: higher than what real-life applications
7350 currently need, and low enough for avoid triggering excessively large
7351 allocations in problematic software. (And yes, somebody should fix
7354 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
7355 to the highest possible values, as separate accounting of file
7356 descriptors is no longer necessary, as memcg tracks them correctly as
7357 part of the memory accounting anyway. Thus, from the four limits on
7358 file descriptors currently enforced (fs.file-max, fs.nr_open,
7359 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
7360 and keep only the latter two. A set of build-time options
7361 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
7362 has been added to revert this change in behaviour, which might be
7363 an option for systems that turn off memcg in the kernel.
7365 * When no /etc/locale.conf file exists (and hence no locale settings
7366 are in place), systemd will now use the "C.UTF-8" locale by default,
7367 and set LANG= to it. This locale is supported by various
7368 distributions including Fedora, with clear indications that upstream
7369 glibc is going to make it available too. This locale enables UTF-8
7370 mode by default, which appears appropriate for 2018.
7372 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
7373 default. This effectively switches the RFC3704 Reverse Path filtering
7374 from Strict mode to Loose mode. This is more appropriate for hosts
7375 that have multiple links with routes to the same networks (e.g.
7376 a client with a Wi-Fi and Ethernet both connected to the internet).
7378 Consult the kernel documentation for details on this sysctl:
7379 https://docs.kernel.org/networking/ip-sysctl.html
7381 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
7384 * CPUAccounting=yes no longer enables the CPU controller when using
7385 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
7386 statistics are now provided independently from the CPU controller.
7388 * Support for disabling a particular cgroup controller within a sub-tree
7389 has been added through the DisableControllers= directive.
7391 * cgroup_no_v1=all on the kernel command line now also implies
7392 using the unified cgroup hierarchy, unless one explicitly passes
7393 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
7395 * The new "MemoryMin=" unit file property may now be used to set the
7396 memory usage protection limit of processes invoked by the unit. This
7397 controls the cgroup v2 memory.min attribute. Similarly, the new
7398 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
7399 cgroup v2 io.latency cgroup property for configuring per-service I/O
7402 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
7403 to the cgroup v1 "devices" cgroup controller.
7405 * systemd-escape now is able to combine --unescape with --template. It
7406 also learnt a new option --instance for extracting and unescaping the
7407 instance part of a unit name.
7409 * sd-bus now provides the sd_bus_message_readv() which is similar to
7410 sd_bus_message_read() but takes a va_list object. The pair
7411 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
7412 has been added for configuring the default method call timeout to
7413 use. sd_bus_error_move() may be used to efficiently move the contents
7414 from one sd_bus_error structure to another, invalidating the
7415 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
7416 be used to control whether a bus connection object is automatically
7417 flushed when an sd-event loop is exited.
7419 * When processing classic BSD syslog log messages, journald will now
7420 save the original time-stamp string supplied in the new
7421 SYSLOG_TIMESTAMP= journal field. This permits consumers to
7422 reconstruct the original BSD syslog message more correctly.
7424 * StandardOutput=/StandardError= in service files gained support for
7425 new "append:…" parameters, for connecting STDOUT/STDERR of a service
7426 to a file, and appending to it.
7428 * The signal to use as last step of killing of unit processes is now
7429 configurable. Previously it was hard-coded to SIGKILL, which may now
7430 be overridden with the new KillSignal= setting. Note that this is the
7431 signal used when regular termination (i.e. SIGTERM) does not suffice.
7432 Similarly, the signal used when aborting a program in case of a
7433 watchdog timeout may now be configured too (WatchdogSignal=).
7435 * The XDG_SESSION_DESKTOP environment variable may now be configured in
7436 the pam_systemd argument line, using the new desktop= switch. This is
7437 useful to initialize it properly from a display manager without
7438 having to touch C code.
7440 * Most configuration options that previously accepted percentage values
7441 now also accept permille values with the '‰' suffix (instead of '%').
7443 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
7446 * systemd-resolved's configuration file resolved.conf gained a new
7447 option ReadEtcHosts= which may be used to turn off processing and
7448 honoring /etc/hosts entries.
7450 * The "--wait" switch may now be passed to "systemctl
7451 is-system-running", in which case the tool will synchronously wait
7452 until the system finished start-up.
7454 * hostnamed gained a new bus call to determine the DMI product UUID.
7456 * On x86-64 systemd will now prefer using the RDRAND processor
7457 instruction over /dev/urandom whenever it requires randomness that
7458 neither has to be crypto-grade nor should be reproducible. This
7459 should substantially reduce the amount of entropy systemd requests
7460 from the kernel during initialization on such systems, though not
7461 reduce it to zero. (Why not zero? systemd still needs to allocate
7462 UUIDs and such uniquely, which require high-quality randomness.)
7464 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
7465 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
7466 for forcing the "Other Information" bit in IPv6 RA messages. The
7467 bonding logic gained four new options AdActorSystemPriority=,
7468 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
7469 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
7470 shuffling of flows. The tunnel logic gained a new
7471 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
7472 Deployment. The policy rule logic gained four new options IPProtocol=,
7473 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
7474 support for the MulticastToUnicast= option. networkd also gained
7475 support for configuring static IPv4 ARP or IPv6 neighbor entries.
7477 * .preset files (as read by 'systemctl preset') may now be used to
7478 instantiate services.
7480 * /etc/crypttab now understands the sector-size= option to configure
7481 the sector size for an encrypted partition.
7483 * Key material for encrypted disks may now be placed on a formatted
7484 medium, and referenced from /etc/crypttab by the UUID of the file
7485 system, followed by "=" suffixed by the path to the key file.
7487 * The "collect" udev component has been removed without replacement, as
7488 it is neither used nor maintained.
7490 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
7491 LogsDirectory=, ConfigurationDirectory= settings are used in a
7492 service the executed processes will now receive a set of environment
7493 variables containing the full paths of these directories.
7494 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
7495 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
7496 are used. Note that these options may be used multiple times per
7497 service in which case the resulting paths will be concatenated and
7498 separated by colons.
7500 * Predictable interface naming has been extended to cover InfiniBand
7501 NICs. They will be exposed with an "ib" prefix.
7503 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
7504 which case the respective line failing is ignored.
7506 * .link files may now be used to configure the equivalent to the
7507 "ethtool advertise" commands.
7509 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
7510 alternative to libudev.h. Previously, the latter was just an internal
7511 wrapper around the former, but now these two APIs are exposed
7514 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
7515 which calculates an app-specific boot ID similar to how
7516 sd_id128_get_machine_app_specific() generates an app-specific machine
7519 * A new tool systemd-id128 has been added that can be used to determine
7520 and generate various 128-bit IDs.
7522 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
7525 * systemd-hibernate-resume-generator will now honor the "noresume"
7526 kernel command line option, in which case it will bypass resuming
7527 from any hibernated image.
7529 * The systemd-sleep.conf configuration file gained new options
7530 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
7531 AllowHybridSleep= for prohibiting specific sleep modes even if the
7532 kernel exports them.
7534 * portablectl is now officially supported and has thus moved to
7537 * bootctl learnt the two new commands "set-default" and "set-oneshot"
7538 for setting the default boot loader item to boot to (either
7539 persistently or only for the next boot). This is currently only
7540 compatible with sd-boot, but may be implemented on other boot loaders
7541 too, that follow the boot loader interface. The updated interface is
7542 now documented here:
7544 https://systemd.io/BOOT_LOADER_INTERFACE
7546 * A new kernel command line option systemd.early_core_pattern= is now
7547 understood which may be used to influence the core_pattern PID 1
7548 installs during early boot.
7550 * busctl learnt two new options -j and --json= for outputting method
7551 call replies, properties and monitoring output in JSON.
7553 * journalctl's JSON output now supports simple ANSI coloring as well as
7554 a new "json-seq" mode for generating RFC7464 output.
7556 * Unit files now support the %g/%G specifiers that resolve to the UNIX
7557 group/GID of the service manager runs as, similar to the existing
7558 %u/%U specifiers that resolve to the UNIX user/UID.
7560 * systemd-logind learnt a new global configuration option
7561 UserStopDelaySec= that may be set in logind.conf. It specifies how
7562 long the systemd --user instance shall remain started after a user
7563 logs out. This is useful to speed up repetitive re-connections of the
7564 same user, as it means the user's service manager doesn't have to be
7565 stopped/restarted on each iteration, but can be reused between
7566 subsequent options. This setting defaults to 10s. systemd-logind also
7567 exports two new properties on its Manager D-Bus objects indicating
7568 whether the system's lid is currently closed, and whether the system
7571 * systemd gained support for a generic boot counting logic, which
7572 generically permits automatic reverting to older boot loader entries
7573 if newer updated ones don't work. The boot loader side is implemented
7574 in sd-boot, but is kept open for other boot loaders too. For details
7577 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
7579 * The SuccessAction=/FailureAction= unit file settings now learnt two
7580 new parameters: "exit" and "exit-force", which result in immediate
7581 exiting of the service manager, and are only useful in systemd --user
7582 and container environments.
7584 * Unit files gained support for a pair of options
7585 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
7586 exit status to use as service manager exit status when
7587 SuccessAction=/FailureAction= is set to exit or exit-force.
7589 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
7590 options may now be used to configure the log rate limiting applied by
7591 journald per-service.
7593 * systemd-analyze gained a new verb "timespan" for parsing and
7594 normalizing time span values (i.e. strings like "5min 7s 8us").
7596 * systemd-analyze also gained a new verb "security" for analyzing the
7597 security and sand-boxing settings of services in order to determine an
7598 "exposure level" for them, indicating whether a service would benefit
7599 from more sand-boxing options turned on for them.
7601 * "systemd-analyze syscall-filter" will now also show system calls
7602 supported by the local kernel but not included in any of the defined
7605 * .nspawn files now understand the Ephemeral= setting, matching the
7606 --ephemeral command line switch.
7608 * sd-event gained the new APIs sd_event_source_get_floating() and
7609 sd_event_source_set_floating() for controlling whether a specific
7610 event source is "floating", i.e. destroyed along with the even loop
7613 * Unit objects on D-Bus gained a new "Refs" property that lists all
7614 clients that currently have a reference on the unit (to ensure it is
7617 * The JoinControllers= option in system.conf is no longer supported, as
7618 it didn't work correctly, is hard to support properly, is legacy (as
7619 the concept only exists on cgroup v1) and apparently wasn't used.
7621 * Journal messages that are generated whenever a unit enters the failed
7622 state are now tagged with a unique MESSAGE_ID. Similarly, messages
7623 generated whenever a service process exits are now made recognizable,
7624 too. A tagged message is also emitted whenever a unit enters the
7625 "dead" state on success.
7627 * systemd-run gained a new switch --working-directory= for configuring
7628 the working directory of the service to start. A shortcut -d is
7629 equivalent, setting the working directory of the service to the
7630 current working directory of the invoking program. The new --shell
7631 (or just -S) option has been added for invoking the $SHELL of the
7632 caller as a service, and implies --pty --same-dir --wait --collect
7633 --service-type=exec. Or in other words, "systemd-run -S" is now the
7634 quickest way to quickly get an interactive in a fully clean and
7635 well-defined system service context.
7637 * machinectl gained a new verb "import-fs" for importing an OS tree
7638 from a directory. Moreover, when a directory or tarball is imported
7639 and single top-level directory found with the OS itself below the OS
7640 tree is automatically mangled and moved one level up.
7642 * systemd-importd will no longer set up an implicit btrfs loop-back
7643 file system on /var/lib/machines. If one is already set up, it will
7644 continue to be used.
7646 * A new generator "systemd-run-generator" has been added. It will
7647 synthesize a unit from one or more program command lines included in
7648 the kernel command line. This is very useful in container managers
7651 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
7653 This will run "systemd-nspawn" on an image, invoke the specified
7654 command line and immediately shut down the container again, returning
7655 the command line's exit code.
7657 * The block device locking logic is now documented:
7659 https://systemd.io/BLOCK_DEVICE_LOCKING
7661 * loginctl and machinectl now optionally output the various tables in
7662 JSON using the --output= switch. It is our intention to add similar
7663 support to systemctl and all other commands.
7665 * udevadm's query and trigger verb now optionally take a .device unit
7668 * systemd-udevd's network naming logic now understands a new
7669 net.naming-scheme= kernel command line switch, which may be used to
7670 pick a specific version of the naming scheme. This helps stabilizing
7671 interface names even as systemd/udev are updated and the naming logic
7674 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
7675 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
7676 initialize one to all 0xFF.
7678 * After loading the SELinux policy systemd will now recursively relabel
7679 all files and directories listed in
7680 /run/systemd/relabel-extra.d/*.relabel (which should be simple
7681 newline separated lists of paths) in addition to the ones it already
7682 implicitly relabels in /run, /dev and /sys. After the relabelling is
7683 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
7684 removed. This is useful to permit initrds (i.e. code running before
7685 the SELinux policy is in effect) to generate files in the host
7686 filesystem safely and ensure that the correct label is applied during
7687 the transition to the host OS.
7689 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
7690 mknod() handling in user namespaces. Previously mknod() would always
7691 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
7692 but device nodes generated that way cannot be opened, and attempts to
7693 open them result in EPERM. This breaks the "graceful fallback" logic
7694 in systemd's PrivateDevices= sand-boxing option. This option is
7695 implemented defensively, so that when systemd detects it runs in a
7696 restricted environment (such as a user namespace, or an environment
7697 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
7698 where device nodes cannot be created the effect of PrivateDevices= is
7699 bypassed (following the logic that 2nd-level sand-boxing is not
7700 essential if the system systemd runs in is itself already sand-boxed
7701 as a whole). This logic breaks with 4.18 in container managers where
7702 user namespacing is used: suddenly PrivateDevices= succeeds setting
7703 up a private /dev/ file system containing devices nodes — but when
7704 these are opened they don't work.
7706 At this point it is recommended that container managers utilizing
7707 user namespaces that intend to run systemd in the payload explicitly
7708 block mknod() with seccomp or similar, so that the graceful fallback
7711 We are very sorry for the breakage and the requirement to change
7712 container configurations for newer kernels. It's purely caused by an
7713 incompatible kernel change. The relevant kernel developers have been
7714 notified about this userspace breakage quickly, but they chose to
7717 * PermissionsStartOnly= setting is deprecated (but is still supported
7718 for backwards compatibility). The same functionality is provided by
7719 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
7722 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
7723 pam_systemd anymore.
7725 * The naming scheme for network devices was changed to always rename
7726 devices, even if they were already renamed by userspace. The "kernel"
7727 policy was changed to only apply as a fallback, if no other naming
7730 * The requirements to build systemd is bumped to meson-0.46 and
7733 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
7734 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
7735 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
7736 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
7737 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
7738 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
7739 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
7740 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
7741 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
7742 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
7743 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
7744 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
7745 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
7746 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
7747 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
7748 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
7749 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
7750 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
7751 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
7752 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
7753 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
7754 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
7755 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
7756 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
7757 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
7758 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
7759 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
7760 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
7761 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
7762 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
7763 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
7764 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
7765 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
7766 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
7767 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
7768 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
7769 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
7770 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
7771 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
7772 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
7773 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
7774 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
7775 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
7776 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
7777 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
7779 — Warsaw, 2018-12-21
7783 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
7784 builtin will name network interfaces differently than in previous
7785 versions for virtual network interfaces created with SR-IOV and NPAR
7786 and for devices where the PCI network controller device does not have
7787 a slot number associated.
7789 SR-IOV virtual devices are now named based on the name of the parent
7790 interface, with a suffix of "v<N>", where <N> is the virtual device
7791 number. Previously those virtual devices were named as if completely
7794 The ninth and later NPAR virtual devices will be named following the
7795 scheme used for the first eight NPAR partitions. Previously those
7796 devices were not renamed and the kernel default (eth<n>) was used.
7798 "net_id" will also generate names for PCI devices where the PCI
7799 network controller device does not have an associated slot number
7800 itself, but one of its parents does. Previously those devices were
7801 not renamed and the kernel default (eth<n>) was used.
7803 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
7804 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
7805 the unit. So, it is expected that the default behavior of
7806 systemd-logind is not changed. However, if distribution packagers or
7807 administrators disabled or modified IPAddressDeny= setting by a
7808 drop-in config file, then it may be necessary to update the file to
7809 re-enable AF_INET and AF_INET6 to support network user name services,
7812 * When the RestrictNamespaces= unit property is specified multiple
7813 times, then the specified types are merged now. Previously, only the
7814 last assignment was used. So, if distribution packagers or
7815 administrators modified the setting by a drop-in config file, then it
7816 may be necessary to update the file.
7818 * When OnFailure= is used in combination with Restart= on a service
7819 unit, then the specified units will no longer be triggered on
7820 failures that result in restarting. Previously, the specified units
7821 would be activated each time the unit failed, even when the unit was
7822 going to be restarted automatically. This behaviour contradicted the
7823 documentation. With this release the code is adjusted to match the
7826 * systemd-tmpfiles will now print a notice whenever it encounters
7827 tmpfiles.d/ lines referencing the /var/run/ directory. It will
7828 recommend reworking them to use the /run/ directory instead (for
7829 which /var/run/ is simply a symlinked compatibility alias). This way
7830 systemd-tmpfiles can properly detect line conflicts and merge lines
7831 referencing the same file by two paths, without having to access
7834 * systemctl disable/unmask/preset/preset-all cannot be used with
7835 --runtime. Previously this was allowed, but resulted in unintuitive
7836 behaviour that wasn't useful. systemctl disable/unmask will now undo
7837 both runtime and persistent enablement/masking, i.e. it will remove
7838 any relevant symlinks both in /run and /etc.
7840 * Note that all long-running system services shipped with systemd will
7841 now default to a system call allow list (rather than a deny list, as
7842 before). In particular, systemd-udevd will now enforce one too. For
7843 most cases this should be safe, however downstream distributions
7844 which disabled sandboxing of systemd-udevd (specifically the
7845 MountFlags= setting), might want to disable this security feature
7846 too, as the default allow-listing will prohibit all mount, swap,
7847 reboot and clock changing operations from udev rules.
7849 * sd-boot acquired new loader configuration settings to optionally turn
7850 off Windows and MacOS boot partition discovery as well as
7851 reboot-into-firmware menu items. It is also able to pick a better
7852 screen resolution for HiDPI systems, and now provides loader
7853 configuration settings to change the resolution explicitly.
7855 * systemd-resolved now supports DNS-over-TLS. It's still
7856 turned off by default, use DNSOverTLS=opportunistic to turn it on in
7857 resolved.conf. We intend to make this the default as soon as couple
7858 of additional techniques for optimizing the initial latency caused by
7859 establishing a TLS/TCP connection are implemented.
7861 * systemd-resolved.service and systemd-networkd.service now set
7862 DynamicUser=yes. The users systemd-resolve and systemd-network are
7863 not created by systemd-sysusers anymore.
7865 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
7866 that embed a network facing module into any process using getpwuid()
7867 or related call: the dynamic allocation of the user ID for
7868 systemd-resolved.service means the service manager has to check NSS
7869 if the user name is already taken when forking off the service. Since
7870 the user in the common case won't be defined in /etc/passwd the
7871 lookup is likely to trigger nss-ldap which in turn might use NSS to
7872 ask systemd-resolved for hostname lookups. This will hence result in
7873 a deadlock: a user name lookup in order to start
7874 systemd-resolved.service will result in a hostname lookup for which
7875 systemd-resolved.service needs to be started already. There are
7876 multiple ways to work around this problem: pre-allocate the
7877 "systemd-resolve" user on such systems, so that nss-ldap won't be
7878 triggered; or use a different NSS package that doesn't do networking
7879 in-process but provides a local asynchronous name cache; or configure
7880 the NSS package to avoid lookups for UIDs in the range `pkg-config
7881 systemd --variable=dynamicuidmin` … `pkg-config systemd
7882 --variable=dynamicuidmax`, so that it does not consider itself
7883 authoritative for the same UID range systemd allocates dynamic users
7886 * The systemd-resolve tool has been renamed to resolvectl (it also
7887 remains available under the old name, for compatibility), and its
7888 interface is now verb-based, similar in style to the other <xyz>ctl
7889 tools, such as systemctl or loginctl.
7891 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
7892 compatibility. It may be symlinked under the 'resolvconf' name, in
7893 which case it will take arguments and input compatible with the
7894 Debian and FreeBSD resolvconf tool.
7896 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
7897 where the system initially suspends, and after a timeout resumes and
7900 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
7901 set the client will only send a DUID as client identifier.
7903 * The nss-systemd glibc NSS module will now enumerate dynamic users and
7904 groups in effect. Previously, it could resolve UIDs/GIDs to user
7905 names/groups and vice versa, but did not support enumeration.
7907 * journald's Compress= configuration setting now optionally accepts a
7908 byte threshold value. All journal objects larger than this threshold
7909 will be compressed, smaller ones will not. Previously this threshold
7910 was not configurable and set to 512.
7912 * A new system.conf setting NoNewPrivileges= is now available which may
7913 be used to turn off acquisition of new privileges system-wide
7914 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
7915 for all its children). Note that turning this option on means setuid
7916 binaries and file system capabilities lose their special powers.
7917 While turning on this option is a big step towards a more secure
7918 system, doing so is likely to break numerous pre-existing UNIX tools,
7919 in particular su and sudo.
7921 * A new service systemd-time-sync-wait.service has been added. If
7922 enabled it will delay the time-sync.target unit at boot until time
7923 synchronization has been received from the network. This
7924 functionality is useful on systems lacking a local RTC or where it is
7925 acceptable that the boot process shall be delayed by external network
7928 * When hibernating, systemd will now inform the kernel of the image
7929 write offset, on kernels new enough to support this. This means swap
7930 files should work for hibernation now.
7932 * When loading unit files, systemd will now look for drop-in unit files
7933 extensions in additional places. Previously, for a unit file name
7934 "foo-bar-baz.service" it would look for dropin files in
7935 "foo-bar-baz.service.d/*.conf". Now, it will also look in
7936 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
7937 service name truncated after all inner dashes. This scheme allows
7938 writing drop-ins easily that apply to a whole set of unit files at
7939 once. It's particularly useful for mount and slice units (as their
7940 naming is prefix based), but is also useful for service and other
7941 units, for packages that install multiple unit files at once,
7942 following a strict naming regime of beginning the unit file name with
7943 the package's name. Two new specifiers are now supported in unit
7944 files to match this: %j and %J are replaced by the part of the unit
7945 name following the last dash.
7947 * Unit files and other configuration files that support specifier
7948 expansion now understand another three new specifiers: %T and %V will
7949 resolve to /tmp and /var/tmp respectively, or whatever temporary
7950 directory has been set for the calling user. %E will expand to either
7951 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
7953 * The ExecStart= lines of unit files are no longer required to
7954 reference absolute paths. If non-absolute paths are specified the
7955 specified binary name is searched within the service manager's
7956 built-in $PATH, which may be queried with 'systemd-path
7957 search-binaries-default'. It's generally recommended to continue to
7958 use absolute paths for all binaries specified in unit files.
7960 * Units gained a new load state "bad-setting", which is used when a
7961 unit file was loaded, but contained fatal errors which prevent it
7962 from being started (for example, a service unit has been defined
7963 lacking both ExecStart= and ExecStop= lines).
7965 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
7966 support alternative debuggers, for example lldb. The old name
7967 continues to be available however, for compatibility reasons. Use the
7968 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
7969 to pick an alternative debugger instead of the default gdb.
7971 * systemctl and the other tools will now output escape sequences that
7972 generate proper clickable hyperlinks in various terminal emulators
7973 where useful (for example, in the "systemctl status" output you can
7974 now click on the unit file name to quickly open it in the
7975 editor/viewer of your choice). Note that not all terminal emulators
7976 support this functionality yet, but many do. Unfortunately, the
7977 "less" pager doesn't support this yet, hence this functionality is
7978 currently automatically turned off when a pager is started (which
7979 happens quite often due to auto-paging). We hope to remove this
7980 limitation as soon as "less" learns these escape sequences. This new
7981 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
7982 environment variable. For details on these escape sequences see:
7983 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
7985 * networkd's .network files now support a new IPv6MTUBytes= option for
7986 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
7987 option in the [Route] section to configure the MTU to use for
7988 specific routes. It also gained support for configuration of the DHCP
7989 "UserClass" option through the new UserClass= setting. It gained
7990 three new options in the new [CAN] section for configuring CAN
7991 networks. The MULTICAST and ALLMULTI interface flags may now be
7992 controlled explicitly with the new Multicast= and AllMulticast=
7995 * networkd will now automatically make use of the kernel's route
7996 expiration feature, if it is available.
7998 * udevd's .link files now support setting the number of receive and
7999 transmit channels, using the RxChannels=, TxChannels=,
8000 OtherChannels=, CombinedChannels= settings.
8002 * Support for UDPSegmentationOffload= has been removed, given its
8003 limited support in hardware, and waning software support.
8005 * networkd's .netdev files now support creating "netdevsim" interfaces.
8007 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
8008 to query the unit belonging to a specific kernel control group.
8010 * systemd-analyze gained a new verb "cat-config", which may be used to
8011 dump the contents of any configuration file, with all its matching
8012 drop-in files added in, and honouring the usual search and masking
8013 logic applied to systemd configuration files. For example use
8014 "systemd-analyze cat-config systemd/system.conf" to get the complete
8015 system configuration file of systemd how it would be loaded by PID 1
8016 itself. Similar to this, various tools such as systemd-tmpfiles or
8017 systemd-sysusers, gained a new option "--cat-config", which does the
8018 corresponding operation for their own configuration settings. For
8019 example, "systemd-tmpfiles --cat-config" will now output the full
8020 list of tmpfiles.d/ lines in place.
8022 * timedatectl gained three new verbs: "show" shows bus properties of
8023 systemd-timedated, "timesync-status" shows the current NTP
8024 synchronization state of systemd-timesyncd, and "show-timesync"
8025 shows bus properties of systemd-timesyncd.
8027 * systemd-timesyncd gained a bus interface on which it exposes details
8030 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
8031 understood by systemd-timedated. It takes a colon-separated list of
8032 unit names of NTP client services. The list is used by
8033 "timedatectl set-ntp".
8035 * systemd-nspawn gained a new --rlimit= switch for setting initial
8036 resource limits for the container payload. There's a new switch
8037 --hostname= to explicitly override the container's hostname. A new
8038 --no-new-privileges= switch may be used to control the
8039 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
8040 --oom-score-adjust= switch controls the OOM scoring adjustment value
8041 for the payload. The new --cpu-affinity= switch controls the CPU
8042 affinity of the container payload. The new --resolv-conf= switch
8043 allows more detailed control of /etc/resolv.conf handling of the
8044 container. Similarly, the new --timezone= switch allows more detailed
8045 control of /etc/localtime handling of the container.
8047 * systemd-detect-virt gained a new --list switch, which will print a
8048 list of all currently known VM and container environments.
8050 * Support for "Portable Services" has been added, see
8051 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
8052 experimental, but this is expected to change soon. Reflecting this
8053 experimental state, the "portablectl" binary is not installed into
8054 /usr/bin yet. The binary has to be called with the full path
8055 /usr/lib/systemd/portablectl instead.
8057 * journalctl's and systemctl's -o switch now knows a new log output
8058 mode "with-unit". The output it generates is very similar to the
8059 regular "short" mode, but displays the unit name instead of the
8060 syslog tag for each log line. Also, the date is shown with timezone
8061 information. This mode is probably more useful than the classic
8062 "short" output mode for most purposes, except where pixel-perfect
8063 compatibility with classic /var/log/messages formatting is required.
8065 * A new --dump-bus-properties switch has been added to the systemd
8066 binary, which may be used to dump all supported D-Bus properties.
8067 (Options which are still supported, but are deprecated, are *not*
8070 * sd-bus gained a set of new calls:
8071 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
8072 enable/disable the "floating" state of a bus slot object,
8073 i.e. whether the slot object pins the bus it is allocated for into
8074 memory or if the bus slot object gets disconnected when the bus goes
8075 away. sd_bus_open_with_description(),
8076 sd_bus_open_user_with_description(),
8077 sd_bus_open_system_with_description() may be used to allocate bus
8078 objects and set their description string already during allocation.
8080 * sd-event gained support for watching inotify events from the event
8081 loop, in an efficient way, sharing inotify handles between multiple
8082 users. For this a new function sd_event_add_inotify() has been added.
8084 * sd-event and sd-bus gained support for calling special user-supplied
8085 destructor functions for userdata pointers associated with
8086 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
8087 functions sd_bus_slot_set_destroy_callback,
8088 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
8089 sd_bus_track_get_destroy_callback,
8090 sd_event_source_set_destroy_callback,
8091 sd_event_source_get_destroy_callback have been added.
8093 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
8095 * PID 1 will now automatically reschedule .timer units whenever the
8096 local timezone changes. (They previously got rescheduled
8097 automatically when the system clock changed.)
8099 * New documentation has been added to document cgroups delegation,
8100 portable services and the various code quality tools we have set up:
8102 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
8103 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
8104 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
8106 * The Boot Loader Specification has been added to the source tree.
8108 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
8110 While moving it into our source tree we have updated it and further
8111 changes are now accepted through the usual github PR workflow.
8113 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
8114 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
8115 earlier PAM modules. The data in these fields is used to initialize
8116 the session scope's resource properties. Thus external PAM modules
8117 may now configure per-session limits, for example sourced from
8118 external user databases.
8120 * socket units with Accept=yes will now maintain a "refused" counter in
8121 addition to the existing "accepted" counter, counting connections
8122 refused due to the enforced limits.
8124 * The "systemd-path search-binaries-default" command may now be use to
8125 query the default, built-in $PATH PID 1 will pass to the services it
8128 * A new unit file setting PrivateMounts= has been added. It's a boolean
8129 option. If enabled the unit's processes are invoked in their own file
8130 system namespace. Note that this behaviour is also implied if any
8131 other file system namespacing options (such as PrivateTmp=,
8132 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
8133 primarily useful for services that do not use any of the other file
8134 system namespacing options. One such service is systemd-udevd.service
8135 where this is now used by default.
8137 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
8138 when the system is booted in UEFI "secure mode".
8140 * A new unit "system-update-pre.target" is added, which defines an
8141 optional synchronization point for offline system updates, as
8142 implemented by the pre-existing "system-update.target" unit. It
8143 allows ordering services before the service that executes the actual
8144 update process in a generic way.
8146 * Systemd now emits warnings whenever .include syntax is used.
8148 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
8149 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
8150 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
8151 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
8152 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
8153 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
8154 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
8155 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
8156 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
8157 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
8158 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
8159 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
8160 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
8161 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
8162 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
8163 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
8164 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
8165 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
8166 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
8167 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
8168 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
8169 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
8170 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
8171 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
8172 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
8173 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
8174 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
8175 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
8176 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8178 — Berlin, 2018-06-22
8182 * The MemoryAccounting= unit property now defaults to on. After
8183 discussions with the upstream control group maintainers we learnt
8184 that the negative impact of cgroup memory accounting on current
8185 kernels is finally relatively minimal, so that it should be safe to
8186 enable this by default without affecting system performance. Besides
8187 memory accounting only task accounting is turned on by default, all
8188 other forms of resource accounting (CPU, IO, IP) remain off for now,
8189 because it's not clear yet that their impact is small enough to move
8190 from opt-in to opt-out. We recommend downstreams to leave memory
8191 accounting on by default if kernel 4.14 or higher is primarily
8192 used. On very resource constrained systems or when support for old
8193 kernels is a necessity, -Dmemory-accounting-default=false can be used
8194 to revert this change.
8196 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
8197 %udev_rules_update) and the journal catalog (%journal_catalog_update)
8198 from the upgrade scriptlets of individual packages now do nothing.
8199 Transfiletriggers have been added which will perform those updates
8200 once at the end of the transaction.
8202 Similar transfiletriggers have been added to execute any sysctl.d
8203 and binfmt.d rules. Thus, it should be unnecessary to provide any
8204 scriptlets to execute this configuration from package installation
8207 * systemd-sysusers gained a mode where the configuration to execute is
8208 specified on the command line, but this configuration is not executed
8209 directly, but instead it is merged with the configuration on disk,
8210 and the result is executed. This is useful for package installation
8211 scripts which want to create the user before installing any files on
8212 disk (in case some of those files are owned by that user), while
8213 still allowing local admin overrides.
8215 This functionality is exposed to rpm scriptlets through a new
8216 %sysusers_create_package macro. Old %sysusers_create and
8217 %sysusers_create_inline macros are deprecated.
8219 A transfiletrigger for sysusers.d configuration is now installed,
8220 which means that it should be unnecessary to call systemd-sysusers from
8221 package installation scripts, unless the package installs any files
8222 owned by those newly-created users, in which case
8223 %sysusers_create_package should be used.
8225 * Analogous change has been done for systemd-tmpfiles: it gained a mode
8226 where the command-line configuration is merged with the configuration
8227 on disk. This is exposed as the new %tmpfiles_create_package macro,
8228 and %tmpfiles_create is deprecated. A transfiletrigger is installed
8229 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
8230 from package installation scripts.
8232 * sysusers.d configuration for a user may now also specify the group
8233 number, in addition to the user number ("u username 123:456"), or
8234 without the user number ("u username -:456").
8236 * Configution items for systemd-sysusers can now be specified as
8237 positional arguments when the new --inline switch is used.
8239 * The login shell of users created through sysusers.d may now be
8240 specified (previously, it was always /bin/sh for root and
8241 /sbin/nologin for other users).
8243 * systemd-analyze gained a new --global switch to look at global user
8244 configuration. It also gained a unit-paths verb to list the unit load
8245 paths that are compiled into systemd (which can be used with
8246 --systemd, --user, or --global).
8248 * udevadm trigger gained a new --settle/-w option to wait for any
8249 triggered events to finish (but just those, and not any other events
8250 which are triggered meanwhile).
8252 * The action that systemd-logind takes when the lid is closed and the
8253 machine is connected to external power can now be configured using
8254 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
8255 was determined by HandleLidSwitch=, and, for backwards compatibility,
8256 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
8258 * journalctl will periodically call sd_journal_process() to make it
8259 resilient against inotify queue overruns when journal files are
8260 rotated very quickly.
8262 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
8263 sd_bus_get_n_queued_write — may be used to check the number of
8264 pending bus messages.
8266 * systemd gained a new
8267 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
8268 which can be used to migrate foreign processes to scope and service
8269 units. The primary user for this new API is systemd itself: the
8270 systemd --user instance uses this call of the systemd --system
8271 instance to migrate processes if it itself gets the request to
8272 migrate processes and the kernel refuses this due to access
8273 restrictions. Thanks to this "systemd-run --scope --user …" works
8274 again in pure cgroup v2 environments when invoked from the user
8277 * A new TemporaryFileSystem= setting can be used to mask out part of
8278 the real file system tree with tmpfs mounts. This may be combined
8279 with BindPaths= and BindReadOnlyPaths= to hide files or directories
8280 not relevant to the unit, while still allowing some paths lower in
8281 the tree to be accessed.
8283 ProtectHome=tmpfs may now be used to hide user home and runtime
8284 directories from units, in a way that is mostly equivalent to
8285 "TemporaryFileSystem=/home /run/user /root".
8287 * Non-service units are now started with KeyringMode=shared by default.
8288 This means that mount and swapon and other mount tools have access
8289 to keys in the main keyring.
8291 * /sys/fs/bpf is now mounted automatically.
8293 * QNX virtualization is now detected by systemd-detect-virt and may
8294 be used in ConditionVirtualization=.
8296 * IPAccounting= may now be enabled also for slice units.
8298 * A new -Dsplit-bin= build configuration switch may be used to specify
8299 whether bin and sbin directories are merged, or if they should be
8300 included separately in $PATH and various listings of executable
8301 directories. The build configuration scripts will try to autodetect
8302 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
8303 system, but distributions are encouraged to configure this
8306 * A new -Dok-color= build configuration switch may be used to change
8307 the colour of "OK" status messages.
8309 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
8310 PrivateNetwork=yes was buggy in previous versions of systemd. This
8311 means that after the upgrade and daemon-reexec, any such units must
8314 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
8315 will not exclude read-only files owned by root from cleanup.
8317 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
8318 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
8319 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
8320 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
8321 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
8322 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
8323 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
8324 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
8325 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
8326 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
8327 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
8328 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
8329 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
8330 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
8331 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
8332 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
8334 — Warsaw, 2018-03-05
8338 * Some keyboards come with a zoom see-saw or rocker which until now got
8339 mapped to the Linux "zoomin/out" keys in hwdb. However, these
8340 keycodes are not recognized by any major desktop. They now produce
8341 Up/Down key events so that they can be used for scrolling.
8343 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
8344 slightly: previously, if an argument was specified for lines of this
8345 type (i.e. the right-most column was set) this string was appended to
8346 existing files each time systemd-tmpfiles was run. This behaviour was
8347 different from what the documentation said, and not particularly
8348 useful, as repeated systemd-tmpfiles invocations would not be
8349 idempotent and grow such files without bounds. With this release
8350 behaviour has been altered to match what the documentation says:
8351 lines of this type only have an effect if the indicated files don't
8352 exist yet, and only then the argument string is written to the file.
8354 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
8355 systemd-tmpfiles behaviour: previously, read-only files owned by root
8356 were always excluded from the file "aging" algorithm (i.e. the
8357 automatic clean-up of directories like /tmp based on
8358 atime/mtime/ctime). We intend to drop this restriction, and age files
8359 by default even when owned by root and read-only. This behaviour was
8360 inherited from older tools, but there have been requests to remove
8361 it, and it's not obvious why this restriction was made in the first
8362 place. Please speak up now, if you are aware of software that requires
8363 this behaviour, otherwise we'll remove the restriction in v238.
8365 * A new environment variable $SYSTEMD_OFFLINE is now understood by
8366 systemctl. It takes a boolean argument. If on, systemctl assumes it
8367 operates on an "offline" OS tree, and will not attempt to talk to the
8368 service manager. Previously, this mode was implicitly enabled if a
8369 chroot() environment was detected, and this new environment variable
8370 now provides explicit control.
8372 * .path and .socket units may now be created transiently, too.
8373 Previously only service, mount, automount and timer units were
8374 supported as transient units. The systemd-run tool has been updated
8375 to expose this new functionality, you may hence use it now to bind
8376 arbitrary commands to path or socket activation on-the-fly from the
8377 command line. Moreover, almost all properties are now exposed for the
8378 unit types that already supported transient operation.
8380 * The systemd-mount command gained support for a new --owner= parameter
8381 which takes a user name, which is then resolved and included in uid=
8382 and gid= mount options string of the file system to mount.
8384 * A new unit condition ConditionControlGroupController= has been added
8385 that checks whether a specific cgroup controller is available.
8387 * Unit files, udev's .link files, and systemd-networkd's .netdev and
8388 .network files all gained support for a new condition
8389 ConditionKernelVersion= for checking against specific kernel
8392 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
8393 support for configuring device flags in the Flags= setting. In the
8394 same files, the [Tunnel] section gained support for configuring
8395 AllowLocalRemote=. The [Route] section in .network files gained
8396 support for configuring InitialCongestionWindow=,
8397 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
8398 understands RapidCommit=.
8400 * systemd-networkd's DHCPv6 support gained support for Prefix
8403 * sd-bus gained support for a new "watch-bind" feature. When this
8404 feature is enabled, an sd_bus connection may be set up to connect to
8405 an AF_UNIX socket in the file system as soon as it is created. This
8406 functionality is useful for writing early-boot services that
8407 automatically connect to the system bus as soon as it is started,
8408 without ugly time-based polling. systemd-networkd and
8409 systemd-resolved have been updated to make use of this
8410 functionality. busctl exposes this functionality in a new
8411 --watch-bind= command line switch.
8413 * sd-bus will now optionally synthesize a local "Connected" signal as
8414 soon as a D-Bus connection is set up fully. This message mirrors the
8415 already existing "Disconnected" signal which is synthesized when the
8416 connection is terminated. This signal is generally useful but
8417 particularly handy in combination with the "watch-bind" feature
8418 described above. Synthesizing of this message has to be requested
8419 explicitly through the new API call sd_bus_set_connected_signal(). In
8420 addition a new call sd_bus_is_ready() has been added that checks
8421 whether a connection is fully set up (i.e. between the "Connected" and
8422 "Disconnected" signals).
8424 * sd-bus gained two new calls sd_bus_request_name_async() and
8425 sd_bus_release_name_async() for asynchronously registering bus
8426 names. Similar, there is now sd_bus_add_match_async() for installing
8427 a signal match asynchronously. All of systemd's own services have
8428 been updated to make use of these calls. Doing these operations
8429 asynchronously has two benefits: it reduces the risk of deadlocks in
8430 case of cyclic dependencies between bus services, and it speeds up
8431 service initialization since synchronization points for bus
8432 round-trips are removed.
8434 * sd-bus gained two new calls sd_bus_match_signal() and
8435 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
8436 and sd_bus_add_match_async() but instead of taking a D-Bus match
8437 string take match fields as normal function parameters.
8439 * sd-bus gained two new calls sd_bus_set_sender() and
8440 sd_bus_message_set_sender() for setting the sender name of outgoing
8441 messages (either for all outgoing messages or for just one specific
8442 one). These calls are only useful in direct connections as on
8443 brokered connections the broker fills in the sender anyway,
8444 overwriting whatever the client filled in.
8446 * sd-event gained a new pseudo-handle that may be specified on all API
8447 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
8448 used this refers to the default event loop object of the calling
8449 thread. Note however that this does not implicitly allocate one —
8450 which has to be done prior by using sd_event_default(). Similarly
8451 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
8452 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
8453 to the default bus of the specified type of the calling thread. Here
8454 too this does not implicitly allocate bus connection objects, this
8455 has to be done prior with sd_bus_default() and friends.
8457 * sd-event gained a new call pair
8458 sd_event_source_{get|set}_io_fd_own(). This may be used to request
8459 automatic closure of the file descriptor an IO event source watches
8460 when the event source is destroyed.
8462 * systemd-networkd gained support for natively configuring WireGuard
8465 * In previous versions systemd synthesized user records both for the
8466 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
8467 internally. In order to simplify distribution-wide renames of the
8468 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
8469 new transitional flag file has been added: if
8470 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
8471 user and group record within the systemd codebase is disabled.
8473 * systemd-notify gained a new --uid= option for selecting the source
8474 user/UID to use for notification messages sent to the service
8477 * journalctl gained a new --grep= option to list only entries in which
8478 the message matches a certain pattern. By default matching is case
8479 insensitive if the pattern is lowercase, and case sensitive
8480 otherwise. Option --case-sensitive=yes|no can be used to override
8481 this an specify case sensitivity or case insensitivity.
8483 * There's now a "systemd-analyze service-watchdogs" command for printing
8484 the current state of the service runtime watchdog, and optionally
8485 enabling or disabling the per-service watchdogs system-wide if given a
8486 boolean argument (i.e. the concept you configure in WatchdogSec=), for
8487 debugging purposes. There's also a kernel command line option
8488 systemd.service_watchdogs= for controlling the same.
8490 * Two new "log-level" and "log-target" options for systemd-analyze were
8491 added that merge the now deprecated get-log-level, set-log-level and
8492 get-log-target, set-log-target pairs. The deprecated options are still
8493 understood for backwards compatibility. The two new options print the
8494 current value when no arguments are given, and set them when a
8495 level/target is given as an argument.
8497 * sysusers.d's "u" lines now optionally accept both a UID and a GID
8498 specification, separated by a ":" character, in order to create users
8499 where UID and GID do not match.
8501 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
8502 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
8503 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
8504 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
8505 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
8506 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
8507 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
8508 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
8509 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
8510 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
8511 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
8512 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
8513 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
8514 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
8515 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
8516 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
8517 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
8518 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
8519 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
8520 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
8527 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
8528 in v235 has been extended to also set the dummy.ko module option
8529 numdummies=0, preventing the kernel from automatically creating
8530 dummy0. All dummy interfaces must now be explicitly created.
8532 * Unknown '%' specifiers in configuration files are now rejected. This
8533 applies to units and tmpfiles.d configuration. Any percent characters
8534 that are followed by a letter or digit that are not supposed to be
8535 interpreted as the beginning of a specifier should be escaped by
8536 doubling ("%%"). (So "size=5%" is still accepted, as well as
8537 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
8538 valid specifiers today.)
8540 * systemd-resolved now maintains a new dynamic
8541 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
8542 recommended to make /etc/resolv.conf a symlink to it. This file
8543 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
8544 includes dynamically acquired search domains, achieving more correct
8545 DNS resolution by software that bypasses local DNS APIs such as NSS.
8547 * The "uaccess" udev tag has been dropped from /dev/kvm and
8548 /dev/dri/renderD*. These devices now have the 0666 permissions by
8549 default (but this may be changed at build-time). /dev/dri/renderD*
8550 will now be owned by the "render" group along with /dev/kfd.
8552 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
8553 systemd-journal-gatewayd.service and
8554 systemd-journal-upload.service. This means "nss-systemd" must be
8555 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
8556 services are resolved properly.
8558 * In /etc/fstab two new mount options are now understood:
8559 x-systemd.makefs and x-systemd.growfs. The former has the effect that
8560 the configured file system is formatted before it is mounted, the
8561 latter that the file system is resized to the full block device size
8562 after it is mounted (i.e. if the file system is smaller than the
8563 partition it resides on, it's grown). This is similar to the fsck
8564 logic in /etc/fstab, and pulls in systemd-makefs@.service and
8565 systemd-growfs@.service as necessary, similar to
8566 systemd-fsck@.service. Resizing is currently only supported on ext4
8569 * In systemd-networkd, the IPv6 RA logic now optionally may announce
8570 DNS server and domain information.
8572 * Support for the LUKS2 on-disk format for encrypted partitions has
8573 been added. This requires libcryptsetup2 during compilation and
8576 * The systemd --user instance will now signal "readiness" when its
8577 basic.target unit has been reached, instead of when the run queue ran
8578 empty for the first time.
8580 * Tmpfiles.d with user configuration are now also supported.
8581 systemd-tmpfiles gained a new --user switch, and snippets placed in
8582 ~/.config/user-tmpfiles.d/ and corresponding directories will be
8583 executed by systemd-tmpfiles --user running in the new
8584 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
8585 running in the user session.
8587 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
8588 %S resolves to the top-level state directory (/var/lib for the system
8589 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
8590 top-level cache directory (/var/cache for the system instance,
8591 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
8592 logs directory (/var/log for the system instance,
8593 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
8594 existing %t specifier, that resolves to the top-level runtime
8595 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
8598 * journalctl learnt a new parameter --output-fields= for limiting the
8599 set of journal fields to output in verbose and JSON output modes.
8601 * systemd-timesyncd's configuration file gained a new option
8602 RootDistanceMaxSec= for setting the maximum root distance of servers
8603 it'll use, as well as the new options PollIntervalMinSec= and
8604 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
8606 * bootctl gained a new command "list" for listing all available boot
8607 menu items on systems that follow the boot loader specification.
8609 * systemctl gained a new --dry-run switch that shows what would be done
8610 instead of doing it, and is currently supported by the shutdown and
8613 * ConditionSecurity= can now detect the TOMOYO security module.
8615 * Unit file [Install] sections are now also respected in unit drop-in
8616 files. This is intended to be used by drop-ins under /usr/lib/.
8618 * systemd-firstboot may now also set the initial keyboard mapping.
8620 * Udev "changed" events for devices which are exposed as systemd
8621 .device units are now propagated to units specified in
8622 ReloadPropagatedFrom= as reload requests.
8624 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
8625 unit template name (i.e. a name in the form of 'foobar@.service',
8626 without the instance component between the '@' and - the '.'), then
8627 the escaped sysfs path of the device is automatically used as the
8630 * SystemCallFilter= in unit files has been extended so that an "errno"
8631 can be specified individually for each system call. Example:
8632 SystemCallFilter=~uname:EILSEQ.
8634 * The cgroup delegation logic has been substantially updated. Delegate=
8635 now optionally takes a list of controllers (instead of a boolean, as
8636 before), which lists the controllers to delegate at least.
8638 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
8640 * A new LogLevelMax= setting configures the maximum log level any
8641 process of the service may log at (i.e. anything with a lesser
8642 priority than what is specified is automatically dropped). A new
8643 LogExtraFields= setting allows configuration of additional journal
8644 fields to attach to all log records generated by any of the unit's
8647 * New StandardInputData= and StandardInputText= settings along with the
8648 new option StandardInput=data may be used to configure textual or
8649 binary data that shall be passed to the executed service process via
8650 standard input, encoded in-line in the unit file.
8652 * StandardInput=, StandardOutput= and StandardError= may now be used to
8653 connect stdin/stdout/stderr of executed processes directly with a
8654 file or AF_UNIX socket in the file system, using the new "file:" option.
8656 * A new unit file option CollectMode= has been added, that allows
8657 tweaking the garbage collection logic for units. It may be used to
8658 tell systemd to garbage collect units that have failed automatically
8659 (normally it only GCs units that exited successfully). systemd-run
8660 and systemd-mount expose this new functionality with a new -G option.
8662 * "machinectl bind" may now be used to bind mount non-directories
8663 (i.e. regularfiles, devices, fifos, sockets).
8665 * systemd-analyze gained a new verb "calendar" for validating and
8666 testing calendar time specifications to use for OnCalendar= in timer
8667 units. Besides validating the expression it will calculate the next
8668 time the specified expression would elapse.
8670 * In addition to the pre-existing FailureAction= unit file setting
8671 there's now SuccessAction=, for configuring a shutdown action to
8672 execute when a unit completes successfully. This is useful in
8673 particular inside containers that shall terminate after some workload
8674 has been completed. Also, both options are now supported for all unit
8675 types, not just services.
8677 * networkds's IP rule support gained two new options
8678 IncomingInterface= and OutgoingInterface= for configuring the incoming
8679 and outgoing interfaces of configured rules. systemd-networkd also
8680 gained support for "vxcan" network devices.
8682 * networkd gained a new setting RequiredForOnline=, taking a
8683 boolean. If set, systemd-wait-online will take it into consideration
8684 when determining that the system is up, otherwise it will ignore the
8685 interface for this purpose.
8687 * The sd_notify() protocol gained support for a new operation: with
8688 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
8689 store again, ahead of POLLHUP or POLLERR when they are removed
8692 * A new document doc/UIDS-GIDS.md has been added to the source tree,
8693 that documents the UID/GID range and assignment assumptions and
8694 requirements of systemd.
8696 * The watchdog device PID 1 will ping may now be configured through the
8697 WatchdogDevice= configuration file setting, or by setting the
8698 systemd.watchdog_service= kernel commandline option.
8700 * systemd-resolved's gained support for registering DNS-SD services on
8701 the local network using MulticastDNS. Services may either be
8702 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
8703 the same dir below /run, /usr/lib), or through its D-Bus API.
8705 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
8706 extend the effective start, runtime, and stop time. The service must
8707 continue to send EXTEND_TIMEOUT_USEC within the period specified to
8708 prevent the service manager from making the service as timedout.
8710 * systemd-resolved's DNSSEC support gained support for RFC 8080
8711 (Ed25519 keys and signatures).
8713 * The systemd-resolve command line tool gained a new set of options
8714 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
8715 --set-nta= and --revert to configure per-interface DNS configuration
8716 dynamically during runtime. It's useful for pushing DNS information
8717 into systemd-resolved from DNS hook scripts that various interface
8718 managing software supports (such as pppd).
8720 * systemd-nspawn gained a new --network-namespace-path= command line
8721 option, which may be used to make a container join an existing
8722 network namespace, by specifying a path to a "netns" file.
8724 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
8725 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
8726 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
8727 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
8728 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
8729 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
8730 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
8731 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
8732 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
8733 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
8734 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
8735 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
8736 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
8737 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
8738 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
8739 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
8740 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
8741 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
8742 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
8743 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
8744 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
8745 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
8746 Jędrzejewski-Szmek, Zeal Jagannatha
8748 — Berlin, 2017-12-14
8752 * INCOMPATIBILITY: systemd-logind.service and other long-running
8753 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
8754 communication with the outside. This generally improves security of
8755 the system, and is in almost all cases a safe and good choice, as
8756 these services do not and should not provide any network-facing
8757 functionality. However, systemd-logind uses the glibc NSS API to
8758 query the user database. This creates problems on systems where NSS
8759 is set up to directly consult network services for user database
8760 lookups. In particular, this creates incompatibilities with the
8761 "nss-nis" module, which attempts to directly contact the NIS/YP
8762 network servers it is configured for, and will now consistently
8763 fail. In such cases, it is possible to turn off IP sandboxing for
8764 systemd-logind.service (set IPAddressDeny= in its [Service] section
8765 to the empty string, via a .d/ unit file drop-in). Downstream
8766 distributions might want to update their nss-nis packaging to include
8767 such a drop-in snippet, accordingly, to hide this incompatibility
8768 from the user. Another option is to make use of glibc's nscd service
8769 to proxy such network requests through a privilege-separated, minimal
8770 local caching daemon, or to switch to more modern technologies such
8771 sssd, whose NSS hook-ups generally do not involve direct network
8772 access. In general, we think it's definitely time to question the
8773 implementation choices of nss-nis, i.e. whether it's a good idea
8774 today to embed a network-facing loadable module into all local
8775 processes that need to query the user database, including the most
8776 trivial and benign ones, such as "ls". For more details about
8777 IPAddressDeny= see below.
8779 * A new modprobe.d drop-in is now shipped by default that sets the
8780 bonding module option max_bonds=0. This overrides the kernel default,
8781 to avoid conflicts and ambiguity as to whether or not bond0 should be
8782 managed by systemd-networkd or not. This resolves multiple issues
8783 with bond0 properties not being applied, when bond0 is configured
8784 with systemd-networkd. Distributors may choose to not package this,
8785 however in that case users will be prevented from correctly managing
8786 bond0 interface using systemd-networkd.
8788 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
8789 which print the logging level and target of the system manager. They
8790 complement the existing "set-log-level" and "set-log-target" verbs
8791 used to change those values.
8793 * journald.conf gained a new boolean setting ReadKMsg= which defaults
8794 to on. If turned off kernel log messages will not be read by
8795 systemd-journald or included in the logs. It also gained a new
8796 setting LineMax= for configuring the maximum line length in
8797 STDOUT/STDERR log streams. The new default for this value is 48K, up
8798 from the previous hardcoded 2048.
8800 * A new unit setting RuntimeDirectoryPreserve= has been added, which
8801 allows more detailed control of what to do with a runtime directory
8802 configured with RuntimeDirectory= (i.e. a directory below /run or
8803 $XDG_RUNTIME_DIR) after a unit is stopped.
8805 * The RuntimeDirectory= setting for units gained support for creating
8806 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
8807 one top-level directory.
8809 * Units gained new options StateDirectory=, CacheDirectory=,
8810 LogsDirectory= and ConfigurationDirectory= which are closely related
8811 to RuntimeDirectory= but manage per-service directories below
8812 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
8813 possible to write unit files which when activated automatically gain
8814 properly owned service specific directories in these locations, thus
8815 making unit files self-contained and increasing compatibility with
8816 stateless systems and factory reset where /etc or /var are
8817 unpopulated at boot. Matching these new settings there's also
8818 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
8819 ConfigurationDirectoryMode= for configuring the access mode of these
8820 directories. These settings are particularly useful in combination
8821 with DynamicUser=yes as they provide secure, properly-owned,
8822 writable, and stateful locations for storage, excluded from the
8823 sandbox that such services live in otherwise.
8825 * Automake support has been removed from this release. systemd is now
8828 * systemd-journald will now aggressively cache client metadata during
8829 runtime, speeding up log write performance under pressure. This comes
8830 at a small price though: as much of the metadata is read
8831 asynchronously from /proc/ (and isn't implicitly attached to log
8832 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
8833 metadata stored alongside a log entry might be slightly
8834 out-of-date. Previously it could only be slightly newer than the log
8835 message. The time window is small however, and given that the kernel
8836 is unlikely to be improved anytime soon in this regard, this appears
8839 * nss-myhostname/systemd-resolved will now by default synthesize an
8840 A/AAAA resource record for the "_gateway" hostname, pointing to the
8841 current default IP gateway. Previously it did that for the "gateway"
8842 name, hampering adoption, as some distributions wanted to leave that
8843 hostname open for local use. The old behaviour may still be
8844 requested at build time.
8846 * systemd-networkd's [Address] section in .network files gained a new
8847 Scope= setting for configuring the IP address scope. The [Network]
8848 section gained a new boolean setting ConfigureWithoutCarrier= that
8849 tells systemd-networkd to ignore link sensing when configuring the
8850 device. The [DHCP] section gained a new Anonymize= boolean option for
8851 turning on a number of options suggested in RFC 7844. A new
8852 [RoutingPolicyRule] section has been added for configuring the IP
8853 routing policy. The [Route] section has gained support for a new
8854 Type= setting which permits configuring
8855 blackhole/unreachable/prohibit routes.
8857 * The [VRF] section in .netdev files gained a new Table= setting for
8858 configuring the routing table to use. The [Tunnel] section gained a
8859 new Independent= boolean field for configuring tunnels independent of
8860 an underlying network interface. The [Bridge] section gained a new
8861 GroupForwardMask= option for configuration of propagation of link
8862 local frames between bridge ports.
8864 * The WakeOnLan= setting in .link files gained support for a number of
8865 new modes. A new TCP6SegmentationOffload= setting has been added for
8866 configuring TCP/IPv6 hardware segmentation offload.
8868 * The IPv6 RA sender implementation may now optionally send out RDNSS
8869 and RDNSSL records to supply DNS configuration to peers.
8871 * systemd-nspawn gained support for a new --system-call-filter= command
8872 line option for adding and removing entries in the default system
8873 call filter it applies. Moreover systemd-nspawn has been changed to
8874 implement a system call allow list instead of a deny list.
8876 * systemd-run gained support for a new --pipe command line option. If
8877 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
8878 are directly passed on to the activated transient service
8879 executable. This allows invoking arbitrary processes as systemd
8880 services (for example to take benefit of dependency management,
8881 accounting management, resource management or log management that is
8882 done automatically for services) — while still allowing them to be
8883 integrated in a classic UNIX shell pipeline.
8885 * When a service sends RELOAD=1 via sd_notify() and reload propagation
8886 using ReloadPropagationTo= is configured, a reload is now propagated
8887 to configured units. (Previously this was only done on explicitly
8888 requested reloads, using "systemctl reload" or an equivalent
8891 * For each service unit a restart counter is now kept: it is increased
8892 each time the service is restarted due to Restart=, and may be
8893 queried using "systemctl show -p NRestarts …".
8895 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
8896 @signal and @timer have been added, for usage with SystemCallFilter=
8897 in unit files and the new --system-call-filter= command line option
8898 of systemd-nspawn (see above).
8900 * ExecStart= lines in unit files gained two new modifiers: when a
8901 command line is prefixed with "!" the command will be executed as
8902 configured, except for the credentials applied by
8903 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
8904 "+", but does still apply namespacing options unlike "+". There's
8905 also "!!" now, which is mostly identical, but becomes a NOP on
8906 systems that support ambient capabilities. This is useful to write
8907 unit files that work with ambient capabilities where possible but
8908 automatically fall back to traditional privilege dropping mechanisms
8909 on systems where this is not supported.
8911 * ListenNetlink= settings in socket units now support RDMA netlink
8914 * A new unit file setting LockPersonality= has been added which permits
8915 locking down the chosen execution domain ("personality") of a service
8918 * A new special target "getty-pre.target" has been added, which is
8919 ordered before all text logins, and may be used to order services
8920 before textual logins acquire access to the console.
8922 * systemd will now attempt to load the virtio-rng.ko kernel module very
8923 early on if a VM environment supporting this is detected. This should
8924 improve entropy during early boot in virtualized environments.
8926 * A _netdev option is now supported in /etc/crypttab that operates in a
8927 similar way as the same option in /etc/fstab: it permits configuring
8928 encrypted devices that need to be ordered after the network is up.
8929 Following this logic, two new special targets
8930 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
8931 added that are to cryptsetup.target what remote-fs.target and
8932 remote-fs-pre.target are to local-fs.target.
8934 * Service units gained a new UnsetEnvironment= setting which permits
8935 unsetting specific environment variables for services that are
8936 normally passed to it (for example in order to mask out locale
8937 settings for specific services that can't deal with it).
8939 * Units acquired a new boolean option IPAccounting=. When turned on, IP
8940 traffic accounting (packet count as well as byte count) is done for
8941 the service, and shown as part of "systemctl status" or "systemd-run
8944 * Service units acquired two new options IPAddressAllow= and
8945 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
8946 for configuring a simple IP access control list for all sockets of
8947 the unit. These options are available also on .slice and .socket
8948 units, permitting flexible access list configuration for individual
8949 services as well as groups of services (as defined by a slice unit),
8950 including system-wide. Note that IP ACLs configured this way are
8951 enforced on every single IPv4 and IPv6 socket created by any process
8952 of the service unit, and apply to ingress as well as egress traffic.
8954 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
8955 structured log message is generated each time the unit is stopped,
8956 containing information about the consumed resources of this
8959 * A new setting KeyringMode= has been added to unit files, which may be
8960 used to control how the kernel keyring is set up for executed
8963 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
8964 "systemctl kexec" and "systemctl exit" are now always asynchronous in
8965 behaviour (that is: these commands return immediately after the
8966 operation was enqueued instead of waiting for the operation to
8967 complete). Previously, "systemctl poweroff" and "systemctl reboot"
8968 were asynchronous on systems using systemd-logind (i.e. almost
8969 always, and like they were on sysvinit), and the other three commands
8970 were unconditionally synchronous. With this release this is cleaned
8971 up, and callers will see the same asynchronous behaviour on all
8972 systems for all five operations.
8974 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
8977 * .timer units now accept calendar specifications in other timezones
8978 than UTC or the local timezone.
8980 * The tmpfiles snippet var.conf has been changed to create
8981 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
8982 the "utmp" group already, and it appears to be generally understood
8983 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
8984 databases. Previously this was implemented correctly for all these
8985 databases excepts btmp, which has been opened up like this now
8986 too. Note that while the other databases are world-readable
8987 (i.e. 0644), btmp is not and remains more restrictive.
8989 * The systemd-resolve tool gained a new --reset-server-features
8990 switch. When invoked like this systemd-resolved will forget
8991 everything it learnt about the features supported by the configured
8992 upstream DNS servers, and restarts the feature probing logic on the
8993 next resolver look-up for them at the highest feature level
8996 * The status dump systemd-resolved sends to the logs upon receiving
8997 SIGUSR1 now also includes information about all DNS servers it is
8998 configured to use, and the features levels it probed for them.
9000 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
9001 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
9002 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
9003 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
9004 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
9005 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
9006 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
9007 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
9008 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
9009 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
9010 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
9011 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
9012 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
9013 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
9014 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
9015 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
9016 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
9017 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
9018 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
9019 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9021 — Berlin, 2017-10-06
9025 * Meson is now supported as build system in addition to Automake. It is
9026 our plan to remove Automake in one of our next releases, so that
9027 Meson becomes our exclusive build system. Hence, please start using
9028 the Meson build system in your downstream packaging. There's plenty
9029 of documentation around how to use Meson, the extremely brief
9032 ./autogen.sh && ./configure && make && sudo make install
9036 meson build && ninja -C build && sudo ninja -C build install
9038 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
9039 which permits configuring a timeout on the time a job is
9040 running. This is particularly useful for setting timeouts on jobs for
9043 * Unit files gained two new options ConditionUser= and ConditionGroup=
9044 for conditionalizing units based on the identity of the user/group
9045 running a systemd user instance.
9047 * systemd-networkd now understands a new FlowLabel= setting in the
9048 [VXLAN] section of .network files, as well as a Priority= in
9049 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
9050 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
9051 gained support for configuration of GENEVE links, and IPv6 address
9052 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
9054 * .link files now understand a new Port= setting.
9056 * systemd-networkd's DHCP support gained support for DHCP option 119
9057 (domain search list).
9059 * systemd-networkd gained support for serving IPv6 address ranges using
9060 the Router Advertisement protocol. The new .network configuration
9061 section [IPv6Prefix] may be used to configure the ranges to
9062 serve. This is implemented based on a new, minimal, native server
9063 implementation of RA.
9065 * journalctl's --output= switch gained support for a new parameter
9066 "short-iso-precise" for a mode where timestamps are shown as precise
9069 * systemd-udevd's "net_id" builtin may now generate stable network
9070 interface names from IBM PowerVM VIO devices as well as ACPI platform
9073 * MulticastDNS support in systemd-resolved may now be explicitly
9074 enabled/disabled using the new MulticastDNS= configuration file
9077 * systemd-resolved may now optionally use libidn2 instead of the libidn
9078 for processing internationalized domain names. Support for libidn2
9079 should be considered experimental and should not be enabled by
9082 * "machinectl pull-tar" and related call may now do verification of
9083 downloaded images using SUSE-style .sha256 checksum files in addition
9084 to the already existing support for validating using Ubuntu-style
9087 * sd-bus gained support for a new sd_bus_message_appendv() call which
9088 is va_list equivalent of sd_bus_message_append().
9090 * sd-boot gained support for validating images using SHIM/MOK.
9092 * The SMACK code learnt support for "onlycap".
9094 * systemd-mount --umount is now much smarter in figuring out how to
9095 properly unmount a device given its mount or device path.
9097 * The code to call libnss_dns as a fallback from libnss_resolve when
9098 the communication with systemd-resolved fails was removed. This
9099 fallback was redundant and interfered with the [!UNAVAIL=return]
9100 suffix. See nss-resolve(8) for the recommended configuration.
9102 * systemd-logind may now be restarted without losing state. It stores
9103 the file descriptors for devices it manages in the system manager
9104 using the FDSTORE= mechanism. Please note that further changes in
9105 other components may be required to make use of this (for example
9106 Xorg has code to listen for stops of systemd-logind and terminate
9107 itself when logind is stopped or restarted, in order to avoid using
9108 stale file descriptors for graphical devices, which is now
9109 counterproductive and must be reverted in order for restarts of
9110 systemd-logind to be safe. See
9111 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
9113 * All kernel-install plugins are called with the environment variable
9114 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
9115 /etc/machine-id. If the machine ID could not be determined,
9116 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
9117 anything in the entry directory (passed as the second argument) if
9118 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
9119 temporary directory is passed as the entry directory and removed
9120 after all the plugins exit.
9122 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
9123 will now use its value as the machine ID instead of the machine ID
9124 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
9125 /etc/machine-info and no machine ID is set in /etc/machine-id,
9126 kernel-install will try to store the current machine ID there as
9127 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
9128 will generate a new UUID, store it in /etc/machine-info as
9129 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
9131 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
9132 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
9133 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
9134 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
9135 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
9136 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
9137 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
9138 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
9139 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
9140 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
9141 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
9142 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
9143 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
9144 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
9145 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
9146 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
9147 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
9148 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
9149 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
9150 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
9151 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
9152 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
9153 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
9154 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
9155 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
9156 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
9157 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
9158 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
9159 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
9162 — Berlin, 2017-07-12
9166 * The "hybrid" control group mode has been modified to improve
9167 compatibility with "legacy" cgroups-v1 setups. Specifically, the
9168 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
9169 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
9170 cgroups-v1 hierarchy), the only externally visible change being that
9171 the cgroups-v2 hierarchy is also mounted, to
9172 /sys/fs/cgroup/unified. This should provide a large degree of
9173 compatibility with "legacy" cgroups-v1, while taking benefit of the
9174 better management capabilities of cgroups-v2.
9176 * The default control group setup mode may be selected both a boot-time
9177 via a set of kernel command line parameters (specifically:
9178 systemd.unified_cgroup_hierarchy= and
9179 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
9180 default selected on the configure command line
9181 (--with-default-hierarchy=). The upstream default is "hybrid"
9182 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
9183 this will change in a future systemd version to be "unified" (pure
9184 cgroups-v2 mode). The third option for the compile time option is
9185 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
9186 distributions to default to "hybrid" mode for release distributions,
9187 starting with v233. We recommend "unified" for development
9188 distributions (specifically: distributions such as Fedora's rawhide)
9189 as that's where things are headed in the long run. Use "legacy" for
9190 greatest stability and compatibility only.
9192 * Note one current limitation of "unified" and "hybrid" control group
9193 setup modes: the kernel currently does not permit the systemd --user
9194 instance (i.e. unprivileged code) to migrate processes between two
9195 disconnected cgroup subtrees, even if both are managed and owned by
9196 the user. This effectively means "systemd-run --user --scope" doesn't
9197 work when invoked from outside of any "systemd --user" service or
9198 scope. Specifically, it is not supported from session scopes. We are
9199 working on fixing this in a future systemd version. (See #3388 for
9200 further details about this.)
9202 * DBus policy files are now installed into /usr rather than /etc. Make
9203 sure your system has dbus >= 1.9.18 running before upgrading to this
9204 version, or override the install path with --with-dbuspolicydir= .
9206 * All python scripts shipped with systemd (specifically: the various
9207 tests written in Python) now require Python 3.
9209 * systemd unit tests can now run standalone (without the source or
9210 build directories), and can be installed into /usr/lib/systemd/tests/
9211 with 'make install-tests'.
9213 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
9214 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
9217 * Support for the %c, %r, %R specifiers in unit files has been
9218 removed. Specifiers are not supposed to be dependent on configuration
9219 in the unit file itself (so that they resolve the same regardless
9220 where used in the unit files), but these specifiers were influenced
9221 by the Slice= option.
9223 * The shell invoked by debug-shell.service now defaults to /bin/sh in
9224 all cases. If distributions want to use a different shell for this
9225 purpose (for example Fedora's /sbin/sushell) they need to specify
9226 this explicitly at configure time using --with-debug-shell=.
9228 * The confirmation spawn prompt has been reworked to offer the
9231 (c)ontinue, proceed without asking anymore
9232 (D)ump, show the state of the unit
9233 (f)ail, don't execute the command and pretend it failed
9235 (i)nfo, show a short summary of the unit
9236 (j)obs, show jobs that are in progress
9237 (s)kip, don't execute the command and pretend it succeeded
9238 (y)es, execute the command
9240 The 'n' choice for the confirmation spawn prompt has been removed,
9241 because its meaning was confusing.
9243 The prompt may now also be redirected to an alternative console by
9244 specifying the console as parameter to systemd.confirm_spawn=.
9246 * Services of Type=notify require a READY=1 notification to be sent
9247 during startup. If no such message is sent, the service now fails,
9248 even if the main process exited with a successful exit code.
9250 * Services that fail to start up correctly now always have their
9251 ExecStopPost= commands executed. Previously, they'd enter "failed"
9252 state directly, without executing these commands.
9254 * The option MulticastDNS= of network configuration files has acquired
9255 an actual implementation. With MulticastDNS=yes a host can resolve
9256 names of remote hosts and reply to mDNS A and AAAA requests.
9258 * When units are about to be started an additional check is now done to
9259 ensure that all dependencies of type BindsTo= (when used in
9260 combination with After=) have been started.
9262 * systemd-analyze gained a new verb "syscall-filter" which shows which
9263 system call groups are defined for the SystemCallFilter= unit file
9264 setting, and which system calls they contain.
9266 * A new system call filter group "@filesystem" has been added,
9267 consisting of various file system related system calls. Group
9268 "@reboot" has been added, covering reboot, kexec and shutdown related
9269 calls. Finally, group "@swap" has been added covering swap
9270 configuration related calls.
9272 * A new unit file option RestrictNamespaces= has been added that may be
9273 used to restrict access to the various process namespace types the
9274 Linux kernel provides. Specifically, it may be used to take away the
9275 right for a service unit to create additional file system, network,
9276 user, and other namespaces. This sandboxing option is particularly
9277 relevant due to the high amount of recently discovered namespacing
9278 related vulnerabilities in the kernel.
9280 * systemd-udev's .link files gained support for a new AutoNegotiation=
9281 setting for configuring Ethernet auto-negotiation.
9283 * systemd-networkd's .network files gained support for a new
9284 ListenPort= setting in the [DHCP] section to explicitly configure the
9285 UDP client port the DHCP client shall listen on.
9287 * .network files gained a new Unmanaged= boolean setting for explicitly
9288 excluding one or more interfaces from management by systemd-networkd.
9290 * The systemd-networkd ProxyARP= option has been renamed to
9291 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
9292 renamed to ReduceARPProxy=. The old names continue to be available
9295 * systemd-networkd gained support for configuring IPv6 Proxy NDP
9296 addresses via the new IPv6ProxyNDPAddress= .network file setting.
9298 * systemd-networkd's bonding device support gained support for two new
9299 configuration options ActiveSlave= and PrimarySlave=.
9301 * The various options in the [Match] section of .network files gained
9302 support for negative matching.
9304 * New systemd-specific mount options are now understood in /etc/fstab:
9306 x-systemd.mount-timeout= may be used to configure the maximum
9307 permitted runtime of the mount command.
9309 x-systemd.device-bound may be set to bind a mount point to its
9310 backing device unit, in order to automatically remove a mount point
9311 if its backing device is unplugged. This option may also be
9312 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
9313 on the block device, which is now automatically set for all CDROM
9314 drives, so that mounted CDs are automatically unmounted when they are
9315 removed from the drive.
9317 x-systemd.after= and x-systemd.before= may be used to explicitly
9318 order a mount after or before another unit or mount point.
9320 * Enqueued start jobs for device units are now automatically garbage
9321 collected if there are no jobs waiting for them anymore.
9323 * systemctl list-jobs gained two new switches: with --after, for every
9324 queued job the jobs it's waiting for are shown; with --before the
9325 jobs which it's blocking are shown.
9327 * systemd-nspawn gained support for ephemeral boots from disk images
9328 (or in other words: --ephemeral and --image= may now be
9329 combined). Moreover, ephemeral boots are now supported for normal
9330 directories, even if the backing file system is not btrfs. Of course,
9331 if the file system does not support file system snapshots or
9332 reflinks, the initial copy operation will be relatively expensive, but
9333 this should still be suitable for many use cases.
9335 * Calendar time specifications in .timer units now support
9336 specifications relative to the end of a month by using "~" instead of
9337 "-" as separator between month and day. For example, "*-02~03" means
9338 "the third last day in February". In addition a new syntax for
9339 repeated events has been added using the "/" character. For example,
9340 "9..17/2:00" means "every two hours from 9am to 5pm".
9342 * systemd-socket-proxyd gained a new parameter --connections-max= for
9343 configuring the maximum number of concurrent connections.
9345 * sd-id128 gained a new API for generating unique IDs for the host in a
9346 way that does not leak the machine ID. Specifically,
9347 sd_id128_get_machine_app_specific() derives an ID based on the
9348 machine ID in a well-defined, non-reversible, stable way. This is
9349 useful whenever an identifier for the host is needed but where the
9350 identifier shall not be useful to identify the system beyond the
9351 scope of the application itself. (Internally this uses HMAC-SHA256 as
9352 keyed hash function using the machine ID as input.)
9354 * NotifyAccess= gained a new supported value "exec". When set
9355 notifications are accepted from all processes systemd itself invoked,
9356 including all control processes.
9358 * .nspawn files gained support for defining overlay mounts using the
9359 Overlay= and OverlayReadOnly= options. Previously this functionality
9360 was only available on the systemd-nspawn command line.
9362 * systemd-nspawn's --bind= and --overlay= options gained support for
9363 bind/overlay mounts whose source lies within the container tree by
9364 prefixing the source path with "+".
9366 * systemd-nspawn's --bind= and --overlay= options gained support for
9367 automatically allocating a temporary source directory in /var/tmp
9368 that is removed when the container dies. Specifically, if the source
9369 directory is specified as empty string this mechanism is selected. An
9370 example usage is --overlay=+/var::/var, which creates an overlay
9371 mount based on the original /var contained in the image, overlaid
9372 with a temporary directory in the host's /var/tmp. This way changes
9373 to /var are automatically flushed when the container shuts down.
9375 * systemd-nspawn --image= option does now permit raw file system block
9376 devices (in addition to images containing partition tables, as
9379 * The disk image dissection logic in systemd-nspawn gained support for
9380 automatically setting up LUKS encrypted as well as Verity protected
9381 partitions. When a container is booted from an encrypted image the
9382 passphrase is queried at start-up time. When a container with Verity
9383 data is started, the root hash is search in a ".roothash" file
9384 accompanying the disk image (alternatively, pass the root hash via
9385 the new --root-hash= command line option).
9387 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
9388 be used to dissect disk images the same way as systemd-nspawn does
9389 it, following the Bootable Partition Specification. It may even be
9390 used to mount disk images with complex partition setups (including
9391 LUKS and Verity partitions) to a local host directory, in order to
9392 inspect them. This tool is not considered public API (yet), and is
9393 thus not installed into /usr/bin. Please do not rely on its
9394 existence, since it might go away or be changed in later systemd
9397 * A new generator "systemd-verity-generator" has been added, similar in
9398 style to "systemd-cryptsetup-generator", permitting automatic setup of
9399 Verity root partitions when systemd boots up. In order to make use of
9400 this your partition setup should follow the Discoverable Partitions
9401 Specification, and the GPT partition ID of the root file system
9402 partition should be identical to the upper 128-bit of the Verity root
9403 hash. The GPT partition ID of the Verity partition protecting it
9404 should be the lower 128-bit of the Verity root hash. If the partition
9405 image follows this model it is sufficient to specify a single
9406 "roothash=" kernel command line argument to both configure which root
9407 image and verity partition to use as well as the root hash for
9408 it. Note that systemd-nspawn's Verity support follows the same
9409 semantics, meaning that disk images with proper Verity data in place
9410 may be booted in containers with systemd-nspawn as well as on
9411 physical systems via the verity generator. Also note that the "mkosi"
9412 tool available at https://github.com/systemd/mkosi has been updated
9413 to generate Verity protected disk images following this scheme. In
9414 fact, it has been updated to generate disk images that optionally
9415 implement a complete UEFI SecureBoot trust chain, involving a signed
9416 kernel and initrd image that incorporates such a root hash as well as
9417 a Verity-enabled root partition.
9419 * The hardware database (hwdb) udev supports has been updated to carry
9420 accelerometer quirks.
9422 * All system services are now run with a fresh kernel keyring set up
9423 for them. The invocation ID is stored by default in it, thus
9424 providing a safe, non-overridable way to determine the invocation
9427 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
9428 options for bind mounting arbitrary paths in a service-specific
9429 way. When these options are used, arbitrary host or service files and
9430 directories may be mounted to arbitrary locations in the service's
9433 * Documentation has been added that lists all of systemd's low-level
9434 environment variables:
9436 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
9438 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
9439 whether a specific socket file descriptor matches a specified socket
9442 * systemd-firstboot has been updated to check for the
9443 systemd.firstboot= kernel command line option. It accepts a boolean
9444 and when set to false the first boot questions are skipped.
9446 * systemd-fstab-generator has been updated to check for the
9447 systemd.volatile= kernel command line option, which either takes an
9448 optional boolean parameter or the special value "state". If used the
9449 system may be booted in a "volatile" boot mode. Specifically,
9450 "systemd.volatile" is used, the root directory will be mounted as
9451 tmpfs, and only /usr is mounted from the actual root file system. If
9452 "systemd.volatile=state" is used, the root directory will be mounted
9453 as usual, but /var is mounted as tmpfs. This concept provides similar
9454 functionality as systemd-nspawn's --volatile= option, but provides it
9455 on physical boots. Use this option for implementing stateless
9456 systems, or testing systems with all state and/or configuration reset
9457 to the defaults. (Note though that many distributions are not
9458 prepared to boot up without a populated /etc or /var, though.)
9460 * systemd-gpt-auto-generator gained support for LUKS encrypted root
9461 partitions. Previously it only supported LUKS encrypted partitions
9462 for all other uses, except for the root partition itself.
9464 * Socket units gained support for listening on AF_VSOCK sockets for
9465 communication in virtualized QEMU environments.
9467 * The "configure" script gained a new option --with-fallback-hostname=
9468 for specifying the fallback hostname to use if none is configured in
9469 /etc/hostname. For example, by specifying
9470 --with-fallback-hostname=fedora it is possible to default to a
9471 hostname of "fedora" on pristine installations.
9473 * systemd-cgls gained support for a new --unit= switch for listing only
9474 the control groups of a specific unit. Similar --user-unit= has been
9475 added for listing only the control groups of a specific user unit.
9477 * systemd-mount gained a new --umount switch for unmounting a mount or
9478 automount point (and all mount/automount points below it).
9480 * systemd will now refuse full configuration reloads (via systemctl
9481 daemon-reload and related calls) unless at least 16MiB of free space
9482 are available in /run. This is a safety precaution in order to ensure
9483 that generators can safely operate after the reload completed.
9485 * A new unit file option RootImage= has been added, which has a similar
9486 effect as RootDirectory= but mounts the service's root directory from
9487 a disk image instead of plain directory. This logic reuses the same
9488 image dissection and mount logic that systemd-nspawn already uses,
9489 and hence supports any disk images systemd-nspawn supports, including
9490 those following the Discoverable Partition Specification, as well as
9491 Verity enabled images. This option enables systemd to run system
9492 services directly off disk images acting as resource bundles,
9493 possibly even including full integrity data.
9495 * A new MountAPIVFS= unit file option has been added, taking a boolean
9496 argument. If enabled /proc, /sys and /dev (collectively called the
9497 "API VFS") will be mounted for the service. This is only relevant if
9498 RootDirectory= or RootImage= is used for the service, as these mounts
9499 are of course in place in the host mount namespace anyway.
9501 * systemd-nspawn gained support for a new --pivot-root= switch. If
9502 specified the root directory within the container image is pivoted to
9503 the specified mount point, while the original root disk is moved to a
9504 different place. This option enables booting of ostree images
9505 directly with systemd-nspawn.
9507 * The systemd build scripts will no longer complain if the NTP server
9508 addresses are not changed from the defaults. Google now supports
9509 these NTP servers officially. We still recommend downstreams to
9510 properly register an NTP pool with the NTP pool project though.
9512 * coredumpctl gained a new "--reverse" option for printing the list
9513 of coredumps in reverse order.
9515 * coredumpctl will now show additional information about truncated and
9516 inaccessible coredumps, as well as coredumps that are still being
9517 processed. It also gained a new --quiet switch for suppressing
9518 additional informational message in its output.
9520 * coredumpctl gained support for only showing coredumps newer and/or
9521 older than specific timestamps, using the new --since= and --until=
9522 options, reminiscent of journalctl's options by the same name.
9524 * The systemd-coredump logic has been improved so that it may be reused
9525 to collect backtraces in non-compiled languages, for example in
9526 scripting languages such as Python.
9528 * machinectl will now show the UID shift of local containers, if user
9529 namespacing is enabled for them.
9531 * systemd will now optionally run "environment generator" binaries at
9532 configuration load time. They may be used to add environment
9533 variables to the environment block passed to services invoked. One
9534 user environment generator is shipped by default that sets up
9535 environment variables based on files dropped into /etc/environment.d
9536 and ~/.config/environment.d/.
9538 * systemd-resolved now includes the new, recently published 2017 DNSSEC
9541 * hostnamed has been updated to report a new chassis type of
9542 "convertible" to cover "foldable" laptops that can both act as a
9543 tablet and as a laptop, such as various Lenovo Yoga devices.
9545 Contributions from: Adrián López, Alexander Galanin, Alexander
9546 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
9547 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
9548 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
9549 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
9550 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
9551 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
9552 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
9553 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
9554 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
9555 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
9556 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
9557 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
9558 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
9559 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
9560 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
9561 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
9562 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
9563 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
9564 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
9565 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
9566 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
9567 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
9568 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
9569 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
9570 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
9571 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
9572 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
9575 — Berlin, 2017-03-01
9579 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
9580 RestrictAddressFamilies= enabled. These sandboxing options should
9581 generally be compatible with the various external udev call-out
9582 binaries we are aware of, however there may be exceptions, in
9583 particular when exotic languages for these call-outs are used. In
9584 this case, consider turning off these settings locally.
9586 * The new RemoveIPC= option can be used to remove IPC objects owned by
9587 the user or group of a service when that service exits.
9589 * The new ProtectKernelModules= option can be used to disable explicit
9590 load and unload operations of kernel modules by a service. In
9591 addition access to /usr/lib/modules is removed if this option is set.
9593 * ProtectSystem= option gained a new value "strict", which causes the
9594 whole file system tree with the exception of /dev, /proc, and /sys,
9595 to be remounted read-only for a service.
9597 * The new ProtectKernelTunables= option can be used to disable
9598 modification of configuration files in /sys and /proc by a service.
9599 Various directories and files are remounted read-only, so access is
9600 restricted even if the file permissions would allow it.
9602 * The new ProtectControlGroups= option can be used to disable write
9603 access by a service to /sys/fs/cgroup.
9605 * Various systemd services have been hardened with
9606 ProtectKernelTunables=yes, ProtectControlGroups=yes,
9607 RestrictAddressFamilies=.
9609 * Support for dynamically creating users for the lifetime of a service
9610 has been added. If DynamicUser=yes is specified, user and group IDs
9611 will be allocated from the range 61184…65519 for the lifetime of the
9612 service. They can be resolved using the new nss-systemd.so NSS
9613 module. The module must be enabled in /etc/nsswitch.conf. Services
9614 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
9615 any resources allocated by the service will be cleaned up when the
9616 service exits. They also have ProtectHome=read-only and
9617 ProtectSystem=strict enabled, so they are not able to make any
9618 permanent modifications to the system.
9620 * The nss-systemd module also always resolves root and nobody, making
9621 it possible to have no /etc/passwd or /etc/group files in minimal
9622 container or chroot environments.
9624 * Services may be started with their own user namespace using the new
9625 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
9626 under which the service is running are mapped. All other users are
9629 * Support for the cgroup namespace has been added to systemd-nspawn. If
9630 supported by kernel, the container system started by systemd-nspawn
9631 will have its own view of the cgroup hierarchy. This new behaviour
9632 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
9634 * The new MemorySwapMax= option can be used to limit the maximum swap
9635 usage under the unified cgroup hierarchy.
9637 * Support for the CPU controller in the unified cgroup hierarchy has
9638 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
9639 options. This controller requires out-of-tree patches for the kernel
9640 and the support is provisional.
9642 * Mount and automount units may now be created transiently
9643 (i.e. dynamically at runtime via the bus API, instead of requiring
9644 unit files in the file system).
9646 * systemd-mount is a new tool which may mount file systems – much like
9647 mount(8), optionally pulling in additional dependencies through
9648 transient .mount and .automount units. For example, this tool
9649 automatically runs fsck on a backing block device before mounting,
9650 and allows the automount logic to be used dynamically from the
9651 command line for establishing mount points. This tool is particularly
9652 useful when dealing with removable media, as it will ensure fsck is
9653 run – if necessary – before the first access and that the file system
9654 is quickly unmounted after each access by utilizing the automount
9655 logic. This maximizes the chance that the file system on the
9656 removable media stays in a clean state, and if it isn't in a clean
9657 state is fixed automatically.
9659 * LazyUnmount=yes option for mount units has been added to expose the
9660 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
9663 * /efi will be used as the mount point of the EFI boot partition, if
9664 the directory is present, and the mount point was not configured
9665 through other means (e.g. fstab). If /efi directory does not exist,
9666 /boot will be used as before. This makes it easier to automatically
9667 mount the EFI partition on systems where /boot is used for something
9670 * When operating on GPT disk images for containers, systemd-nspawn will
9671 now mount the ESP to /boot or /efi according to the same rules as PID
9672 1 running on a host. This allows tools like "bootctl" to operate
9673 correctly within such containers, in order to make container images
9674 bootable on physical systems.
9676 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
9678 * Two new user session targets have been added to support running
9679 graphical sessions under the systemd --user instance:
9680 graphical-session.target and graphical-session-pre.target. See
9681 systemd.special(7) for a description of how those targets should be
9684 * The vconsole initialization code has been significantly reworked to
9685 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
9686 support unicode keymaps. Font and keymap configuration will now be
9687 copied to all allocated virtual consoles.
9689 * FreeBSD's bhyve virtualization is now detected.
9691 * Information recorded in the journal for core dumps now includes the
9692 contents of /proc/mountinfo and the command line of the process at
9693 the top of the process hierarchy (which is usually the init process
9696 * systemd-journal-gatewayd learned the --directory= option to serve
9697 files from the specified location.
9699 * journalctl --root=… can be used to peruse the journal in the
9700 /var/log/ directories inside of a container tree. This is similar to
9701 the existing --machine= option, but does not require the container to
9704 * The hardware database has been extended to support
9705 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
9708 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
9709 specify the click rate for mice which include a horizontal wheel with
9710 a click rate that is different than the one for the vertical wheel.
9712 * systemd-run gained a new --wait option that makes service execution
9713 synchronous. (Specifically, the command will not return until the
9714 specified service binary exited.)
9716 * systemctl gained a new --wait option that causes the start command to
9717 wait until the units being started have terminated again.
9719 * A new journal output mode "short-full" has been added which displays
9720 timestamps with abbreviated English day names and adds a timezone
9721 suffix. Those timestamps include more information than the default
9722 "short" output mode, and can be passed directly to journalctl's
9723 --since= and --until= options.
9725 * /etc/resolv.conf will be bind-mounted into containers started by
9726 systemd-nspawn, if possible, so any changes to resolv.conf contents
9727 are automatically propagated to the container.
9729 * The number of instances for socket-activated services originating
9730 from a single IP address can be limited with
9731 MaxConnectionsPerSource=, extending the existing setting of
9734 * systemd-networkd gained support for vcan ("Virtual CAN") interface
9737 * .netdev and .network configuration can now be extended through
9740 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
9741 Segmentation Offload, Generic Receive Offload, Large Receive Offload
9742 can be enabled and disabled using the new UDPSegmentationOffload=,
9743 TCPSegmentationOffload=, GenericSegmentationOffload=,
9744 GenericReceiveOffload=, LargeReceiveOffload= options in the
9745 [Link] section of .link files.
9747 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
9748 Port VLAN ID can be configured for bridge devices using the new STP=,
9749 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
9750 section of .netdev files.
9752 * The route table to which routes received over DHCP or RA should be
9753 added can be configured with the new RouteTable= option in the [DHCP]
9754 and [IPv6AcceptRA] sections of .network files.
9756 * The Address Resolution Protocol can be disabled on links managed by
9757 systemd-networkd using the ARP=no setting in the [Link] section of
9760 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
9761 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
9762 encode information about the result and exit codes of the current
9763 service runtime cycle.
9765 * systemd-sysctl will now configure kernel parameters in the order
9766 they occur in the configuration files. This matches what sysctl
9767 has been traditionally doing.
9769 * kernel-install "plugins" that are executed to perform various
9770 tasks after a new kernel is added and before an old one is removed
9771 can now return a special value to terminate the procedure and
9772 prevent any later plugins from running.
9774 * Journald's SplitMode=login setting has been deprecated. It has been
9775 removed from documentation, and its use is discouraged. In a future
9776 release it will be completely removed, and made equivalent to current
9777 default of SplitMode=uid.
9779 * Storage=both option setting in /etc/systemd/coredump.conf has been
9780 removed. With fast LZ4 compression storing the core dump twice is not
9783 * The --share-system systemd-nspawn option has been replaced with an
9784 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
9785 this functionality is discouraged. In addition the variables
9786 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
9787 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
9788 individual namespaces.
9790 * "machinectl list" now shows the IP address of running containers in
9791 the output, as well as OS release information.
9793 * "loginctl list" now shows the TTY of each session in the output.
9795 * sd-bus gained new API calls sd_bus_track_set_recursive(),
9796 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
9797 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
9798 tracking objects in a "recursive" mode, where a single client can be
9799 counted multiple times, if it takes multiple references.
9801 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
9802 sd_bus_get_exit_on_disconnect(). They may be used to make a
9803 process using sd-bus automatically exit if the bus connection is
9806 * Bus clients of the service manager may now "pin" loaded units into
9807 memory, by taking an explicit reference on them. This is useful to
9808 ensure the client can retrieve runtime data about the service even
9809 after the service completed execution. Taking such a reference is
9810 available only for privileged clients and should be helpful to watch
9811 running services in a race-free manner, and in particular collect
9812 information about exit statuses and results.
9814 * The nss-resolve module has been changed to strictly return UNAVAIL
9815 when communication via D-Bus with resolved failed, and NOTFOUND when
9816 a lookup completed but was negative. This means it is now possible to
9817 neatly configure fallbacks using nsswitch.conf result checking
9818 expressions. Taking benefit of this, the new recommended
9819 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
9821 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
9823 * A new setting CtrlAltDelBurstAction= has been added to
9824 /etc/systemd/system.conf which may be used to configure the precise
9825 behaviour if the user on the console presses Ctrl-Alt-Del more often
9826 than 7 times in 2s. Previously this would unconditionally result in
9827 an expedited, immediate reboot. With this new setting the precise
9828 operation may be configured in more detail, and also turned off
9831 * In .netdev files two new settings RemoteChecksumTx= and
9832 RemoteChecksumRx= are now understood that permit configuring the
9833 remote checksumming logic for VXLAN networks.
9835 * The service manager learnt a new "invocation ID" concept for invoked
9836 services. Each runtime cycle of a service will get a new invocation
9837 ID (a 128-bit random UUID) assigned that identifies the current
9838 run of the service uniquely and globally. A new invocation ID
9839 is generated each time a service starts up. The journal will store
9840 the invocation ID of a service along with any logged messages, thus
9841 making the invocation ID useful for matching the online runtime of a
9842 service with the offline log data it generated in a safe way without
9843 relying on synchronized timestamps. In many ways this new service
9844 invocation ID concept is similar to the kernel's boot ID concept that
9845 uniquely and globally identifies the runtime of each boot. The
9846 invocation ID of a service is passed to the service itself via an
9847 environment variable ($INVOCATION_ID). A new bus call
9848 GetUnitByInvocationID() has been added that is similar to GetUnit()
9849 but instead of retrieving the bus path for a unit by its name
9850 retrieves it by its invocation ID. The returned path is valid only as
9851 long as the passed invocation ID is current.
9853 * systemd-resolved gained a new "DNSStubListener" setting in
9854 resolved.conf. It either takes a boolean value or the special values
9855 "udp" and "tcp", and configures whether to enable the stub DNS
9856 listener on 127.0.0.53:53.
9858 * IP addresses configured via networkd may now carry additional
9859 configuration settings supported by the kernel. New options include:
9860 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
9861 PrefixRoute=, AutoJoin=.
9863 * The PAM configuration fragment file for "user@.service" shipped with
9864 systemd (i.e. the --user instance of systemd) has been stripped to
9865 the minimum necessary to make the system boot. Previously, it
9866 contained Fedora-specific stanzas that did not apply to other
9867 distributions. It is expected that downstream distributions add
9868 additional configuration lines, matching their needs to this file,
9869 using it only as rough template of what systemd itself needs. Note
9870 that this reduced fragment does not even include an invocation of
9871 pam_limits which most distributions probably want to add, even though
9872 systemd itself does not need it. (There's also the new build time
9873 option --with-pamconfdir=no to disable installation of the PAM
9876 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
9877 capability is now also dropped from its set (in addition to
9878 CAP_SYS_MKNOD as before).
9880 * In service unit files it is now possible to connect a specific named
9881 file descriptor with stdin/stdout/stdout of an executed service. The
9882 name may be specified in matching .socket units using the
9883 FileDescriptorName= setting.
9885 * A number of journal settings may now be configured on the kernel
9886 command line. Specifically, the following options are now understood:
9887 systemd.journald.max_level_console=,
9888 systemd.journald.max_level_store=,
9889 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
9890 systemd.journald.max_level_wall=.
9892 * "systemctl is-enabled --full" will now show by which symlinks a unit
9893 file is enabled in the unit dependency tree.
9895 * Support for VeraCrypt encrypted partitions has been added to the
9896 "cryptsetup" logic and /etc/crypttab.
9898 * systemd-detect-virt gained support for a new --private-users switch
9899 that checks whether the invoking processes are running inside a user
9900 namespace. Similar, a new special value "private-users" for the
9901 existing ConditionVirtualization= setting has been added, permitting
9902 skipping of specific units in user namespace environments.
9904 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
9905 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
9906 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
9907 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
9908 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
9909 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
9910 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
9911 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
9912 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
9913 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
9914 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
9915 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
9916 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
9917 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
9918 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
9919 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
9920 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
9921 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
9922 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
9923 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
9924 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
9925 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
9926 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
9927 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
9928 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
9929 Jędrzejewski-Szmek, Zeal Jagannatha
9931 — Santa Fe, 2016-11-03
9935 * In service units the various ExecXYZ= settings have been extended
9936 with an additional special character as first argument of the
9937 assigned value: if the character '+' is used the specified command
9938 line it will be run with full privileges, regardless of User=,
9939 Group=, CapabilityBoundingSet= and similar options. The effect is
9940 similar to the existing PermissionsStartOnly= option, but allows
9941 configuration of this concept for each executed command line
9944 * Services may now alter the service watchdog timeout at runtime by
9945 sending a WATCHDOG_USEC= message via sd_notify().
9947 * MemoryLimit= and related unit settings now optionally take percentage
9948 specifications. The percentage is taken relative to the amount of
9949 physical memory in the system (or in case of containers, the assigned
9950 amount of memory). This allows scaling service resources neatly with
9951 the amount of RAM available on the system. Similarly, systemd-logind's
9952 RuntimeDirectorySize= option now also optionally takes percentage
9955 * In similar fashion TasksMax= takes percentage values now, too. The
9956 value is taken relative to the configured maximum number of processes
9957 on the system. The per-service task maximum has been changed to 15%
9958 using this functionality. (Effectively this is an increase of 512 →
9959 4915 for service units, given the kernel's default pid_max setting.)
9961 * Calendar time specifications in .timer units now understand a ".."
9962 syntax for time ranges. Example: "4..7:10" may now be used for
9963 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
9966 * The InaccessableDirectories=, ReadOnlyDirectories= and
9967 ReadWriteDirectories= unit file settings have been renamed to
9968 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
9969 applied to all kinds of file nodes, and not just directories, with
9970 the exception of symlinks. Specifically these settings may now be
9971 used on block and character device nodes, UNIX sockets and FIFOS as
9972 well as regular files. The old names of these settings remain
9973 available for compatibility.
9975 * systemd will now log about all service processes it kills forcibly
9976 (using SIGKILL) because they remained after the clean shutdown phase
9977 of the service completed. This should help identifying services that
9978 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
9979 systemd-logind's configuration a similar log message is generated for
9980 processes killed at the end of each session due to this setting.
9982 * systemd will now set the $JOURNAL_STREAM environment variable for all
9983 services whose stdout/stderr are connected to the Journal (which
9984 effectively means by default: all services). The variable contains
9985 the device and inode number of the file descriptor used for
9986 stdout/stderr. This may be used by invoked programs to detect whether
9987 their stdout/stderr is connected to the Journal, in which case they
9988 can switch over to direct Journal communication, thus being able to
9989 pass extended, structured metadata along with their log messages. As
9990 one example, this is now used by glib's logging primitives.
9992 * When using systemd's default tmp.mount unit for /tmp, the mount point
9993 will now be established with the "nosuid" and "nodev" options. This
9994 avoids privilege escalation attacks that put traps and exploits into
9995 /tmp. However, this might cause problems if you e.g. put container
9996 images or overlays into /tmp; if you need this, override tmp.mount's
9997 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
10000 * systemd now supports the "memory" cgroup controller also on
10003 * The systemd-cgtop tool now optionally takes a control group path as
10004 command line argument. If specified, the control group list shown is
10005 limited to subgroups of that group.
10007 * The SystemCallFilter= unit file setting gained support for
10008 pre-defined, named system call filter sets. For example
10009 SystemCallFilter=@clock is now an effective way to make all clock
10010 changing-related system calls unavailable to a service. A number of
10011 similar pre-defined groups are defined. Writing system call filters
10012 for system services is simplified substantially with this new
10013 concept. Accordingly, all of systemd's own, long-running services now
10014 enable system call filtering based on this, by default.
10016 * A new service setting MemoryDenyWriteExecute= has been added, taking
10017 a boolean value. If turned on, a service may no longer create memory
10018 mappings that are writable and executable at the same time. This
10019 enhances security for services where this is enabled as it becomes
10020 harder to dynamically write and then execute memory in exploited
10021 service processes. This option has been enabled for all of systemd's
10022 own long-running services.
10024 * A new RestrictRealtime= service setting has been added, taking a
10025 boolean argument. If set the service's processes may no longer
10026 acquire realtime scheduling. This improves security as realtime
10027 scheduling may otherwise be used to easily freeze the system.
10029 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
10030 value. This may be used for requesting that the system manager inside
10031 of the container reports start-up completion to nspawn which then
10032 propagates this notification further to the service manager
10033 supervising nspawn itself. A related option NotifyReady= in .nspawn
10034 files has been added too. This functionality allows ordering of the
10035 start-up of multiple containers using the usual systemd ordering
10038 * machinectl gained a new command "stop" that is an alias for
10041 * systemd-resolved gained support for contacting DNS servers on
10042 link-local IPv6 addresses.
10044 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
10045 its caches. A method call for requesting the same operation has been
10046 added to the bus API too, and is made available via "systemd-resolve
10049 * systemd-resolve gained a new --status switch. If passed a brief
10050 summary of the used DNS configuration with per-interface information
10053 * resolved.conf gained a new Cache= boolean option, defaulting to
10054 on. If turned off local DNS caching is disabled. This comes with a
10055 performance penalty in particular when DNSSEC is enabled. Note that
10056 resolved disables its internal caching implicitly anyway, when the
10057 configured DNS server is on a host-local IP address such as ::1 or
10058 127.0.0.1, thus automatically avoiding double local caching.
10060 * systemd-resolved now listens on the local IP address 127.0.0.53:53
10061 for DNS requests. This improves compatibility with local programs
10062 that do not use the libc NSS or systemd-resolved's bus APIs for name
10063 resolution. This minimal DNS service is only available to local
10064 programs and does not implement the full DNS protocol, but enough to
10065 cover local DNS clients. A new, static resolv.conf file, listing just
10066 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
10067 now recommended to make /etc/resolv.conf a symlink to this file in
10068 order to route all DNS lookups to systemd-resolved, regardless if
10069 done via NSS, the bus API or raw DNS packets. Note that this local
10070 DNS service is not as fully featured as the libc NSS or
10071 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
10072 used to deliver link-local address information (as this implies
10073 sending a local interface index along), LLMNR/mDNS support via this
10074 interface is severely restricted. It is thus strongly recommended for
10075 all applications to use the libc NSS API or native systemd-resolved
10078 * systemd-networkd's bridge support learned a new setting
10079 VLANFiltering= for controlling VLAN filtering. Moreover a new section
10080 in .network files has been added for configuring VLAN bridging in
10081 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
10083 * systemd-networkd's IPv6 Router Advertisement code now makes use of
10084 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
10085 now be acquired without relying on DHCPv6. Two new options
10086 UseDomains= and UseDNS= have been added to configure this behaviour.
10088 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
10089 renamed IPv6AcceptRA=, without altering its behaviour. The old
10090 setting name remains available for compatibility reasons.
10092 * The systemd-networkd VTI/VTI6 tunneling support gained new options
10093 Key=, InputKey= and OutputKey=.
10095 * systemd-networkd gained support for VRF ("Virtual Routing Function")
10096 interface configuration.
10098 * "systemctl edit" may now be used to create new unit files by
10099 specifying the --force switch.
10101 * sd-event gained a new function sd_event_get_iteration() for
10102 requesting the current iteration counter of the event loop. It starts
10103 at zero and is increased by one with each event loop iteration.
10105 * A new rpm macro %systemd_ordering is provided by the macros.systemd
10106 file. It can be used in lieu of %systemd_requires in packages which
10107 don't use any systemd functionality and are intended to be installed
10108 in minimal containers without systemd present. This macro provides
10109 ordering dependencies to ensure that if the package is installed in
10110 the same rpm transaction as systemd, systemd will be installed before
10111 the scriptlets for the package are executed, allowing unit presets
10114 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
10115 been added to simplify packaging of generators.
10117 * The os-release file gained VERSION_CODENAME field for the
10118 distribution nickname (e.g. VERSION_CODENAME=woody).
10120 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
10121 can be set to disable parsing of metadata and the creation
10122 of persistent symlinks for that device.
10124 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
10125 to make them available to logged-in users has been reverted.
10127 * Much of the common code of the various systemd components is now
10128 built into an internal shared library libsystemd-shared-231.so
10129 (incorporating the systemd version number in the name, to be updated
10130 with future releases) that the components link to. This should
10131 decrease systemd footprint both in memory during runtime and on
10132 disk. Note that the shared library is not for public use, and is
10133 neither API nor ABI stable, but is likely to change with every new
10134 released update. Packagers need to make sure that binaries
10135 linking to libsystemd-shared.so are updated in step with the
10138 * Configuration for "mkosi" is now part of the systemd
10139 repository. mkosi is a tool to easily build legacy-free OS images,
10140 and is available on github: https://github.com/systemd/mkosi. If
10141 "mkosi" is invoked in the build tree a new raw OS image is generated
10142 incorporating the systemd sources currently being worked on and a
10143 clean, fresh distribution installation. The generated OS image may be
10144 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
10145 UEFI PC. This functionality is particularly useful to easily test
10146 local changes made to systemd in a pristine, defined environment. See
10147 doc/HACKING for details.
10149 * configure learned the --with-support-url= option to specify the
10150 distribution's bugtracker.
10152 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
10153 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
10154 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
10155 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
10156 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
10157 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
10158 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
10159 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
10160 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
10161 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
10162 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
10163 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
10164 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
10165 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
10166 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
10167 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
10168 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
10169 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
10170 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
10172 — Berlin, 2016-07-25
10176 * DNSSEC is now turned on by default in systemd-resolved (in
10177 "allow-downgrade" mode), but may be turned off during compile time by
10178 passing "--with-default-dnssec=no" to "configure" (and of course,
10179 during runtime with DNSSEC= in resolved.conf). We recommend
10180 downstreams to leave this on at least during development cycles and
10181 report any issues with the DNSSEC logic upstream. We are very
10182 interested in collecting feedback about the DNSSEC validator and its
10183 limitations in the wild. Note however, that DNSSEC support is
10184 probably nothing downstreams should turn on in stable distros just
10185 yet, as it might create incompatibilities with a few DNS servers and
10186 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
10187 automatically whenever we detect such incompatible setups, but there
10188 might be systems we do not cover yet. Hence: please help us testing
10189 the DNSSEC code, leave this on where you can, report back, but then
10190 again don't consider turning this on in your stable, LTS or
10191 production release just yet. (Note that you have to enable
10192 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
10193 and its DNSSEC mode for hostname resolution from local
10196 * systemd-resolve conveniently resolves DANE records with the --tlsa
10197 option and OPENPGPKEY records with the --openpgp option. It also
10198 supports dumping raw DNS record data via the new --raw= switch.
10200 * systemd-logind will now by default terminate user processes that are
10201 part of the user session scope unit (session-XX.scope) when the user
10202 logs out. This behavior is controlled by the KillUserProcesses=
10203 setting in logind.conf, and the previous default of "no" is now
10204 changed to "yes". This means that user sessions will be properly
10205 cleaned up after, but additional steps are necessary to allow
10206 intentionally long-running processes to survive logout.
10208 While the user is logged in at least once, user@.service is running,
10209 and any service that should survive the end of any individual login
10210 session can be started at a user service or scope using systemd-run.
10211 systemd-run(1) man page has been extended with an example which shows
10212 how to run screen in a scope unit underneath user@.service. The same
10213 command works for tmux.
10215 After the user logs out of all sessions, user@.service will be
10216 terminated too, by default, unless the user has "lingering" enabled.
10217 To effectively allow users to run long-term tasks even if they are
10218 logged out, lingering must be enabled for them. See loginctl(1) for
10219 details. The default polkit policy was modified to allow users to
10220 set lingering for themselves without authentication.
10222 Previous defaults can be restored at compile time by the
10223 --without-kill-user-processes option to "configure".
10225 * systemd-logind gained new configuration settings SessionsMax= and
10226 InhibitorsMax=, both with a default of 8192. It will not register new
10227 user sessions or inhibitors above this limit.
10229 * systemd-logind will now reload configuration on SIGHUP.
10231 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
10232 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
10233 enable. Also, support for the "io" cgroup controller in the unified
10234 hierarchy has been added, so that the "memory", "pids" and "io" are
10235 now the controllers that are supported on the unified hierarchy.
10237 WARNING: it is not possible to use previous systemd versions with
10238 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
10239 is necessary to also update systemd in the initramfs if using the
10240 unified hierarchy. An updated SELinux policy is also required.
10242 * LLDP support has been extended, and both passive (receive-only) and
10243 active (sender) modes are supported. Passive mode ("routers-only") is
10244 enabled by default in systemd-networkd. Active LLDP mode is enabled
10245 by default for containers on the internal network. The "networkctl
10246 lldp" command may be used to list information gathered. "networkctl
10247 status" will also show basic LLDP information on connected peers now.
10249 * The IAID and DUID unique identifier sent in DHCP requests may now be
10250 configured for the system and each .network file managed by
10251 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
10253 * systemd-networkd gained support for configuring proxy ARP support for
10254 each interface, via the ProxyArp= setting in .network files. It also
10255 gained support for configuring the multicast querier feature of
10256 bridge devices, via the new MulticastQuerier= setting in .netdev
10257 files. Similarly, snooping on the IGMP traffic can be controlled
10258 via the new setting MulticastSnooping=.
10260 A new setting PreferredLifetime= has been added for addresses
10261 configured in .network file to configure the lifetime intended for an
10264 The systemd-networkd DHCP server gained the option EmitRouter=, which
10265 defaults to yes, to configure whether the DHCP Option 3 (Router)
10268 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
10269 systemd-socket-activate and installed into /usr/bin. It is now fully
10272 * systemd-journald now uses separate threads to flush changes to disk
10273 when closing journal files, thus reducing impact of slow disk I/O on
10274 logging performance.
10276 * The sd-journal API gained two new calls
10277 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
10278 can be used to open journal files using file descriptors instead of
10279 file or directory paths. sd_journal_open_container() has been
10280 deprecated, sd_journal_open_directory_fd() should be used instead
10281 with the flag SD_JOURNAL_OS_ROOT.
10283 * journalctl learned a new output mode "-o short-unix" that outputs log
10284 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
10285 UTC). It also gained support for a new --no-hostname setting to
10286 suppress the hostname column in the family of "short" output modes.
10288 * systemd-ask-password now optionally skips printing of the password to
10289 stdout with --no-output which can be useful in scripts.
10291 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
10292 (devices tagged with ID_MAKER_TOOL) are now tagged with
10293 "uaccess" and are available to logged in users.
10295 * The DeviceAllow= unit setting now supports specifiers (with "%").
10297 * "systemctl show" gained a new --value switch, which allows print a
10298 only the contents of a specific unit property, without also printing
10299 the property's name. Similar support was added to "show*" verbs
10300 of loginctl and machinectl that output "key=value" lists.
10302 * A new unit type "generated" was added for files dynamically generated
10303 by generator tools. Similarly, a new unit type "transient" is used
10304 for unit files created using the runtime API. "systemctl enable" will
10305 refuse to operate on such files.
10307 * A new command "systemctl revert" has been added that may be used to
10308 revert to the vendor version of a unit file, in case local changes
10309 have been made by adding drop-ins or overriding the unit file.
10311 * "machinectl clean" gained a new verb to automatically remove all or
10312 just hidden container images.
10314 * systemd-tmpfiles gained support for a new line type "e" for emptying
10315 directories, if they exist, without creating them if they don't.
10317 * systemd-nspawn gained support for automatically patching the UID/GIDs
10318 of the owners and the ACLs of all files and directories in a
10319 container tree to match the UID/GID user namespacing range selected
10320 for the container invocation. This mode is enabled via the new
10321 --private-users-chown switch. It also gained support for
10322 automatically choosing a free, previously unused UID/GID range when
10323 starting a container, via the new --private-users=pick setting (which
10324 implies --private-users-chown). Together, these options for the first
10325 time make user namespacing for nspawn containers fully automatic and
10326 thus deployable. The systemd-nspawn@.service template unit file has
10327 been changed to use this functionality by default.
10329 * systemd-nspawn gained a new --network-zone= switch, that allows
10330 creating ad-hoc virtual Ethernet links between multiple containers,
10331 that only exist as long as at least one container referencing them is
10332 running. This allows easy connecting of multiple containers with a
10333 common link that implements an Ethernet broadcast domain. Each of
10334 these network "zones" may be named relatively freely by the user, and
10335 may be referenced by any number of containers, but each container may
10336 only reference one of these "zones". On the lower level, this is
10337 implemented by an automatically managed bridge network interface for
10338 each zone, that is created when the first container referencing its
10339 zone is created and removed when the last one referencing its zone
10342 * The default start timeout may now be configured on the kernel command
10343 line via systemd.default_timeout_start_sec=. It was already
10344 configurable via the DefaultTimeoutStartSec= option in
10345 /etc/systemd/system.conf.
10347 * Socket units gained a new TriggerLimitIntervalSec= and
10348 TriggerLimitBurst= setting to configure a limit on the activation
10349 rate of the socket unit.
10351 * The LimitNICE= setting now optionally takes normal UNIX nice values
10352 in addition to the raw integer limit value. If the specified
10353 parameter is prefixed with "+" or "-" and is in the range -20…19 the
10354 value is understood as UNIX nice value. If not prefixed like this it
10355 is understood as raw RLIMIT_NICE limit.
10357 * Note that the effect of the PrivateDevices= unit file setting changed
10358 slightly with this release: the per-device /dev file system will be
10359 mounted read-only from this version on, and will have "noexec"
10360 set. This (minor) change of behavior might cause some (exceptional)
10361 legacy software to break, when PrivateDevices=yes is set for its
10362 service. Please leave PrivateDevices= off if you run into problems
10365 * systemd-bootchart has been split out to a separate repository:
10366 https://github.com/systemd/systemd-bootchart
10368 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
10369 merged into the kernel in its current form.
10371 * The compatibility libraries libsystemd-daemon.so,
10372 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
10373 which have been deprecated since systemd-209 have been removed along
10374 with the corresponding pkg-config files. All symbols provided by
10375 those libraries are provided by libsystemd.so.
10377 * The Capabilities= unit file setting has been removed (it is ignored
10378 for backwards compatibility). AmbientCapabilities= and
10379 CapabilityBoundingSet= should be used instead.
10381 * A new special target has been added, initrd-root-device.target,
10382 which creates a synchronization point for dependencies of the root
10383 device in early userspace. Initramfs builders must ensure that this
10384 target is now included in early userspace.
10386 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
10387 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
10388 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
10389 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
10390 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
10391 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
10392 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
10393 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
10394 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
10395 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
10396 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
10397 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
10398 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
10399 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
10400 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
10401 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
10402 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
10403 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
10404 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
10405 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
10406 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
10407 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
10408 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
10409 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
10410 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
10413 — Fairfax, 2016-05-21
10417 * The systemd-resolved DNS resolver service has gained a substantial
10418 set of new features, most prominently it may now act as a DNSSEC
10419 validating stub resolver. DNSSEC mode is currently turned off by
10420 default, but is expected to be turned on by default in one of the
10421 next releases. For now, we invite everybody to test the DNSSEC logic
10422 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
10423 service also gained a full set of D-Bus interfaces, including calls
10424 to configure DNS and DNSSEC settings per link (for use by external
10425 network management software). systemd-resolved and systemd-networkd
10426 now distinguish between "search" and "routing" domains. The former
10427 are used to qualify single-label names, the latter are used purely
10428 for routing lookups within certain domains to specific links.
10429 resolved now also synthesizes RRs for all entries from /etc/hosts.
10431 * The systemd-resolve tool (which is a client utility for
10432 systemd-resolved) has been improved considerably and is now fully
10433 supported and documented. Hence it has moved from /usr/lib/systemd to
10436 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
10439 * The coredump collection logic has been reworked: when a coredump is
10440 collected it is now written to disk, compressed and processed
10441 (including stacktrace extraction) from a new instantiated service
10442 systemd-coredump@.service, instead of directly from the
10443 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
10444 processing large coredumps can take up a substantial amount of
10445 resources and time, and this previously happened entirely outside of
10446 systemd's service supervision. With the new logic the core_pattern
10447 hook only does minimal metadata collection before passing off control
10448 to the new instantiated service, which is configured with a time
10449 limit, a nice level and other settings to minimize negative impact on
10450 the rest of the system. Also note that the new logic will honour the
10451 RLIMIT_CORE setting of the crashed process, which now allows users
10452 and processes to turn off coredumping for their processes by setting
10455 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
10456 and all forked processes by default. Previously, PID 1 would leave
10457 the setting at "0" for all processes, as set by the kernel. Note that
10458 the resource limit traditionally has no effect on the generated
10459 coredumps on the system if the /proc/sys/kernel/core_pattern hook
10460 logic is used. Since the limit is now honoured (see above) its
10461 default has been changed so that the coredumping logic is enabled by
10462 default for all processes, while allowing specific opt-out.
10464 * When the stacktrace is extracted from processes of system users, this
10465 is now done as "systemd-coredump" user, in order to sandbox this
10466 potentially security sensitive parsing operation. (Note that when
10467 processing coredumps of normal users this is done under the user ID
10468 of process that crashed, as before.) Packagers should take notice
10469 that it is now necessary to create the "systemd-coredump" system user
10470 and group at package installation time.
10472 * The systemd-activate socket activation testing tool gained support
10473 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
10474 and --seqpacket switches. It also has been extended to support both
10475 new-style and inetd-style file descriptor passing. Use the new
10476 --inetd switch to request inetd-style file descriptor passing.
10478 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
10479 variable, which takes a boolean value. If set to false, ANSI color
10480 output is disabled in the tools even when run on a terminal that
10483 * The VXLAN support in networkd now supports two new settings
10484 DestinationPort= and PortRange=.
10486 * A new systemd.machine_id= kernel command line switch has been added,
10487 that may be used to set the machine ID in /etc/machine-id if it is
10488 not initialized yet. This command line option has no effect if the
10489 file is already initialized.
10491 * systemd-nspawn gained a new --as-pid2 switch that invokes any
10492 specified command line as PID 2 rather than PID 1 in the
10493 container. In this mode PID 1 is a minimal stub init process that
10494 implements the special POSIX and Linux semantics of PID 1 regarding
10495 signal and child process management. Note that this stub init process
10496 is implemented in nspawn itself and requires no support from the
10497 container image. This new logic is useful to support running
10498 arbitrary commands in the container, as normal processes are
10499 generally not prepared to run as PID 1.
10501 * systemd-nspawn gained a new --chdir= switch for setting the current
10502 working directory for the process started in the container.
10504 * "journalctl /dev/sda" will now output all kernel log messages for
10505 specified device from the current boot, in addition to all devices
10506 that are parents of it. This should make log output about devices
10507 pretty useful, as long as kernel drivers attach enough metadata to
10508 the log messages. (The usual SATA drivers do.)
10510 * The sd-journal API gained two new calls
10511 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
10512 that report whether log data from /run or /var has been found.
10514 * journalctl gained a new switch "--fields" that prints all journal
10515 record field names currently in use in the journal. This is backed
10516 by two new sd-journal API calls sd_journal_enumerate_fields() and
10517 sd_journal_restart_fields().
10519 * Most configurable timeouts in systemd now expect an argument of
10520 "infinity" to turn them off, instead of "0" as before. The semantics
10521 from now on is that a timeout of "0" means "now", and "infinity"
10522 means "never". To maintain backwards compatibility, "0" continues to
10523 turn off previously existing timeout settings.
10525 * "systemctl reload-or-try-restart" has been renamed to "systemctl
10526 try-reload-or-restart" to clarify what it actually does: the "try"
10527 logic applies to both reloading and restarting, not just restarting.
10528 The old name continues to be accepted for compatibility.
10530 * On boot-up, when PID 1 detects that the system clock is behind the
10531 release date of the systemd version in use, the clock is now set
10532 to the latter. Previously, this was already done in timesyncd, in order
10533 to avoid running with clocks set to the various clock epochs such as
10534 1902, 1938 or 1970. With this change the logic is now done in PID 1
10535 in addition to timesyncd during early boot-up, so that it is enforced
10536 before the first process is spawned by systemd. Note that the logic
10537 in timesyncd remains, as it is more comprehensive and ensures
10538 clock monotonicity by maintaining a persistent timestamp file in
10539 /var. Since /var is generally not available in earliest boot or the
10540 initrd, this part of the logic remains in timesyncd, and is not done
10543 * Support for tweaking details in net_cls.class_id through the
10544 NetClass= configuration directive has been removed, as the kernel
10545 people have decided to deprecate that controller in cgroup v2.
10546 Userspace tools such as nftables are moving over to setting rules
10547 that are specific to the full cgroup path of a task, which obsoletes
10548 these controllers anyway. The NetClass= directive is kept around for
10549 legacy compatibility reasons. For a more in-depth description of the
10550 kernel change, please refer to the respective upstream commit:
10552 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
10554 * A new service setting RuntimeMaxSec= has been added that may be used
10555 to specify a maximum runtime for a service. If the timeout is hit, the
10556 service is terminated and put into a failure state.
10558 * A new service setting AmbientCapabilities= has been added. It allows
10559 configuration of additional Linux process capabilities that are
10560 passed to the activated processes. This is only available on very
10563 * The process resource limit settings in service units may now be used
10564 to configure hard and soft limits individually.
10566 * The various libsystemd APIs such as sd-bus or sd-event now publicly
10567 expose support for gcc's __attribute__((cleanup())) C extension.
10568 Specifically, for many object destructor functions alternative
10569 versions have been added that have names suffixed with "p" and take a
10570 pointer to a pointer to the object to destroy, instead of just a
10571 pointer to the object itself. This is useful because these destructor
10572 functions may be used directly as parameters to the cleanup
10573 construct. Internally, systemd has been a heavy user of this GCC
10574 extension for a long time, and with this change similar support is
10575 now available to consumers of the library outside of systemd. Note
10576 that by using this extension in your sources compatibility with old
10577 and strictly ANSI compatible C compilers is lost. However, all gcc or
10578 LLVM versions of recent years support this extension.
10580 * Timer units gained support for a new setting RandomizedDelaySec= that
10581 allows configuring some additional randomized delay to the configured
10582 time. This is useful to spread out timer events to avoid load peaks in
10583 clusters or larger setups.
10585 * Calendar time specifications now support sub-second accuracy.
10587 * Socket units now support listening on SCTP and UDP-lite protocol
10590 * The sd-event API now comes with a full set of man pages.
10592 * Older versions of systemd contained experimental support for
10593 compressing journal files and coredumps with the LZ4 compressor that
10594 was not compatible with the lz4 binary (due to API limitations of the
10595 lz4 library). This support has been removed; only support for files
10596 compatible with the lz4 binary remains. This LZ4 logic is now
10597 officially supported and no longer considered experimental.
10599 * The dkr image import logic has been removed again from importd. dkr's
10600 micro-services focus doesn't fit into the machine image focus of
10601 importd, and quickly got out of date with the upstream dkr API.
10603 * Creation of the /run/lock/lockdev/ directory was dropped from
10604 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
10605 been available for many years. If you still need this, you need to
10606 create your own tmpfiles.d config file with:
10608 d /run/lock/lockdev 0775 root lock -
10610 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
10611 and RebootArgument= have been moved from the [Service] section of
10612 unit files to [Unit], and they are now supported on all unit types,
10613 not just service units. Of course, systemd will continue to
10614 understand these settings also at the old location, in order to
10615 maintain compatibility.
10617 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
10618 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
10619 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
10620 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
10621 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
10622 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
10623 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
10624 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
10625 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
10626 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
10627 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
10628 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
10629 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
10630 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
10631 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
10632 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
10633 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
10634 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
10635 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
10637 — Berlin, 2016-02-11
10641 * A number of properties previously only settable in unit
10642 files are now also available as properties to set when
10643 creating transient units programmatically via the bus, as it
10644 is exposed with systemd-run's --property=
10645 setting. Specifically, these are: SyslogIdentifier=,
10646 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
10647 EnvironmentFile=, ReadWriteDirectories=,
10648 ReadOnlyDirectories=, InaccessibleDirectories=,
10649 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
10651 * When creating transient services via the bus API it is now
10652 possible to pass in a set of file descriptors to use as
10653 STDIN/STDOUT/STDERR for the invoked process.
10655 * Slice units may now be created transiently via the bus APIs,
10656 similar to the way service and scope units may already be
10657 created transiently.
10659 * Wherever systemd expects a calendar timestamp specification
10660 (like in journalctl's --since= and --until= switches) UTC
10661 timestamps are now supported. Timestamps suffixed with "UTC"
10662 are now considered to be in Universal Time Coordinated
10663 instead of the local timezone. Also, timestamps may now
10664 optionally be specified with sub-second accuracy. Both of
10665 these additions also apply to recurring calendar event
10666 specification, such as OnCalendar= in timer units.
10668 * journalctl gained a new "--sync" switch that asks the
10669 journal daemon to write all so far unwritten log messages to
10670 disk and sync the files, before returning.
10672 * systemd-tmpfiles learned two new line types "q" and "Q" that
10673 operate like "v", but also set up a basic btrfs quota
10674 hierarchy when used on a btrfs file system with quota
10677 * tmpfiles' "v", "q" and "Q" will now create a plain directory
10678 instead of a subvolume (even on a btrfs file system) if the
10679 root directory is a plain directory, and not a
10680 subvolume. This should simplify things with certain chroot()
10681 environments which are not aware of the concept of btrfs
10684 * systemd-detect-virt gained a new --chroot switch to detect
10685 whether execution takes place in a chroot() environment.
10687 * CPUAffinity= now takes CPU index ranges in addition to
10688 individual indexes.
10690 * The various memory-related resource limit settings (such as
10691 LimitAS=) now understand the usual K, M, G, … suffixes to
10692 the base of 1024 (IEC). Similar, the time-related resource
10693 limit settings understand the usual min, h, day, … suffixes
10696 * There's a new system.conf setting DefaultTasksMax= to
10697 control the default TasksMax= setting for services and
10698 scopes running on the system. (TasksMax= is the primary
10699 setting that exposes the "pids" cgroup controller on systemd
10700 and was introduced in the previous systemd release.) The
10701 setting now defaults to 512, which means services that are
10702 not explicitly configured otherwise will only be able to
10703 create 512 processes or threads at maximum, from this
10704 version on. Note that this means that thread- or
10705 process-heavy services might need to be reconfigured to set
10706 TasksMax= to a higher value. It is sufficient to set
10707 TasksMax= in these specific unit files to a higher value, or
10708 even "infinity". Similar, there's now a logind.conf setting
10709 UserTasksMax= that defaults to 4096 and limits the total
10710 number of processes or tasks each user may own
10711 concurrently. nspawn containers also have the TasksMax=
10712 value set by default now, to 8192. Note that all of this
10713 only has an effect if the "pids" cgroup controller is
10714 enabled in the kernel. The general benefit of these changes
10715 should be a more robust and safer system, that provides a
10716 certain amount of per-service fork() bomb protection.
10718 * systemd-nspawn gained the new --network-veth-extra= switch
10719 to define additional and arbitrarily-named virtual Ethernet
10720 links between the host and the container.
10722 * A new service execution setting PassEnvironment= has been
10723 added that allows importing select environment variables
10724 from PID1's environment block into the environment block of
10727 * Timer units gained support for a new RemainAfterElapse=
10728 setting which takes a boolean argument. It defaults to on,
10729 exposing behaviour unchanged to previous releases. If set to
10730 off, timer units are unloaded after they elapsed if they
10731 cannot elapse again. This is particularly useful for
10732 transient timer units, which shall not stay around longer
10733 than until they first elapse.
10735 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
10736 default now (the kernel default is 16). This is beneficial
10737 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
10738 allows substantially larger numbers of queued
10739 datagrams. This should increase the capability of systemd to
10740 parallelize boot-up, as logging and sd_notify() are unlikely
10741 to stall execution anymore. If you need to change the value
10742 from the new defaults, use the usual sysctl.d/ snippets.
10744 * The compression framing format used by the journal or
10745 coredump processing has changed to be in line with what the
10746 official LZ4 tools generate. LZ4 compression support in
10747 systemd was considered unsupported previously, as the format
10748 was not compatible with the normal tools. With this release
10749 this has changed now, and it is hence safe for downstream
10750 distributions to turn it on. While not compressing as well
10751 as the XZ, LZ4 is substantially faster, which makes
10752 it a good default choice for the compression logic in the
10753 journal and in coredump handling.
10755 * Any reference to /etc/mtab has been dropped from
10756 systemd. The file has been obsolete since a while, but
10757 systemd refused to work on systems where it was incorrectly
10758 set up (it should be a symlink or non-existent). Please make
10759 sure to update to util-linux 2.27.1 or newer in conjunction
10760 with this systemd release, which also drops any reference to
10761 /etc/mtab. If you maintain a distribution make sure that no
10762 software you package still references it, as this is a
10763 likely source of bugs. There's also a glibc bug pending,
10764 asking for removal of any reference to this obsolete file:
10766 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
10768 Note that only util-linux versions built with
10769 --enable-libmount-force-mountinfo are supported.
10771 * Support for the ".snapshot" unit type has been removed. This
10772 feature turned out to be little useful and little used, and
10773 has now been removed from the core and from systemctl.
10775 * The dependency types RequiresOverridable= and
10776 RequisiteOverridable= have been removed from systemd. They
10777 have been used only very sparingly to our knowledge and
10778 other options that provide a similar effect (such as
10779 systemctl --mode=ignore-dependencies) are much more useful
10780 and commonly used. Moreover, they were only half-way
10781 implemented as the option to control behaviour regarding
10782 these dependencies was never added to systemctl. By removing
10783 these dependency types the execution engine becomes a bit
10784 simpler. Unit files that use these dependencies should be
10785 changed to use the non-Overridable dependency types
10786 instead. In fact, when parsing unit files with these
10787 options, that's what systemd will automatically convert them
10788 too, but it will also warn, asking users to fix the unit
10789 files accordingly. Removal of these dependency types should
10790 only affect a negligible number of unit files in the wild.
10792 * Behaviour of networkd's IPForward= option changed
10793 (again). It will no longer maintain a per-interface setting,
10794 but propagate one way from interfaces where this is enabled
10795 to the global kernel setting. The global setting will be
10796 enabled when requested by a network that is set up, but
10797 never be disabled again. This change was made to make sure
10798 IPv4 and IPv6 behaviour regarding packet forwarding is
10799 similar (as the Linux IPv6 stack does not support
10800 per-interface control of this setting) and to minimize
10803 * In unit files the behaviour of %u, %U, %h, %s has
10804 changed. These specifiers will now unconditionally resolve
10805 to the various user database fields of the user that the
10806 systemd instance is running as, instead of the user
10807 configured in the specific unit via User=. Note that this
10808 effectively doesn't change much, as resolving of these
10809 specifiers was already turned off in the --system instance
10810 of systemd, as we cannot do NSS lookups from PID 1. In the
10811 --user instance of systemd these specifiers where correctly
10812 resolved, but hardly made any sense, since the user instance
10813 lacks privileges to do user switches anyway, and User= is
10814 hence useless. Moreover, even in the --user instance of
10815 systemd behaviour was awkward as it would only take settings
10816 from User= assignment placed before the specifier into
10817 account. In order to unify and simplify the logic around
10818 this the specifiers will now always resolve to the
10819 credentials of the user invoking the manager (which in case
10820 of PID 1 is the root user).
10822 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
10823 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
10824 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
10825 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
10826 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
10827 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
10828 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
10829 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
10830 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
10831 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
10832 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
10833 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
10834 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
10835 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
10838 — Berlin, 2015-11-18
10842 * systemd now depends on util-linux v2.27. More specifically,
10843 the newly added mount monitor feature in libmount now
10844 replaces systemd's former own implementation.
10846 * libmount mandates /etc/mtab not to be regular file, and
10847 systemd now enforces this condition at early boot.
10848 /etc/mtab has been deprecated and warned about for a very
10849 long time, so systems running systemd should already have
10850 stopped having this file around as anything else than a
10851 symlink to /proc/self/mounts.
10853 * Support for the "pids" cgroup controller has been added. It
10854 allows accounting the number of tasks in a cgroup and
10855 enforcing limits on it. This adds two new setting
10856 TasksAccounting= and TasksMax= to each unit, as well as a
10857 global option DefaultTasksAccounting=.
10859 * Support for the "net_cls" cgroup controller has been added.
10860 It allows assigning a net class ID to each task in the
10861 cgroup, which can then be used in firewall rules and traffic
10862 shaping configurations. Note that the kernel netfilter net
10863 class code does not currently work reliably for ingress
10864 packets on unestablished sockets.
10866 This adds a new config directive called NetClass= to CGroup
10867 enabled units. Allowed values are positive numbers for fixed
10868 assignments and "auto" for picking a free value
10871 * 'systemctl is-system-running' now returns 'offline' if the
10872 system is not booted with systemd. This command can now be
10873 used as a substitute for 'systemd-notify --booted'.
10875 * Watchdog timeouts have been increased to 3 minutes for all
10876 in-tree service files. Apparently, disk IO issues are more
10877 frequent than we hoped, and user reported >1 minute waiting
10880 * 'machine-id-commit' functionality has been merged into
10881 'machine-id-setup --commit'. The separate binary has been
10884 * The WorkingDirectory= directive in unit files may now be set
10885 to the special value '~'. In this case, the working
10886 directory is set to the home directory of the user
10887 configured in User=.
10889 * "machinectl shell" will now open the shell in the home
10890 directory of the selected user by default.
10892 * The CrashChVT= configuration file setting is renamed to
10893 CrashChangeVT=, following our usual logic of not
10894 abbreviating unnecessarily. The old directive is still
10895 supported for compat reasons. Also, this directive now takes
10896 an integer value between 1 and 63, or a boolean value. The
10897 formerly supported '-1' value for disabling stays around for
10900 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
10901 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
10902 RootDirectory= properties can now be set for transient
10905 * The systemd-analyze tool gained a new "set-log-target" verb
10906 to change the logging target the system manager logs to
10907 dynamically during runtime. This is similar to how
10908 "systemd-analyze set-log-level" already changes the log
10911 * In nspawn /sys is now mounted as tmpfs, with only a selected
10912 set of subdirectories mounted in from the real sysfs. This
10913 enhances security slightly, and is useful for ensuring user
10914 namespaces work correctly.
10916 * Support for USB FunctionFS activation has been added. This
10917 allows implementation of USB gadget services that are
10918 activated as soon as they are requested, so that they don't
10919 have to run continuously, similar to classic socket
10922 * The "systemctl exit" command now optionally takes an
10923 additional parameter that sets the exit code to return from
10924 the systemd manager when exiting. This is only relevant when
10925 running the systemd user instance, or when running the
10926 system instance in a container.
10928 * sd-bus gained the new API calls sd_bus_path_encode_many()
10929 and sd_bus_path_decode_many() that allow easy encoding and
10930 decoding of multiple identifier strings inside a D-Bus
10931 object path. Another new call sd_bus_default_flush_close()
10932 has been added to flush and close per-thread default
10935 * systemd-cgtop gained support for a -M/--machine= switch to
10936 show the control groups within a certain container only.
10938 * "systemctl kill" gained support for an optional --fail
10939 switch. If specified the requested operation will fail of no
10940 processes have been killed, because the unit had no
10941 processes attached, or similar.
10943 * A new systemd.crash_reboot=1 kernel command line option has
10944 been added that triggers a reboot after crashing. This can
10945 also be set through CrashReboot= in systemd.conf.
10947 * The RuntimeDirectory= setting now understands unit
10948 specifiers like %i or %f.
10950 * A new (still internal) library API sd-ipv4acd has been added,
10951 that implements address conflict detection for IPv4. It's
10952 based on code from sd-ipv4ll, and will be useful for
10953 detecting DHCP address conflicts.
10955 * File descriptors passed during socket activation may now be
10956 named. A new API sd_listen_fds_with_names() is added to
10957 access the names. The default names may be overridden,
10958 either in the .socket file using the FileDescriptorName=
10959 parameter, or by passing FDNAME= when storing the file
10960 descriptors using sd_notify().
10962 * systemd-networkd gained support for:
10964 - Setting the IPv6 Router Advertisement settings via
10965 IPv6AcceptRouterAdvertisements= in .network files.
10967 - Configuring the HelloTimeSec=, MaxAgeSec= and
10968 ForwardDelaySec= bridge parameters in .netdev files.
10970 - Configuring PreferredSource= for static routes in
10973 * The "ask-password" framework used to query for LUKS harddisk
10974 passwords or SSL passwords during boot gained support for
10975 caching passwords in the kernel keyring, if it is
10976 available. This makes sure that the user only has to type in
10977 a passphrase once if there are multiple objects to unlock
10978 with the same one. Previously, such password caching was
10979 available only when Plymouth was used; this moves the
10980 caching logic into the systemd codebase itself. The
10981 "systemd-ask-password" utility gained a new --keyname=
10982 switch to control which kernel keyring key to use for
10983 caching a password in. This functionality is also useful for
10984 enabling display managers such as gdm to automatically
10985 unlock the user's GNOME keyring if its passphrase, the
10986 user's password and the harddisk password are the same, if
10987 gdm-autologin is used.
10989 * When downloading tar or raw images using "machinectl
10990 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
10991 file is now also downloaded, if it is available and stored
10992 next to the image file.
10994 * Units of type ".socket" gained a new boolean setting
10995 Writable= which is only useful in conjunction with
10996 ListenSpecial=. If true, enables opening the specified
10997 special file in O_RDWR mode rather than O_RDONLY mode.
10999 * systemd-rfkill has been reworked to become a singleton
11000 service that is activated through /dev/rfkill on each rfkill
11001 state change and saves the settings to disk. This way,
11002 systemd-rfkill is now compatible with devices that exist
11003 only intermittendly, and even restores state if the previous
11004 system shutdown was abrupt rather than clean.
11006 * The journal daemon gained support for vacuuming old journal
11007 files controlled by the number of files that shall remain,
11008 in addition to the already existing control by size and by
11009 date. This is useful as journal interleaving performance
11010 degrades with too many separate journal files, and allows
11011 putting an effective limit on them. The new setting defaults
11012 to 100, but this may be changed by setting SystemMaxFiles=
11013 and RuntimeMaxFiles= in journald.conf. Also, the
11014 "journalctl" tool gained the new --vacuum-files= switch to
11015 manually vacuum journal files to leave only the specified
11016 number of files in place.
11018 * udev will now create /dev/disk/by-path links for ATA devices
11019 on kernels where that is supported.
11021 * Galician, Serbian, Turkish and Korean translations were added.
11023 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
11024 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
11025 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
11026 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
11027 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
11028 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
11029 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
11030 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
11031 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
11032 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
11033 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
11034 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
11035 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
11036 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
11037 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
11038 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
11039 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
11040 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
11042 — Berlin, 2015-10-07
11046 * The DHCP implementation of systemd-networkd gained a set of
11049 - The DHCP server now supports emitting DNS and NTP
11050 information. It may be enabled and configured via
11051 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
11052 and NTP information is enabled, but no servers are
11053 configured, the corresponding uplink information (if there
11054 is any) is propagated.
11056 - Server and client now support transmission and reception
11057 of timezone information. It can be configured via the
11058 newly introduced network options UseTimezone=,
11059 EmitTimezone=, and Timezone=. Transmission of timezone
11060 information is enabled between host and containers by
11061 default now: the container will change its local timezone
11062 to what the host has set.
11064 - Lease timeouts can now be configured via
11065 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
11067 - The DHCP server improved on the stability of
11068 leases. Clients are more likely to get the same lease
11069 information back, even if the server loses state.
11071 - The DHCP server supports two new configuration options to
11072 control the lease address pool metrics, PoolOffset= and
11075 * The encapsulation limit of tunnels in systemd-networkd may
11076 now be configured via 'EncapsulationLimit='. It allows
11077 modifying the maximum additional levels of encapsulation
11078 that are permitted to be prepended to a packet.
11080 * systemd now supports the concept of user buses replacing
11081 session buses, if used with dbus-1.10 (and enabled via dbus
11082 --enable-user-session). It previously only supported this on
11083 kdbus-enabled systems, and this release expands this to
11084 'dbus-daemon' systems.
11086 * systemd-networkd now supports predictable interface names
11087 for virtio devices.
11089 * systemd now optionally supports the new Linux kernel
11090 "unified" control group hierarchy. If enabled via the kernel
11091 command-line option 'systemd.unified_cgroup_hierarchy=1',
11092 systemd will try to mount the unified cgroup hierarchy
11093 directly on /sys/fs/cgroup. If not enabled, or not
11094 available, systemd will fall back to the legacy cgroup
11095 hierarchy setup, as before. Host system and containers can
11096 mix and match legacy and unified hierarchies as they
11097 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
11098 environment variable to individually select the hierarchy to
11099 use for executed containers. By default, nspawn will use the
11100 unified hierarchy for the containers if the host uses the
11101 unified hierarchy, and the legacy hierarchy otherwise.
11102 Please note that at this point the unified hierarchy is an
11103 experimental kernel feature and is likely to change in one
11104 of the next kernel releases. Therefore, it should not be
11105 enabled by default in downstream distributions yet. The
11106 minimum required kernel version for the unified hierarchy to
11107 work is 4.2. Note that when the unified hierarchy is used
11108 for the first time delegated access to controllers is
11109 safe. Because of this systemd-nspawn containers will get
11110 access to controllers now, as will systemd user
11111 sessions. This means containers and user sessions may now
11112 manage their own resources, partitioning up what the system
11115 * A new special scope unit "init.scope" has been introduced
11116 that encapsulates PID 1 of the system. It may be used to
11117 determine resource usage and enforce resource limits on PID
11118 1 itself. PID 1 hence moved out of the root of the control
11121 * The cgtop tool gained support for filtering out kernel
11122 threads when counting tasks in a control group. Also, the
11123 count of processes is now recursively summed up by
11124 default. Two options -k and --recursive= have been added to
11125 revert to old behaviour. The tool has also been updated to
11126 work correctly in containers now.
11128 * systemd-nspawn's --bind= and --bind-ro= options have been
11129 extended to allow creation of non-recursive bind mounts.
11131 * libsystemd gained two new calls sd_pid_get_cgroup() and
11132 sd_peer_get_cgroup() which return the control group path of
11133 a process or peer of a connected AF_UNIX socket. This
11134 function call is particularly useful when implementing
11135 delegated subtrees support in the control group hierarchy.
11137 * The "sd-event" event loop API of libsystemd now supports
11138 correct dequeuing of real-time signals, without losing
11141 * When systemd requests a polkit decision when managing units it
11142 will now add additional fields to the request, including unit
11143 name and desired operation. This enables more powerful polkit
11144 policies, that make decisions depending on these parameters.
11146 * nspawn learnt support for .nspawn settings files, that may
11147 accompany the image files or directories of containers, and
11148 may contain additional settings for the container. This is
11149 an alternative to configuring container parameters via the
11150 nspawn command line.
11152 Contributions from: Cristian Rodríguez, Daniel Mack, David
11153 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
11154 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
11155 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
11156 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
11157 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
11158 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
11159 Andersen, Tom Gundersen, Torstein Husebø
11161 — Berlin, 2015-09-08
11165 * machinectl gained a new verb 'shell' which opens a fresh
11166 shell on the target container or the host. It is similar to
11167 the existing 'login' command of machinectl, but spawns the
11168 shell directly without prompting for username or
11169 password. The pseudo machine '.host' now refers to the local
11170 host and is used by default. Hence, 'machinectl shell' can
11171 be used as replacement for 'su -' which spawns a session as
11172 a fresh systemd unit in a way that is fully isolated from
11173 the originating session.
11175 * systemd-networkd learned to cope with private-zone DHCP
11176 options and allows other programs to query the values.
11178 * SELinux access control when enabling/disabling units is no
11179 longer enforced with this release. The previous implementation
11180 was incorrect, and a new corrected implementation is not yet
11181 available. As unit file operations are still protected via
11182 polkit and D-Bus policy this is not a security problem. Yet,
11183 distributions which care about optimal SELinux support should
11184 probably not stabilize on this release.
11186 * sd-bus gained support for matches of type "arg0has=", that
11187 test for membership of strings in string arrays sent in bus
11190 * systemd-resolved now dumps the contents of its DNS and LLMNR
11191 caches to the logs on reception of the SIGUSR1 signal. This
11192 is useful to debug DNS behaviour.
11194 * The coredumpctl tool gained a new --directory= option to
11195 operate on journal files in a specific directory.
11197 * "systemctl reboot" and related commands gained a new
11198 "--message=" option which may be used to set a free-text
11199 wall message when shutting down or rebooting the
11200 system. This message is also logged, which is useful for
11201 figuring out the reason for a reboot or shutdown a
11204 * The "systemd-resolve-host" tool's -i switch now takes
11205 network interface numbers as alternative to interface names.
11207 * A new unit file setting for services has been introduced:
11208 UtmpMode= allows configuration of how precisely systemd
11209 handles utmp and wtmp entries for the service if this is
11210 enabled. This allows writing services that appear similar to
11211 user sessions in the output of the "w", "who", "last" and
11214 * systemd-resolved will now locally synthesize DNS resource
11215 records for the "localhost" and "gateway" domains as well as
11216 the local hostname. This should ensure that clients querying
11217 RRs via resolved will get similar results as those going via
11218 NSS, if nss-myhostname is enabled.
11220 Contributions from: Alastair Hughes, Alex Crawford, Daniel
11221 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
11222 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
11223 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
11224 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
11225 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
11226 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
11227 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
11228 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
11229 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
11230 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
11231 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11233 — Berlin, 2015-08-27
11237 * The systemd-efi-boot-generator functionality was merged into
11238 systemd-gpt-auto-generator.
11240 * systemd-networkd now supports Group Policy for vxlan
11241 devices. It can be enabled via the new boolean configuration
11242 option called 'GroupPolicyExtension='.
11244 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
11245 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
11246 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
11248 — Berlin, 2015-07-31
11252 * The python-systemd code has been removed from the systemd repository.
11253 A new repository has been created which accommodates the code from
11254 now on, and we kindly ask distributions to create a separate package
11255 for this: https://github.com/systemd/python-systemd
11257 * The systemd daemon will now reload its main configuration
11258 (/etc/systemd/system.conf) on daemon-reload.
11260 * sd-dhcp now exposes vendor specific extensions via
11261 sd_dhcp_lease_get_vendor_specific().
11263 * systemd-networkd gained a number of new configuration options.
11265 - A new boolean configuration option for TAP devices called
11266 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
11267 device, thus allowing to send and receive GSO packets.
11269 - A new tunnel configuration option called 'CopyDSCP='.
11270 If enabled, the DSCP field of ip6 tunnels is copied into the
11271 decapsulated packet.
11273 - A set of boolean bridge configuration options were added.
11274 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
11275 and 'UnicastFlood=' are now parsed by networkd and applied to the
11276 respective bridge link device via the respective IFLA_BRPORT_*
11279 - A new string configuration option to override the hostname sent
11280 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
11281 is true, networkd will use the configured hostname instead of the
11282 system hostname when sending DHCP requests.
11284 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
11285 networkd will configure the IPv6 flow-label of the tunnel device
11286 according to RFC2460.
11288 - The 'macvtap' virtual network devices are now supported, similar to
11289 the already supported 'macvlan' devices.
11291 * systemd-resolved now implements RFC5452 to improve resilience against
11292 cache poisoning. Additionally, source port randomization is enabled
11293 by default to further protect against DNS spoofing attacks.
11295 * nss-mymachines now supports translating UIDs and GIDs of running
11296 containers with user-namespaces enabled. If a container 'foo'
11297 translates a host uid 'UID' to the container uid 'TUID', then
11298 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
11299 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
11300 mapped as 'vg-foo-TGID'.
11302 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
11303 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
11304 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
11305 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
11306 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
11307 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
11308 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
11309 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
11310 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
11311 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
11313 — Berlin, 2015-07-29
11317 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
11318 There are no known issues with current sysfs, and udev does not need
11319 or should be used to work around such bugs.
11321 * udev does no longer enable USB HID power management. Several reports
11322 indicate, that some devices cannot handle that setting.
11324 * The udev accelerometer helper was removed. The functionality
11325 is now fully included in iio-sensor-proxy. But this means,
11326 older iio-sensor-proxy versions will no longer provide
11327 accelerometer/orientation data with this systemd version.
11328 Please upgrade iio-sensor-proxy to version 1.0.
11330 * networkd gained a new configuration option IPv6PrivacyExtensions=
11331 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
11332 for Stateless Address") on selected networks.
11334 * For the sake of fewer build-time dependencies and less code in the
11335 main repository, the python bindings are about to be removed in the
11336 next release. A new repository has been created which accommodates
11337 the code from now on, and we kindly ask distributions to create a
11338 separate package for this. The removal will take place in v223.
11340 https://github.com/systemd/python-systemd
11342 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
11343 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
11344 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
11345 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
11346 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
11347 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
11348 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
11349 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
11350 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
11351 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11353 — Berlin, 2015-07-07
11357 * The sd-bus.h and sd-event.h APIs have now been declared
11358 stable and have been added to the official interface of
11359 libsystemd.so. sd-bus implements an alternative D-Bus client
11360 library, that is relatively easy to use, very efficient and
11361 supports both classic D-Bus as well as kdbus as transport
11362 backend. sd-event is a generic event loop abstraction that
11363 is built around Linux epoll, but adds features such as event
11364 prioritization or efficient timer handling. Both APIs are good
11365 choices for C programs looking for a bus and/or event loop
11366 implementation that is minimal and does not have to be
11367 portable to other kernels.
11369 * kdbus support is no longer compile-time optional. It is now
11370 always built-in. However, it can still be disabled at
11371 runtime using the kdbus=0 kernel command line setting, and
11372 that setting may be changed to default to off, by specifying
11373 --disable-kdbus at build-time. Note though that the kernel
11374 command line setting has no effect if the kdbus.ko kernel
11375 module is not installed, in which case kdbus is (obviously)
11376 also disabled. We encourage all downstream distributions to
11377 begin testing kdbus by adding it to the kernel images in the
11378 development distributions, and leaving kdbus support in
11381 * The minimal required util-linux version has been bumped to
11384 * Support for chkconfig (--enable-chkconfig) was removed in
11385 favor of calling an abstraction tool
11386 /lib/systemd/systemd-sysv-install. This needs to be
11387 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
11388 in README for details.
11390 * If there's a systemd unit and a SysV init script for the
11391 same service name, and the user executes "systemctl enable"
11392 for it (or a related call), then this will now enable both
11393 (or execute the related operation on both), not just the
11396 * The libudev API documentation has been converted from gtkdoc
11399 * gudev has been removed from the systemd tree, it is now an
11402 * The systemd-cgtop tool learnt a new --raw switch to generate
11403 "raw" (machine parsable) output.
11405 * networkd's IPForwarding= .network file setting learnt the
11406 new setting "kernel", which ensures that networkd does not
11407 change the IP forwarding sysctl from the default kernel
11410 * The systemd-logind bus API now exposes a new boolean
11411 property "Docked" that reports whether logind considers the
11412 system "docked", i.e. connected to a docking station or not.
11414 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
11415 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
11416 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
11417 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
11418 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
11419 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
11420 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
11421 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
11422 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
11423 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
11424 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
11425 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
11426 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
11427 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
11428 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
11429 Fink, Zbigniew Jędrzejewski-Szmek
11431 — Berlin, 2015-06-19
11435 * The gudev library has been extracted into a separate repository
11436 available at: https://git.gnome.org/browse/libgudev/
11437 It is now managed as part of the Gnome project. Distributions
11438 are recommended to pass --disable-gudev to systemd and use
11439 gudev from the Gnome project instead. gudev is still included
11440 in systemd, for now. It will be removed soon, though. Please
11441 also see the announcement-thread on systemd-devel:
11442 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
11444 * systemd now exposes a CPUUsageNSec= property for each
11445 service unit on the bus, that contains the overall consumed
11446 CPU time of a service (the sum of what each process of the
11447 service consumed). This value is only available if
11448 CPUAccounting= is turned on for a service, and is then shown
11449 in the "systemctl status" output.
11451 * Support for configuring alternative mappings of the old SysV
11452 runlevels to systemd targets has been removed. They are now
11453 hardcoded in a way that runlevels 2, 3, 4 all map to
11454 multi-user.target and 5 to graphical.target (which
11455 previously was already the default behaviour).
11457 * The auto-mounter logic gained support for mount point
11458 expiry, using a new TimeoutIdleSec= setting in .automount
11459 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
11461 * The EFI System Partition (ESP) as mounted to /boot by
11462 systemd-efi-boot-generator will now be unmounted
11463 automatically after 2 minutes of not being used. This should
11464 minimize the risk of ESP corruptions.
11466 * New /etc/fstab options x-systemd.requires= and
11467 x-systemd.requires-mounts-for= are now supported to express
11468 additional dependencies for mounts. This is useful for
11469 journaling file systems that support external journal
11470 devices or overlay file systems that require underlying file
11471 systems to be mounted.
11473 * systemd does not support direct live-upgrades (via systemctl
11474 daemon-reexec) from versions older than v44 anymore. As no
11475 distribution we are aware of shipped such old versions in a
11476 stable release this should not be problematic.
11478 * When systemd forks off a new per-connection service instance
11479 it will now set the $REMOTE_ADDR environment variable to the
11480 remote IP address, and $REMOTE_PORT environment variable to
11481 the remote IP port. This behaviour is similar to the
11482 corresponding environment variables defined by CGI.
11484 * systemd-networkd gained support for uplink failure
11485 detection. The BindCarrier= option allows binding interface
11486 configuration dynamically to the link sense of other
11487 interfaces. This is useful to achieve behaviour like in
11490 * systemd-networkd gained support for configuring the DHCP
11491 client identifier to use when requesting leases.
11493 * systemd-networkd now has a per-network UseNTP= option to
11494 configure whether NTP server information acquired via DHCP
11495 is passed on to services like systemd-timesyncd.
11497 * systemd-networkd gained support for vti6 tunnels.
11499 * Note that systemd-networkd manages the sysctl variable
11500 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
11501 it is configured for since v219. The variable controls IP
11502 forwarding, and is a per-interface alternative to the global
11503 /proc/sys/net/ipv[46]/ip_forward. This setting is
11504 configurable in the IPForward= option, which defaults to
11505 "no". This means if networkd is used for an interface it is
11506 no longer sufficient to set the global sysctl option to turn
11507 on IP forwarding! Instead, the .network file option
11508 IPForward= needs to be turned on! Note that the
11509 implementation of this behaviour was broken in v219 and has
11510 been fixed in v220.
11512 * Many bonding and vxlan options are now configurable in
11515 * systemd-nspawn gained a new --property= setting to set unit
11516 properties for the container scope. This is useful for
11517 setting resource parameters (e.g. "CPUShares=500") on
11518 containers started from the command line.
11520 * systemd-nspawn gained a new --private-users= switch to make
11521 use of user namespacing available on recent Linux kernels.
11523 * systemd-nspawn may now be called as part of a shell pipeline
11524 in which case the pipes used for stdin and stdout are passed
11525 directly to the process invoked in the container, without
11526 indirection via a pseudo tty.
11528 * systemd-nspawn gained a new switch to control the UNIX
11529 signal to use when killing the init process of the container
11530 when shutting down.
11532 * systemd-nspawn gained a new --overlay= switch for mounting
11533 overlay file systems into the container using the new kernel
11536 * When a container image is imported via systemd-importd and
11537 the host file system is not btrfs, a loopback block device
11538 file is created in /var/lib/machines.raw with a btrfs file
11539 system inside. It is then mounted to /var/lib/machines to
11540 enable btrfs features for container management. The loopback
11541 file and btrfs file system is grown as needed when container
11542 images are imported via systemd-importd.
11544 * systemd-machined/systemd-importd gained support for btrfs
11545 quota, to enforce container disk space limits on disk. This
11546 is exposed in "machinectl set-limit".
11548 * systemd-importd now can import containers from local .tar,
11549 .raw and .qcow2 images, and export them to .tar and .raw. It
11550 can also import dkr v2 images now from the network (on top
11553 * systemd-importd gained support for verifying downloaded
11554 images with gpg2 (previously only gpg1 was supported).
11556 * systemd-machined, systemd-logind, systemd: most bus calls are
11557 now accessible to unprivileged processes via polkit. Also,
11558 systemd-logind will now allow users to kill their own sessions
11559 without further privileges or authorization.
11561 * systemd-shutdownd has been removed. This service was
11562 previously responsible for implementing scheduled shutdowns
11563 as exposed in /usr/bin/shutdown's time parameter. This
11564 functionality has now been moved into systemd-logind and is
11565 accessible via a bus interface.
11567 * "systemctl reboot" gained a new switch --firmware-setup that
11568 can be used to reboot into the EFI firmware setup, if that
11569 is available. systemd-logind now exposes an API on the bus
11570 to trigger such reboots, in case graphical desktop UIs want
11571 to cover this functionality.
11573 * "systemctl enable", "systemctl disable" and "systemctl mask"
11574 now support a new "--now" switch. If specified the units
11575 that are enabled will also be started, and the ones
11576 disabled/masked also stopped.
11578 * The Gummiboot EFI boot loader tool has been merged into
11579 systemd, and renamed to "systemd-boot". The bootctl tool has been
11580 updated to support systemd-boot.
11582 * An EFI kernel stub has been added that may be used to create
11583 kernel EFI binaries that contain not only the actual kernel,
11584 but also an initrd, boot splash, command line and OS release
11585 information. This combined binary can then be signed as a
11586 single image, so that the firmware can verify it all in one
11587 step. systemd-boot has special support for EFI binaries created
11588 like this and can extract OS release information from them
11589 and show them in the boot menu. This functionality is useful
11590 to implement cryptographically verified boot schemes.
11592 * Optional support has been added to systemd-fsck to pass
11593 fsck's progress report to an AF_UNIX socket in the file
11596 * udev will no longer create device symlinks for all block devices by
11597 default. A deny list for excluding special block devices from this
11598 logic has been turned into an allow list that requires picking block
11599 devices explicitly that require device symlinks.
11601 * A new (currently still internal) API sd-device.h has been
11602 added to libsystemd. This modernized API is supposed to
11603 replace libudev eventually. In fact, already much of libudev
11604 is now just a wrapper around sd-device.h.
11606 * A new hwdb database for storing metadata about pointing
11607 stick devices has been added.
11609 * systemd-tmpfiles gained support for setting file attributes
11610 similar to the "chattr" tool with new 'h' and 'H' lines.
11612 * systemd-journald will no longer unconditionally set the
11613 btrfs NOCOW flag on new journal files. This is instead done
11614 with tmpfiles snippet using the new 'h' line type. This
11615 allows easy disabling of this logic, by masking the
11616 journal-nocow.conf tmpfiles file.
11618 * systemd-journald will now translate audit message types to
11619 human readable identifiers when writing them to the
11620 journal. This should improve readability of audit messages.
11622 * The LUKS logic gained support for the offset= and skip=
11623 options in /etc/crypttab, as previously implemented by
11626 * /usr/lib/os-release gained a new optional field VARIANT= for
11627 distributions that support multiple variants (such as a
11628 desktop edition, a server edition, …)
11630 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
11631 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
11632 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
11633 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
11634 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
11635 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
11636 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
11637 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
11638 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
11639 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
11640 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
11641 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
11642 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
11643 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
11644 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
11645 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
11646 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
11647 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
11648 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
11649 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
11650 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
11651 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
11652 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
11653 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
11654 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
11655 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
11656 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
11658 — Berlin, 2015-05-22
11662 * Introduce a new API "sd-hwdb.h" for querying the hardware
11663 metadata database. With this minimal interface one can query
11664 and enumerate the udev hwdb, decoupled from the old libudev
11665 library. libudev's interface for this is now only a wrapper
11666 around sd-hwdb. A new tool systemd-hwdb has been added to
11667 interface with and update the database.
11669 * When any of systemd's tools copies files (for example due to
11670 tmpfiles' C lines) a btrfs reflink will attempted first,
11671 before bytewise copying is done.
11673 * systemd-nspawn gained a new --ephemeral switch. When
11674 specified a btrfs snapshot is taken of the container's root
11675 directory, and immediately removed when the container
11676 terminates again. Thus, a container can be started whose
11677 changes never alter the container's root directory, and are
11678 lost on container termination. This switch can also be used
11679 for starting a container off the root file system of the
11680 host without affecting the host OS. This switch is only
11681 available on btrfs file systems.
11683 * systemd-nspawn gained a new --template= switch. It takes the
11684 path to a container tree to use as template for the tree
11685 specified via --directory=, should that directory be
11686 missing. This allows instantiating containers dynamically,
11687 on first run. This switch is only available on btrfs file
11690 * When a .mount unit refers to a mount point on which multiple
11691 mounts are stacked, and the .mount unit is stopped all of
11692 the stacked mount points will now be unmounted until no
11693 mount point remains.
11695 * systemd now has an explicit notion of supported and
11696 unsupported unit types. Jobs enqueued for unsupported unit
11697 types will now fail with an "unsupported" error code. More
11698 specifically .swap, .automount and .device units are not
11699 supported in containers, .busname units are not supported on
11700 non-kdbus systems. .swap and .automount are also not
11701 supported if their respective kernel compile time options
11704 * machinectl gained support for two new "copy-from" and
11705 "copy-to" commands for copying files from a running
11706 container to the host or vice versa.
11708 * machinectl gained support for a new "bind" command to bind
11709 mount host directories into local containers. This is
11710 currently only supported for nspawn containers.
11712 * networkd gained support for configuring bridge forwarding
11713 database entries (fdb) from .network files.
11715 * A new tiny daemon "systemd-importd" has been added that can
11716 download container images in tar, raw, qcow2 or dkr formats,
11717 and make them available locally in /var/lib/machines, so
11718 that they can run as nspawn containers. The daemon can GPG
11719 verify the downloads (not supported for dkr, since it has no
11720 provisions for verifying downloads). It will transparently
11721 decompress bz2, xz, gzip compressed downloads if necessary,
11722 and restore sparse files on disk. The daemon uses privilege
11723 separation to ensure the actual download logic runs with
11724 fewer privileges than the daemon itself. machinectl has
11725 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
11726 make the functionality of importd available to the
11727 user. With this in place the Fedora and Ubuntu "Cloud"
11728 images can be downloaded and booted as containers unmodified
11729 (the Fedora images lack the appropriate GPG signature files
11730 currently, so they cannot be verified, but this will change
11731 soon, hopefully). Note that downloading images is currently
11732 only fully supported on btrfs.
11734 * machinectl is now able to list container images found in
11735 /var/lib/machines, along with some metadata about sizes of
11736 disk and similar. If the directory is located on btrfs and
11737 quota is enabled, this includes quota display. A new command
11738 "image-status" has been added that shows additional
11739 information about images.
11741 * machinectl is now able to clone container images
11742 efficiently, if the underlying file system (btrfs) supports
11743 it, with the new "machinectl clone" command. It also
11744 gained commands for renaming and removing images, as well as
11745 marking them read-only or read-write (supported also on
11746 legacy file systems).
11748 * networkd gained support for collecting LLDP network
11749 announcements, from hardware that supports this. This is
11750 shown in networkctl output.
11752 * systemd-run gained support for a new -t (--pty) switch for
11753 invoking a binary on a pty whose input and output is
11754 connected to the invoking terminal. This allows executing
11755 processes as system services while interactively
11756 communicating with them via the terminal. Most interestingly
11757 this is supported across container boundaries. Invoking
11758 "systemd-run -t /bin/bash" is an alternative to running a
11759 full login session, the difference being that the former
11760 will not register a session, nor go through the PAM session
11763 * tmpfiles gained support for a new "v" line type for creating
11764 btrfs subvolumes. If the underlying file system is a legacy
11765 file system, this automatically degrades to creating a
11766 normal directory. Among others /var/lib/machines is now
11767 created like this at boot, should it be missing.
11769 * The directory /var/lib/containers/ has been deprecated and
11770 been replaced by /var/lib/machines. The term "machines" has
11771 been used in the systemd context as generic term for both
11772 VMs and containers, and hence appears more appropriate for
11773 this, as the directory can also contain raw images bootable
11776 * systemd-nspawn when invoked with -M but without --directory=
11777 or --image= is now capable of searching for the container
11778 root directory, subvolume or disk image automatically, in
11779 /var/lib/machines. systemd-nspawn@.service has been updated
11780 to make use of this, thus allowing it to be used for raw
11783 * A new machines.target unit has been introduced that is
11784 supposed to group all containers/VMs invoked as services on
11785 the system. systemd-nspawn@.service has been updated to
11786 integrate with that.
11788 * machinectl gained a new "start" command, for invoking a
11789 container as a service. "machinectl start foo" is mostly
11790 equivalent to "systemctl start systemd-nspawn@foo.service",
11791 but handles escaping in a nicer way.
11793 * systemd-nspawn will now mount most of the cgroupfs tree
11794 read-only into each container, with the exception of the
11795 container's own subtree in the name=systemd hierarchy.
11797 * journald now sets the special FS_NOCOW file flag for its
11798 journal files. This should improve performance on btrfs, by
11799 avoiding heavy fragmentation when journald's write-pattern
11800 is used on COW file systems. It degrades btrfs' data
11801 integrity guarantees for the files to the same levels as for
11802 ext3/ext4 however. This should be OK though as journald does
11803 its own data integrity checks and all its objects are
11804 checksummed on disk. Also, journald should handle btrfs disk
11805 full events a lot more gracefully now, by processing SIGBUS
11806 errors, and not relying on fallocate() anymore.
11808 * When journald detects that journal files it is writing to
11809 have been deleted it will immediately start new journal
11812 * systemd now provides a way to store file descriptors
11813 per-service in PID 1. This is useful for daemons to ensure
11814 that fds they require are not lost during a daemon
11815 restart. The fds are passed to the daemon on the next
11816 invocation in the same way socket activation fds are
11817 passed. This is now used by journald to ensure that the
11818 various sockets connected to all the system's stdout/stderr
11819 are not lost when journald is restarted. File descriptors
11820 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
11821 an extension to sd_notify(). Note that a limit is enforced
11822 on the number of fds a service can store in PID 1, and it
11823 defaults to 0, so that no fds may be stored, unless this is
11824 explicitly turned on.
11826 * The default TERM variable to use for units connected to a
11827 terminal, when no other value is explicitly is set is now
11828 vt220 rather than vt102. This should be fairly safe still,
11829 but allows PgUp/PgDn work.
11831 * The /etc/crypttab option header= as known from Debian is now
11834 * "loginctl user-status" and "loginctl session-status" will
11835 now show the last 10 lines of log messages of the
11836 user/session following the status output. Similar,
11837 "machinectl status" will show the last 10 log lines
11838 associated with a virtual machine or container
11839 service. (Note that this is usually not the log messages
11840 done in the VM/container itself, but simply what the
11841 container manager logs. For nspawn this includes all console
11844 * "loginctl session-status" without further argument will now
11845 show the status of the session of the caller. Similar,
11846 "lock-session", "unlock-session", "activate",
11847 "enable-linger", "disable-linger" may now be called without
11848 session/user parameter in which case they apply to the
11849 caller's session/user.
11851 * An X11 session scriptlet is now shipped that uploads
11852 $DISPLAY and $XAUTHORITY into the environment of the systemd
11853 --user daemon if a session begins. This should improve
11854 compatibility with X11 enabled applications run as systemd
11857 * Generators are now subject to masking via /etc and /run, the
11858 same way as unit files.
11860 * networkd .network files gained support for configuring
11861 per-link IPv4/IPv6 packet forwarding as well as IPv4
11862 masquerading. This is by default turned on for veth links to
11863 containers, as registered by systemd-nspawn. This means that
11864 nspawn containers run with --network-veth will now get
11865 automatic routed access to the host's networks without any
11866 further configuration or setup, as long as networkd runs on
11869 * systemd-nspawn gained the --port= (-p) switch to expose TCP
11870 or UDP posts of a container on the host. With this in place
11871 it is possible to run containers with private veth links
11872 (--network-veth), and have their functionality exposed on
11873 the host as if their services were running directly on the
11876 * systemd-nspawn's --network-veth switch now gained a short
11877 version "-n", since with the changes above it is now truly
11878 useful out-of-the-box. The systemd-nspawn@.service has been
11879 updated to make use of it too by default.
11881 * systemd-nspawn will now maintain a per-image R/W lock, to
11882 ensure that the same image is not started more than once
11883 writable. (It's OK to run an image multiple times
11884 simultaneously in read-only mode.)
11886 * systemd-nspawn's --image= option is now capable of
11887 dissecting and booting MBR and GPT disk images that contain
11888 only a single active Linux partition. Previously it
11889 supported only GPT disk images with proper GPT type
11890 IDs. This allows running cloud images from major
11891 distributions directly with systemd-nspawn, without
11894 * In addition to collecting mouse dpi data in the udev
11895 hardware database, there's now support for collecting angle
11896 information for mouse scroll wheels. The database is
11897 supposed to guarantee similar scrolling behavior on mice
11898 that it knows about. There's also support for collecting
11899 information about Touchpad types.
11901 * udev's input_id built-in will now also collect touch screen
11902 dimension data and attach it to probed devices.
11904 * /etc/os-release gained support for a Distribution Privacy
11907 * networkd gained support for creating "ipvlan", "gretap",
11908 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
11910 * systemd-tmpfiles gained support for "a" lines for setting
11913 * systemd-nspawn will now mount /tmp in the container to
11914 tmpfs, automatically.
11916 * systemd now exposes the memory.usage_in_bytes cgroup
11917 attribute and shows it for each service in the "systemctl
11918 status" output, if available.
11920 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
11921 immediate reboot is triggered. This useful if shutdown is
11922 hung and is unable to complete, to expedite the
11923 operation. Note that this kind of reboot will still unmount
11924 all file systems, and hence should not result in fsck being
11925 run on next reboot.
11927 * A .device unit for an optical block device will now be
11928 considered active only when a medium is in the drive. Also,
11929 mount units are now bound to their backing devices thus
11930 triggering automatic unmounting when devices become
11931 unavailable. With this in place systemd will now
11932 automatically unmount left-over mounts when a CD-ROM is
11933 ejected or a USB stick is yanked from the system.
11935 * networkd-wait-online now has support for waiting for
11936 specific interfaces only (with globbing), and for giving up
11937 after a configurable timeout.
11939 * networkd now exits when idle. It will be automatically
11940 restarted as soon as interfaces show up, are removed or
11941 change state. networkd will stay around as long as there is
11942 at least one DHCP state machine or similar around, that keep
11945 * networkd may now configure IPv6 link-local addressing in
11946 addition to IPv4 link-local addressing.
11948 * The IPv6 "token" for use in SLAAC may now be configured for
11949 each .network interface in networkd.
11951 * Routes configured with networkd may now be assigned a scope
11954 * networkd's [Match] sections now support globbing and lists
11955 of multiple space-separated matches per item.
11957 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
11958 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
11959 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
11960 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
11961 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
11962 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
11963 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
11964 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
11965 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
11966 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
11967 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
11968 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
11969 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
11970 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
11971 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
11972 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
11973 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
11974 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
11975 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
11976 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
11977 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
11978 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
11979 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
11980 Hoffmann, Zbigniew Jędrzejewski-Szmek
11982 — Berlin, 2015-02-16
11986 * When querying unit file enablement status (for example via
11987 "systemctl is-enabled"), a new state "indirect" is now known
11988 which indicates that a unit might not be enabled itself, but
11989 another unit listed in its Also= setting might be.
11991 * Similar to the various existing ConditionXYZ= settings for
11992 units, there are now matching AssertXYZ= settings. While
11993 failing conditions cause a unit to be skipped, but its job
11994 to succeed, failing assertions declared like this will cause
11995 a unit start operation and its job to fail.
11997 * hostnamed now knows a new chassis type "embedded".
11999 * systemctl gained a new "edit" command. When used on a unit
12000 file, this allows extending unit files with .d/ drop-in
12001 configuration snippets or editing the full file (after
12002 copying it from /usr/lib to /etc). This will invoke the
12003 user's editor (as configured with $EDITOR), and reload the
12004 modified configuration after editing.
12006 * "systemctl status" now shows the suggested enablement state
12007 for a unit, as declared in the (usually vendor-supplied)
12008 system preset files.
12010 * nss-myhostname will now resolve the single-label hostname
12011 "gateway" to the locally configured default IP routing
12012 gateways, ordered by their metrics. This assigns a stable
12013 name to the used gateways, regardless which ones are
12014 currently configured. Note that the name will only be
12015 resolved after all other name sources (if nss-myhostname is
12016 configured properly) and should hence not negatively impact
12017 systems that use the single-label hostname "gateway" in
12020 * systemd-inhibit now allows filtering by mode when listing
12023 * Scope and service units gained a new "Delegate" boolean
12024 property, which, when set, allows processes running inside the
12025 unit to further partition resources. This is primarily
12026 useful for systemd user instances as well as container
12029 * journald will now pick up audit messages directly from
12030 the kernel, and log them like any other log message. The
12031 audit fields are split up and fully indexed. This means that
12032 journalctl in many ways is now a (nicer!) alternative to
12033 ausearch, the traditional audit client. Note that this
12034 implements only a minimal audit client. If you want the
12035 special audit modes like reboot-on-log-overflow, please use
12036 the traditional auditd instead, which can be used in
12037 parallel to journald.
12039 * The ConditionSecurity= unit file option now understands the
12040 special string "audit" to check whether auditing is
12043 * journalctl gained two new commands --vacuum-size= and
12044 --vacuum-time= to delete old journal files until the
12045 remaining ones take up no more than the specified size on disk,
12046 or are not older than the specified time.
12048 * A new, native PPPoE library has been added to sd-network,
12049 systemd's library of light-weight networking protocols. This
12050 library will be used in a future version of networkd to
12051 enable PPPoE communication without an external pppd daemon.
12053 * The busctl tool now understands a new "capture" verb that
12054 works similar to "monitor", but writes a packet capture
12055 trace to STDOUT that can be redirected to a file which is
12056 compatible with libcap's capture file format. This can then
12057 be loaded in Wireshark and similar tools to inspect bus
12060 * The busctl tool now understands a new "tree" verb that shows
12061 the object trees of a specific service on the bus, or of all
12064 * The busctl tool now understands a new "introspect" verb that
12065 shows all interfaces and members of objects on the bus,
12066 including their signature and values. This is particularly
12067 useful to get more information about bus objects shown by
12068 the new "busctl tree" command.
12070 * The busctl tool now understands new verbs "call",
12071 "set-property" and "get-property" for invoking bus method
12072 calls, setting and getting bus object properties in a
12075 * busctl gained a new --augment-creds= argument that controls
12076 whether the tool shall augment credential information it
12077 gets from the bus with data from /proc, in a possibly
12080 * nspawn's --link-journal= switch gained two new values
12081 "try-guest" and "try-host" that work like "guest" and
12082 "host", but do not fail if the host has no persistent
12083 journaling enabled. -j is now equivalent to
12084 --link-journal=try-guest.
12086 * macvlan network devices created by nspawn will now have
12087 stable MAC addresses.
12089 * A new SmackProcessLabel= unit setting has been added, which
12090 controls the SMACK security label processes forked off by
12091 the respective unit shall use.
12093 * If compiled with --enable-xkbcommon, systemd-localed will
12094 verify x11 keymap settings by compiling the given keymap. It
12095 will spew out warnings if the compilation fails. This
12096 requires libxkbcommon to be installed.
12098 * When a coredump is collected, a larger number of metadata
12099 fields is now collected and included in the journal records
12100 created for it. More specifically, control group membership,
12101 environment variables, memory maps, working directory,
12102 chroot directory, /proc/$PID/status, and a list of open file
12103 descriptors is now stored in the log entry.
12105 * The udev hwdb now contains DPI information for mice. For
12108 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
12110 * All systemd programs that read standalone configuration
12111 files in /etc now also support a corresponding series of
12112 .conf.d configuration directories in /etc/, /run/,
12113 /usr/local/lib/, /usr/lib/, and (if configured with
12114 --enable-split-usr) /lib/. In particular, the following
12115 configuration files now have corresponding configuration
12116 directories: system.conf user.conf, logind.conf,
12117 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
12118 resolved.conf, timesyncd.conf, journal-remote.conf, and
12119 journal-upload.conf. Note that distributions should use the
12120 configuration directories in /usr/lib/; the directories in
12121 /etc/ are reserved for the system administrator.
12123 * systemd-rfkill will no longer take the rfkill device name
12124 into account when storing rfkill state on disk, as the name
12125 might be dynamically assigned and not stable. Instead, the
12126 ID_PATH udev variable combined with the rfkill type (wlan,
12127 bluetooth, …) is used.
12129 * A new service systemd-machine-id-commit.service has been
12130 added. When used on systems where /etc is read-only during
12131 boot, and /etc/machine-id is not initialized (but an empty
12132 file), this service will copy the temporary machine ID
12133 created as replacement into /etc after the system is fully
12134 booted up. This is useful for systems that are freshly
12135 installed with a non-initialized machine ID, but should get
12136 a fixed machine ID for subsequent boots.
12138 * networkd's .netdev files now provide a large set of
12139 configuration parameters for VXLAN devices. Similarly, the
12140 bridge port cost parameter is now configurable in .network
12141 files. There's also new support for configuring IP source
12142 routing. networkd .link files gained support for a new
12143 OriginalName= match that is useful to match against the
12144 original interface name the kernel assigned. .network files
12145 may include MTU= and MACAddress= fields for altering the MTU
12146 and MAC address while being connected to a specific network
12149 * The LUKS logic gained supported for configuring
12150 UUID-specific key files. There's also new support for naming
12151 LUKS device from the kernel command line, using the new
12152 luks.name= argument.
12154 * Timer units may now be transiently created via the bus API
12155 (this was previously already available for scope and service
12156 units). In addition it is now possible to create multiple
12157 transient units at the same time with a single bus call. The
12158 "systemd-run" tool has been updated to make use of this for
12159 running commands on a specified time, in at(1)-style.
12161 * tmpfiles gained support for "t" lines, for assigning
12162 extended attributes to files. Among other uses this may be
12163 used to assign SMACK labels to files.
12165 Contributions from: Alin Rauta, Alison Chaiken, Andrej
12166 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
12167 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
12168 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
12169 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
12170 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
12171 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
12172 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
12173 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
12174 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
12175 Michael Biebl, Michael Chapman, Michael Marineau, Michal
12176 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
12177 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
12178 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
12179 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
12180 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
12181 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
12182 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
12184 — Berlin, 2014-12-10
12188 * journalctl gained the new options -t/--identifier= to match
12189 on the syslog identifier (aka "tag"), as well as --utc to
12190 show log timestamps in the UTC timezone. journalctl now also
12191 accepts -n/--lines=all to disable line capping in a pager.
12193 * journalctl gained a new switch, --flush, that synchronously
12194 flushes logs from /run/log/journal to /var/log/journal if
12195 persistent storage is enabled. systemd-journal-flush.service
12196 now waits until the operation is complete.
12198 * Services can notify the manager before they start a reload
12199 (by sending RELOADING=1) or shutdown (by sending
12200 STOPPING=1). This allows the manager to track and show the
12201 internal state of daemons and closes a race condition when
12202 the process is still running but has closed its D-Bus
12205 * Services with Type=oneshot do not have to have any ExecStart
12208 * User units are now loaded also from
12209 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
12210 /run/systemd/user directory that was already previously
12211 supported, but is under the control of the user.
12213 * Job timeouts (i.e. timeouts on the time a job that is
12214 queued stays in the run queue) can now optionally result in
12215 immediate reboot or power-off actions (JobTimeoutAction= and
12216 JobTimeoutRebootArgument=). This is useful on ".target"
12217 units, to limit the maximum time a target remains
12218 undispatched in the run queue, and to trigger an emergency
12219 operation in such a case. This is now used by default to
12220 turn off the system if boot-up (as defined by everything in
12221 basic.target) hangs and does not complete for at least
12222 15min. Also, if power-off or reboot hang for at least 30min
12223 an immediate power-off/reboot operation is triggered. This
12224 functionality is particularly useful to increase reliability
12225 on embedded devices, but also on laptops which might
12226 accidentally get powered on when carried in a backpack and
12227 whose boot stays stuck in a hard disk encryption passphrase
12230 * systemd-logind can be configured to also handle lid switch
12231 events even when the machine is docked or multiple displays
12232 are attached (HandleLidSwitchDocked= option).
12234 * A helper binary and a service have been added which can be
12235 used to resume from hibernation in the initramfs. A
12236 generator will parse the resume= option on the kernel
12237 command line to trigger resume.
12239 * A user console daemon systemd-consoled has been
12240 added. Currently, it is a preview, and will so far open a
12241 single terminal on each session of the user marked as
12242 Desktop=systemd-console.
12244 * Route metrics can be specified for DHCP routes added by
12247 * The SELinux context of socket-activated services can be set
12248 from the information provided by the networking stack
12249 (SELinuxContextFromNet= option).
12251 * Userspace firmware loading support has been removed and
12252 the minimum supported kernel version is thus bumped to 3.7.
12254 * Timeout for udev workers has been increased from 1 to 3
12255 minutes, but a warning will be printed after 1 minute to
12256 help diagnose kernel modules that take a long time to load.
12258 * Udev rules can now remove tags on devices with TAG-="foobar".
12260 * systemd's readahead implementation has been removed. In many
12261 circumstances it didn't give expected benefits even for
12262 rotational disk drives and was becoming less relevant in the
12263 age of SSDs. As none of the developers has been using
12264 rotating media anymore, and nobody stepped up to actively
12265 maintain this component of systemd it has now been removed.
12267 * Swap units can use Options= to specify discard options.
12268 Discard options specified for swaps in /etc/fstab are now
12271 * Docker containers are now detected as a separate type of
12274 * The Password Agent protocol gained support for queries where
12275 the user input is shown, useful e.g. for user names.
12276 systemd-ask-password gained a new --echo option to turn that
12279 * The default sysctl.d/ snippets will now set:
12281 net.core.default_qdisc = fq_codel
12283 This selects Fair Queuing Controlled Delay as the default
12284 queuing discipline for network interfaces. fq_codel helps
12285 fight the network bufferbloat problem. It is believed to be
12286 a good default with no tuning required for most workloads.
12287 Downstream distributions may override this choice. On 10Gbit
12288 servers that do not do forwarding, "fq" may perform better.
12289 Systems without a good clocksource should use "pfifo_fast".
12291 * If kdbus is enabled during build a new option BusPolicy= is
12292 available for service units, that allows locking all service
12293 processes into a stricter bus policy, in order to limit
12294 access to various bus services, or even hide most of them
12295 from the service's view entirely.
12297 * networkctl will now show the .network and .link file
12298 networkd has applied to a specific interface.
12300 * sd-login gained a new API call sd_session_get_desktop() to
12301 query which desktop environment has been selected for a
12304 * UNIX utmp support is now compile-time optional to support
12305 legacy-free systems.
12307 * systemctl gained two new commands "add-wants" and
12308 "add-requires" for pulling in units from specific targets
12311 * If the word "rescue" is specified on the kernel command line
12312 the system will now boot into rescue mode (aka
12313 rescue.target), which was previously available only by
12314 specifying "1" or "systemd.unit=rescue.target" on the kernel
12315 command line. This new kernel command line option nicely
12316 mirrors the already existing "emergency" kernel command line
12319 * New kernel command line options mount.usr=, mount.usrflags=,
12320 mount.usrfstype= have been added that match root=, rootflags=,
12321 rootfstype= but allow mounting a specific file system to
12324 * The $NOTIFY_SOCKET is now also passed to control processes of
12325 services, not only the main process.
12327 * This version reenables support for fsck's -l switch. This
12328 means at least version v2.25 of util-linux is required for
12329 operation, otherwise dead-locks on device nodes may
12330 occur. Again: you need to update util-linux to at least
12331 v2.25 when updating systemd to v217.
12333 * The "multi-seat-x" tool has been removed from systemd, as
12334 its functionality has been integrated into X servers 1.16,
12335 and the tool is hence redundant. It is recommended to update
12336 display managers invoking this tool to simply invoke X
12337 directly from now on, again.
12339 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
12340 message flag has been added for all of systemd's polkit
12341 authenticated method calls has been added. In particular this
12342 now allows optional interactive authorization via polkit for
12343 many of PID1's privileged operations such as unit file
12344 enabling and disabling.
12346 * "udevadm hwdb --update" learnt a new switch "--usr" for
12347 placing the rebuilt hardware database in /usr instead of
12348 /etc. When used only hardware database entries stored in
12349 /usr will be used, and any user database entries in /etc are
12350 ignored. This functionality is useful for vendors to ship a
12351 pre-built database on systems where local configuration is
12352 unnecessary or unlikely.
12354 * Calendar time specifications in .timer units now also
12355 understand the strings "semi-annually", "quarterly" and
12356 "minutely" as shortcuts (in addition to the preexisting
12357 "annually", "hourly", …).
12359 * systemd-tmpfiles will now correctly create files in /dev
12360 at boot which are marked for creation only at boot. It is
12361 recommended to always create static device nodes with 'c!'
12362 and 'b!', so that they are created only at boot and not
12363 overwritten at runtime.
12365 * When the watchdog logic is used for a service (WatchdogSec=)
12366 and the watchdog timeout is hit the service will now be
12367 terminated with SIGABRT (instead of just SIGTERM), in order
12368 to make sure a proper coredump and backtrace is
12369 generated. This ensures that hanging services will result in
12370 similar coredump/backtrace behaviour as services that hit a
12371 segmentation fault.
12373 Contributions from: Andreas Henriksson, Andrei Borzenkov,
12374 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
12375 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
12376 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
12377 Herrmann, David Sommerseth, David Strauss, Emil Renner
12378 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
12379 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
12380 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
12381 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
12382 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
12383 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
12384 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
12385 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
12386 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
12387 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
12388 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
12389 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
12390 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
12391 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
12392 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
12393 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
12396 — Berlin, 2014-10-28
12400 * timedated no longer reads NTP implementation unit names from
12401 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
12402 implementations should add a
12404 Conflicts=systemd-timesyncd.service
12406 to their unit files to take over and replace systemd's NTP
12407 default functionality.
12409 * systemd-sysusers gained a new line type "r" for configuring
12410 which UID/GID ranges to allocate system users/groups
12411 from. Lines of type "u" may now add an additional column
12412 that specifies the home directory for the system user to be
12413 created. Also, systemd-sysusers may now optionally read user
12414 information from STDIN instead of a file. This is useful for
12415 invoking it from RPM preinst scriptlets that need to create
12416 users before the first RPM file is installed since these
12417 files might need to be owned by them. A new
12418 %sysusers_create_inline RPM macro has been introduced to do
12419 just that. systemd-sysusers now updates the shadow files as
12420 well as the user/group databases, which should enhance
12421 compatibility with certain tools like grpck.
12423 * A number of bus APIs of PID 1 now optionally consult polkit to
12424 permit access for otherwise unprivileged clients under certain
12425 conditions. Note that this currently doesn't support
12426 interactive authentication yet, but this is expected to be
12427 added eventually, too.
12429 * /etc/machine-info now has new fields for configuring the
12430 deployment environment of the machine, as well as the
12431 location of the machine. hostnamectl has been updated with
12432 new command to update these fields.
12434 * systemd-timesyncd has been updated to automatically acquire
12435 NTP server information from systemd-networkd, which might
12436 have been discovered via DHCP.
12438 * systemd-resolved now includes a caching DNS stub resolver
12439 and a complete LLMNR name resolution implementation. A new
12440 NSS module "nss-resolve" has been added which can be used
12441 instead of glibc's own "nss-dns" to resolve hostnames via
12442 systemd-resolved. Hostnames, addresses and arbitrary RRs may
12443 be resolved via systemd-resolved D-Bus APIs. In contrast to
12444 the glibc internal resolver systemd-resolved is aware of
12445 multi-homed system, and keeps DNS server and caches separate
12446 and per-interface. Queries are sent simultaneously on all
12447 interfaces that have DNS servers configured, in order to
12448 properly handle VPNs and local LANs which might resolve
12449 separate sets of domain names. systemd-resolved may acquire
12450 DNS server information from systemd-networkd automatically,
12451 which in turn might have discovered them via DHCP. A tool
12452 "systemd-resolve-host" has been added that may be used to
12453 query the DNS logic in resolved. systemd-resolved implements
12454 IDNA and automatically uses IDNA or UTF-8 encoding depending
12455 on whether classic DNS or LLMNR is used as transport. In the
12456 next releases we intend to add a DNSSEC and mDNS/DNS-SD
12457 implementation to systemd-resolved.
12459 * A new NSS module nss-mymachines has been added, that
12460 automatically resolves the names of all local registered
12461 containers to their respective IP addresses.
12463 * A new client tool "networkctl" for systemd-networkd has been
12464 added. It currently is entirely passive and will query
12465 networking configuration from udev, rtnetlink and networkd,
12466 and present it to the user in a very friendly
12467 way. Eventually, we hope to extend it to become a full
12468 control utility for networkd.
12470 * .socket units gained a new DeferAcceptSec= setting that
12471 controls the kernels' TCP_DEFER_ACCEPT sockopt for
12472 TCP. Similarly, support for controlling TCP keep-alive
12473 settings has been added (KeepAliveTimeSec=,
12474 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
12475 turning off Nagle's algorithm on TCP has been added
12478 * logind learned a new session type "web", for use in projects
12479 like Cockpit which register web clients as PAM sessions.
12481 * timer units with at least one OnCalendar= setting will now
12482 be started only after time-sync.target has been
12483 reached. This way they will not elapse before the system
12484 clock has been corrected by a local NTP client or
12485 similar. This is particular useful on RTC-less embedded
12486 machines, that come up with an invalid system clock.
12488 * systemd-nspawn's --network-veth= switch should now result in
12489 stable MAC addresses for both the outer and the inner side
12492 * systemd-nspawn gained a new --volatile= switch for running
12493 container instances with /etc or /var unpopulated.
12495 * The kdbus client code has been updated to use the new Linux
12496 3.17 memfd subsystem instead of the old kdbus-specific one.
12498 * systemd-networkd's DHCP client and server now support
12499 FORCERENEW. There are also new configuration options to
12500 configure the vendor client identifier and broadcast mode
12503 * systemd will no longer inform the kernel about the current
12504 timezone, as this is necessarily incorrect and racy as the
12505 kernel has no understanding of DST and similar
12506 concepts. This hence means FAT timestamps will be always
12507 considered UTC, similar to what Android is already
12508 doing. Also, when the RTC is configured to the local time
12509 (rather than UTC) systemd will never synchronize back to it,
12510 as this might confuse Windows at a later boot.
12512 * systemd-analyze gained a new command "verify" for offline
12513 validation of unit files.
12515 * systemd-networkd gained support for a couple of additional
12516 settings for bonding networking setups. Also, the metric for
12517 statically configured routes may now be configured. For
12518 network interfaces where this is appropriate the peer IP
12519 address may now be configured.
12521 * systemd-networkd's DHCP client will no longer request
12522 broadcasting by default, as this tripped up some networks.
12523 For hardware where broadcast is required the feature should
12524 be switched back on using RequestBroadcast=yes.
12526 * systemd-networkd will now set up IPv4LL addresses (when
12527 enabled) even if DHCP is configured successfully.
12529 * udev will now default to respect network device names given
12530 by the kernel when the kernel indicates that these are
12531 predictable. This behavior can be tweaked by changing
12532 NamePolicy= in the relevant .link file.
12534 * A new library systemd-terminal has been added that
12535 implements full TTY stream parsing and rendering. This
12536 library is supposed to be used later on for implementing a
12537 full userspace VT subsystem, replacing the current kernel
12540 * A new tool systemd-journal-upload has been added to push
12541 journal data to a remote system running
12542 systemd-journal-remote.
12544 * journald will no longer forward all local data to another
12545 running syslog daemon. This change has been made because
12546 rsyslog (which appears to be the most commonly used syslog
12547 implementation these days) no longer makes use of this, and
12548 instead pulls the data out of the journal on its own. Since
12549 forwarding the messages to a non-existent syslog server is
12550 more expensive than we assumed we have now turned this
12551 off. If you run a syslog server that is not a recent rsyslog
12552 version, you have to turn this option on again
12553 (ForwardToSyslog= in journald.conf).
12555 * journald now optionally supports the LZ4 compressor for
12556 larger journal fields. This compressor should perform much
12557 better than XZ which was the previous default.
12559 * machinectl now shows the IP addresses of local containers,
12560 if it knows them, plus the interface name of the container.
12562 * A new tool "systemd-escape" has been added that makes it
12563 easy to escape strings to build unit names and similar.
12565 * sd_notify() messages may now include a new ERRNO= field
12566 which is parsed and collected by systemd and shown among the
12567 "systemctl status" output for a service.
12569 * A new component "systemd-firstboot" has been added that
12570 queries the most basic systemd information (timezone,
12571 hostname, root password) interactively on first
12572 boot. Alternatively it may also be used to provision these
12573 things offline on OS images installed into directories.
12575 * The default sysctl.d/ snippets will now set
12577 net.ipv4.conf.default.promote_secondaries=1
12579 This has the benefit of no flushing secondary IP addresses
12580 when primary addresses are removed.
12582 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
12583 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
12584 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
12585 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
12586 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
12587 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
12588 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
12589 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
12590 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
12591 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
12592 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
12593 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
12594 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
12595 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
12596 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
12598 — Berlin, 2014-08-19
12602 * A new tool systemd-sysusers has been added. This tool
12603 creates system users and groups in /etc/passwd and
12604 /etc/group, based on static declarative system user/group
12605 definitions in /usr/lib/sysusers.d/. This is useful to
12606 enable factory resets and volatile systems that boot up with
12607 an empty /etc directory, and thus need system users and
12608 groups created during early boot. systemd now also ships
12609 with two default sysusers.d/ files for the most basic
12610 users and groups systemd and the core operating system
12613 * A new tmpfiles snippet has been added that rebuilds the
12614 essential files in /etc on boot, should they be missing.
12616 * A directive for ensuring automatic clean-up of
12617 /var/cache/man/ has been removed from the default
12618 configuration. This line should now be shipped by the man
12619 implementation. The necessary change has been made to the
12620 man-db implementation. Note that you need to update your man
12621 implementation to one that ships this line, otherwise no
12622 automatic clean-up of /var/cache/man will take place.
12624 * A new condition ConditionNeedsUpdate= has been added that
12625 may conditionalize services to only run when /etc or /var
12626 are "older" than the vendor operating system resources in
12627 /usr. This is useful for reconstructing or updating /etc
12628 after an offline update of /usr or a factory reset, on the
12629 next reboot. Services that want to run once after such an
12630 update or reset should use this condition and order
12631 themselves before the new systemd-update-done.service, which
12632 will mark the two directories as fully updated. A number of
12633 service files have been added making use of this, to rebuild
12634 the udev hardware database, the journald message catalog and
12635 dynamic loader cache (ldconfig). The systemd-sysusers tool
12636 described above also makes use of this now. With this in
12637 place it is now possible to start up a minimal operating
12638 system with /etc empty cleanly. For more information on the
12639 concepts involved see this recent blog story:
12641 https://0pointer.de/blog/projects/stateless.html
12643 * A new system group "input" has been introduced, and all
12644 input device nodes get this group assigned. This is useful
12645 for system-level software to get access to input devices. It
12646 complements what is already done for "audio" and "video".
12648 * systemd-networkd learnt minimal DHCPv4 server support in
12649 addition to the existing DHCPv4 client support. It also
12650 learnt DHCPv6 client and IPv6 Router Solicitation client
12651 support. The DHCPv4 client gained support for static routes
12652 passed in from the server. Note that the [DHCPv4] section
12653 known in older systemd-networkd versions has been renamed to
12654 [DHCP] and is now also used by the DHCPv6 client. Existing
12655 .network files using settings of this section should be
12656 updated, though compatibility is maintained. Optionally, the
12657 client hostname may now be sent to the DHCP server.
12659 * networkd gained support for vxlan virtual networks as well
12660 as tun/tap and dummy devices.
12662 * networkd gained support for automatic allocation of address
12663 ranges for interfaces from a system-wide pool of
12664 addresses. This is useful for dynamically managing a large
12665 number of interfaces with a single network configuration
12666 file. In particular this is useful to easily assign
12667 appropriate IP addresses to the veth links of a large number
12668 of nspawn instances.
12670 * RPM macros for processing sysusers, sysctl and binfmt
12671 drop-in snippets at package installation time have been
12674 * The /etc/os-release file should now be placed in
12675 /usr/lib/os-release. The old location is automatically
12676 created as symlink. /usr/lib is the more appropriate
12677 location of this file, since it shall actually describe the
12678 vendor operating system shipped in /usr, and not the
12679 configuration stored in /etc.
12681 * .mount units gained a new boolean SloppyOptions= setting
12682 that maps to mount(8)'s -s option which enables permissive
12683 parsing of unknown mount options.
12685 * tmpfiles learnt a new "L+" directive which creates a symlink
12686 but (unlike "L") deletes a pre-existing file first, should
12687 it already exist and not already be the correct
12688 symlink. Similarly, "b+", "c+" and "p+" directives have been
12689 added as well, which create block and character devices, as
12690 well as fifos in the filesystem, possibly removing any
12691 pre-existing files of different types.
12693 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
12694 'argument' field (which so far specified the source to
12695 symlink/copy the files from) is now optional. If omitted the
12696 same file os copied from /usr/share/factory/ suffixed by the
12697 full destination path. This is useful for populating /etc
12698 with essential files, by copying them from vendor defaults
12699 shipped in /usr/share/factory/etc.
12701 * A new command "systemctl preset-all" has been added that
12702 applies the service preset settings to all installed unit
12703 files. A new switch --preset-mode= has been added that
12704 controls whether only enable or only disable operations
12707 * A new command "systemctl is-system-running" has been added
12708 that allows checking the overall state of the system, for
12709 example whether it is fully up and running.
12711 * When the system boots up with an empty /etc, the equivalent
12712 to "systemctl preset-all" is executed during early boot, to
12713 make sure all default services are enabled after a factory
12716 * systemd now contains a minimal preset file that enables the
12717 most basic services systemd ships by default.
12719 * Unit files' [Install] section gained a new DefaultInstance=
12720 field for defining the default instance to create if a
12721 template unit is enabled with no instance specified.
12723 * A new passive target cryptsetup-pre.target has been added
12724 that may be used by services that need to make they run and
12725 finish before the first LUKS cryptographic device is set up.
12727 * The /dev/loop-control and /dev/btrfs-control device nodes
12728 are now owned by the "disk" group by default, opening up
12729 access to this group.
12731 * systemd-coredump will now automatically generate a
12732 stack trace of all core dumps taking place on the system,
12733 based on elfutils' libdw library. This stack trace is logged
12736 * systemd-coredump may now optionally store coredumps directly
12737 on disk (in /var/lib/systemd/coredump, possibly compressed),
12738 instead of storing them unconditionally in the journal. This
12739 mode is the new default. A new configuration file
12740 /etc/systemd/coredump.conf has been added to configure this
12741 and other parameters of systemd-coredump.
12743 * coredumpctl gained a new "info" verb to show details about a
12744 specific coredump. A new switch "-1" has also been added
12745 that makes sure to only show information about the most
12746 recent entry instead of all entries. Also, as the tool is
12747 generally useful now the "systemd-" prefix of the binary
12748 name has been removed. Distributions that want to maintain
12749 compatibility with the old name should add a symlink from
12750 the old name to the new name.
12752 * journald's SplitMode= now defaults to "uid". This makes sure
12753 that unprivileged users can access their own coredumps with
12754 coredumpctl without restrictions.
12756 * New kernel command line options "systemd.wants=" (for
12757 pulling an additional unit during boot), "systemd.mask="
12758 (for masking a specific unit for the boot), and
12759 "systemd.debug-shell" (for enabling the debug shell on tty9)
12760 have been added. This is implemented in the new generator
12761 "systemd-debug-generator".
12763 * systemd-nspawn will now by default filter a couple of
12764 syscalls for containers, among them those required for
12765 kernel module loading, direct x86 IO port access, swap
12766 management, and kexec. Most importantly though
12767 open_by_handle_at() is now prohibited for containers,
12768 closing a hole similar to a recently discussed vulnerability
12769 in docker regarding access to files on file hierarchies the
12770 container should normally not have access to. Note that, for
12771 nspawn, we generally make no security claims anyway (and
12772 this is explicitly documented in the man page), so this is
12773 just a fix for one of the most obvious problems.
12775 * A new man page file-hierarchy(7) has been added that
12776 contains a minimized, modernized version of the file system
12777 layout systemd expects, similar in style to the FHS
12778 specification or hier(5). A new tool systemd-path(1) has
12779 been added to query many of these paths for the local
12782 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
12783 longer done. Since the directory now has a per-user size
12784 limit, and is cleaned on logout this appears unnecessary,
12785 in particular since this now brings the lifecycle of this
12786 directory closer in line with how IPC objects are handled.
12788 * systemd.pc now exports a number of additional directories,
12789 including $libdir (which is useful to identify the library
12790 path for the primary architecture of the system), and a
12791 couple of drop-in directories.
12793 * udev's predictable network interface names now use the dev_port
12794 sysfs attribute, introduced in linux 3.15 instead of dev_id to
12795 distinguish between ports of the same PCI function. dev_id should
12796 only be used for ports using the same HW address, hence the need
12799 * machined has been updated to export the OS version of a
12800 container (read from /etc/os-release and
12801 /usr/lib/os-release) on the bus. This is now shown in
12802 "machinectl status" for a machine.
12804 * A new service setting RestartForceExitStatus= has been
12805 added. If configured to a set of exit signals or process
12806 return values, the service will be restarted when the main
12807 daemon process exits with any of them, regardless of the
12810 * systemctl's -H switch for connecting to remote systemd
12811 machines has been extended so that it may be used to
12812 directly connect to a specific container on the
12813 host. "systemctl -H root@foobar:waldi" will now connect as
12814 user "root" to host "foobar", and then proceed directly to
12815 the container named "waldi". Note that currently you have to
12816 authenticate as user "root" for this to work, as entering
12817 containers is a privileged operation.
12819 Contributions from: Andreas Henriksson, Benjamin Steinwender,
12820 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
12821 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
12822 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
12823 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
12824 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
12825 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
12826 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
12827 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
12828 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
12829 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
12830 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
12832 — Berlin, 2014-07-03
12836 * As an experimental feature, udev now tries to lock the
12837 disk device node (flock(LOCK_SH|LOCK_NB)) while it
12838 executes events for the disk or any of its partitions.
12839 Applications like partitioning programs can lock the
12840 disk device node (flock(LOCK_EX)) and claim temporary
12841 device ownership that way; udev will entirely skip all event
12842 handling for this disk and its partitions. If the disk
12843 was opened for writing, the close will trigger a partition
12844 table rescan in udev's "watch" facility, and if needed
12845 synthesize "change" events for the disk and all its partitions.
12846 This is now unconditionally enabled, and if it turns out to
12847 cause major problems, we might turn it on only for specific
12848 devices, or might need to disable it entirely. Device Mapper
12849 devices are excluded from this logic.
12851 * We temporarily dropped the "-l" switch for fsck invocations,
12852 since they collide with the flock() logic above. util-linux
12853 upstream has been changed already to avoid this conflict,
12854 and we will re-add "-l" as soon as util-linux with this
12855 change has been released.
12857 * The dependency on libattr has been removed. Since a long
12858 time, the extended attribute calls have moved to glibc, and
12859 libattr is thus unnecessary.
12861 * Virtualization detection works without privileges now. This
12862 means the systemd-detect-virt binary no longer requires
12863 CAP_SYS_PTRACE file capabilities, and our daemons can run
12864 with fewer privileges.
12866 * systemd-networkd now runs under its own "systemd-network"
12867 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
12868 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
12869 loses the ability to write to files owned by root this way.
12871 * Similarly, systemd-resolved now runs under its own
12872 "systemd-resolve" user with no capabilities remaining.
12874 * Similarly, systemd-bus-proxyd now runs under its own
12875 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
12877 * systemd-networkd gained support for setting up "veth"
12878 virtual Ethernet devices for container connectivity, as well
12879 as GRE and VTI tunnels.
12881 * systemd-networkd will no longer automatically attempt to
12882 manually load kernel modules necessary for certain tunnel
12883 transports. Instead, it is assumed the kernel loads them
12884 automatically when required. This only works correctly on
12885 very new kernels. On older kernels, please consider adding
12886 the kernel modules to /etc/modules-load.d/ as a work-around.
12888 * The resolv.conf file systemd-resolved generates has been
12889 moved to /run/systemd/resolve/. If you have a symlink from
12890 /etc/resolv.conf, it might be necessary to correct it.
12892 * Two new service settings, ProtectHome= and ProtectSystem=,
12893 have been added. When enabled, they will make the user data
12894 (such as /home) inaccessible or read-only and the system
12895 (such as /usr) read-only, for specific services. This allows
12896 very light-weight per-service sandboxing to avoid
12897 modifications of user data or system files from
12898 services. These two new switches have been enabled for all
12899 of systemd's long-running services, where appropriate.
12901 * Socket units gained new SocketUser= and SocketGroup=
12902 settings to set the owner user and group of AF_UNIX sockets
12903 and FIFOs in the file system.
12905 * Socket units gained a new RemoveOnStop= setting. If enabled,
12906 all FIFOS and sockets in the file system will be removed
12907 when the specific socket unit is stopped.
12909 * Socket units gained a new Symlinks= setting. It takes a list
12910 of symlinks to create to file system sockets or FIFOs
12911 created by the specific Unix sockets. This is useful to
12912 manage symlinks to socket nodes with the same lifecycle as
12915 * The /dev/log socket and /dev/initctl FIFO have been moved to
12916 /run, and have been replaced by symlinks. This allows
12917 connecting to these facilities even if PrivateDevices=yes is
12918 used for a service (which makes /dev/log itself unavailable,
12919 but /run is left). This also has the benefit of ensuring
12920 that /dev only contains device nodes, directories and
12921 symlinks, and nothing else.
12923 * sd-daemon gained two new calls sd_pid_notify() and
12924 sd_pid_notifyf(). They are similar to sd_notify() and
12925 sd_notifyf(), but allow overriding of the source PID of
12926 notification messages if permissions permit this. This is
12927 useful to send notify messages on behalf of a different
12928 process (for example, the parent process). The
12929 systemd-notify tool has been updated to make use of this
12930 when sending messages (so that notification messages now
12931 originate from the shell script invoking systemd-notify and
12932 not the systemd-notify process itself. This should minimize
12933 a race where systemd fails to associate notification
12934 messages to services when the originating process already
12937 * A new "on-abnormal" setting for Restart= has been added. If
12938 set, it will result in automatic restarts on all "abnormal"
12939 reasons for a process to exit, which includes unclean
12940 signals, core dumps, timeouts and watchdog timeouts, but
12941 does not include clean and unclean exit codes or clean
12942 signals. Restart=on-abnormal is an alternative for
12943 Restart=on-failure for services that shall be able to
12944 terminate and avoid restarts on certain errors, by
12945 indicating so with an unclean exit code. Restart=on-failure
12946 or Restart=on-abnormal is now the recommended setting for
12947 all long-running services.
12949 * If the InaccessibleDirectories= service setting points to a
12950 mount point (or if there are any submounts contained within
12951 it), it is now attempted to completely unmount it, to make
12952 the file systems truly unavailable for the respective
12955 * The ReadOnlyDirectories= service setting and
12956 systemd-nspawn's --read-only parameter are now recursively
12957 applied to all submounts, too.
12959 * Mount units may now be created transiently via the bus APIs.
12961 * The support for SysV and LSB init scripts has been removed
12962 from the systemd daemon itself. Instead, it is now
12963 implemented as a generator that creates native systemd units
12964 from these scripts when needed. This enables us to remove a
12965 substantial amount of legacy code from PID 1, following the
12966 fact that many distributions only ship a very small number
12967 of LSB/SysV init scripts nowadays.
12969 * Privileged Xen (dom0) domains are not considered
12970 virtualization anymore by the virtualization detection
12971 logic. After all, they generally have unrestricted access to
12972 the hardware and usually are used to manage the unprivileged
12975 * systemd-tmpfiles gained a new "C" line type, for copying
12976 files or entire directories.
12978 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
12979 lines. So far, they have been non-globbing versions of the
12980 latter, and have thus been redundant. In future, it is
12981 recommended to only use "z". "m" has hence been removed
12982 from the documentation, even though it stays supported.
12984 * A tmpfiles snippet to recreate the most basic structure in
12985 /var has been added. This is enough to create the /var/run →
12986 /run symlink and create a couple of structural
12987 directories. This allows systems to boot up with an empty or
12988 volatile /var. Of course, while with this change, the core OS
12989 now is capable with dealing with a volatile /var, not all
12990 user services are ready for it. However, we hope that sooner
12991 or later, many service daemons will be changed upstream so
12992 that they are able to automatically create their necessary
12993 directories in /var at boot, should they be missing. This is
12994 the first step to allow state-less systems that only require
12995 the vendor image for /usr to boot.
12997 * systemd-nspawn has gained a new --tmpfs= switch to mount an
12998 empty tmpfs instance to a specific directory. This is
12999 particularly useful for making use of the automatic
13000 reconstruction of /var (see above), by passing --tmpfs=/var.
13002 * Access modes specified in tmpfiles snippets may now be
13003 prefixed with "~", which indicates that they shall be masked
13004 by whether the existing file or directory is currently
13005 writable, readable or executable at all. Also, if specified,
13006 the sgid/suid/sticky bits will be masked for all
13009 * A new passive target unit "network-pre.target" has been
13010 added which is useful for services that shall run before any
13011 network is configured, for example firewall scripts.
13013 * The "floppy" group that previously owned the /dev/fd*
13014 devices is no longer used. The "disk" group is now used
13015 instead. Distributions should probably deprecate usage of
13018 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
13019 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
13020 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
13021 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
13022 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
13023 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
13026 — Berlin, 2014-06-11
13030 * A new "systemd-timesyncd" daemon has been added for
13031 synchronizing the system clock across the network. It
13032 implements an SNTP client. In contrast to NTP
13033 implementations such as chrony or the NTP reference server,
13034 this only implements a client side, and does not bother with
13035 the full NTP complexity, focusing only on querying time from
13036 one remote server and synchronizing the local clock to
13037 it. Unless you intend to serve NTP to networked clients or
13038 want to connect to local hardware clocks, this simple NTP
13039 client should be more than appropriate for most
13040 installations. The daemon runs with minimal privileges, and
13041 has been hooked up with networkd to only operate when
13042 network connectivity is available. The daemon saves the
13043 current clock to disk every time a new NTP sync has been
13044 acquired, and uses this to possibly correct the system clock
13045 early at bootup, in order to accommodate for systems that
13046 lack an RTC such as the Raspberry Pi and embedded devices,
13047 and to make sure that time monotonically progresses on these
13048 systems, even if it is not always correct. To make use of
13049 this daemon, a new system user and group "systemd-timesync"
13050 needs to be created on installation of systemd.
13052 * The queue "seqnum" interface of libudev has been disabled, as
13053 it was generally incompatible with device namespacing as
13054 sequence numbers of devices go "missing" if the devices are
13055 part of a different namespace.
13057 * "systemctl list-timers" and "systemctl list-sockets" gained
13058 a --recursive switch for showing units of these types also
13059 for all local containers, similar in style to the already
13060 supported --recursive switch for "systemctl list-units".
13062 * A new RebootArgument= setting has been added for service
13063 units, which may be used to specify a kernel reboot argument
13064 to use when triggering reboots with StartLimitAction=.
13066 * A new FailureAction= setting has been added for service
13067 units which may be used to specify an operation to trigger
13068 when a service fails. This works similarly to
13069 StartLimitAction=, but unlike it, controls what is done
13070 immediately rather than only after several attempts to
13071 restart the service in question.
13073 * hostnamed got updated to also expose the kernel name,
13074 release, and version on the bus. This is useful for
13075 executing commands like hostnamectl with the -H switch.
13076 systemd-analyze makes use of this to properly display
13077 details when running non-locally.
13079 * The bootchart tool can now show cgroup information in the
13080 graphs it generates.
13082 * The CFS CPU quota cgroup attribute is now exposed for
13083 services. The new CPUQuota= switch has been added for this
13084 which takes a percentage value. Setting this will have the
13085 result that a service may never get more CPU time than the
13086 specified percentage, even if the machine is otherwise idle.
13088 * systemd-networkd learned IPIP and SIT tunnel support.
13090 * LSB init scripts exposing a dependency on $network will now
13091 get a dependency on network-online.target rather than simply
13092 network.target. This should bring LSB handling closer to
13093 what it was on SysV systems.
13095 * A new fsck.repair= kernel option has been added to control
13096 how fsck shall deal with unclean file systems at boot.
13098 * The (.ini) configuration file parser will now silently ignore
13099 sections whose names begin with "X-". This may be used to maintain
13100 application-specific extension sections in unit files.
13102 * machined gained a new API to query the IP addresses of
13103 registered containers. "machinectl status" has been updated
13104 to show these addresses in its output.
13106 * A new call sd_uid_get_display() has been added to the
13107 sd-login APIs for querying the "primary" session of a
13108 user. The "primary" session of the user is elected from the
13109 user's sessions and generally a graphical session is
13110 preferred over a text one.
13112 * A minimal systemd-resolved daemon has been added. It
13113 currently simply acts as a companion to systemd-networkd and
13114 manages resolv.conf based on per-interface DNS
13115 configuration, possibly supplied via DHCP. In the long run
13116 we hope to extend this into a local DNSSEC enabled DNS and
13119 * The systemd-networkd-wait-online tool is now enabled by
13120 default. It will delay network-online.target until a network
13121 connection has been configured. The tool primarily integrates
13122 with networkd, but will also make a best effort to make sense
13123 of network configuration performed in some other way.
13125 * Two new service options StartupCPUShares= and
13126 StartupBlockIOWeight= have been added that work similarly to
13127 CPUShares= and BlockIOWeight= however only apply during
13128 system startup. This is useful to prioritize certain services
13129 differently during bootup than during normal runtime.
13131 * hostnamed has been changed to prefer the statically
13132 configured hostname in /etc/hostname (unless set to
13133 'localhost' or empty) over any dynamic one supplied by
13134 dhcp. With this change, the rules for picking the hostname
13135 match more closely the rules of other configuration settings
13136 where the local administrator's configuration in /etc always
13137 overrides any other settings.
13139 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
13140 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
13141 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
13142 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
13143 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
13144 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
13145 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
13146 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
13147 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
13148 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
13149 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
13150 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
13151 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
13152 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
13153 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
13154 Lindskog, WaLyong Cho, Will Woods, Zbigniew
13157 — Beijing, 2014-05-28
13161 * When restoring the screen brightness at boot, stay away from
13162 the darkest setting or from the lowest 5% of the available
13163 range, depending on which is the larger value of both. This
13164 should effectively protect the user from rebooting into a
13165 black screen, should the brightness have been set to minimum
13168 * sd-login gained a new sd_machine_get_class() call to
13169 determine the class ("vm" or "container") of a machine
13170 registered with machined.
13172 * sd-login gained new calls
13173 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
13174 to query the identity of the peer of a local AF_UNIX
13175 connection. They operate similarly to their sd_pid_get_xyz()
13178 * PID 1 will now maintain a system-wide system state engine
13179 with the states "starting", "running", "degraded",
13180 "maintenance", "stopping". These states are bound to system
13181 startup, normal runtime, runtime with at least one failed
13182 service, rescue/emergency mode and system shutdown. This
13183 state is shown in the "systemctl status" output when no unit
13184 name is passed. It is useful to determine system state, in
13185 particularly when doing so for many systems or containers at
13188 * A new command "list-machines" has been added to "systemctl"
13189 that lists all local OS containers and shows their system
13190 state (see above), if systemd runs inside of them.
13192 * systemctl gained a new "-r" switch to recursively enumerate
13193 units on all local containers, when used with the
13194 "list-unit" command (which is the default one that is
13195 executed when no parameters are specified).
13197 * The GPT automatic partition discovery logic will now honour
13198 two GPT partition flags: one may be set on a partition to
13199 cause it to be mounted read-only, and the other may be set
13200 on a partition to ignore it during automatic discovery.
13202 * Two new GPT type UUIDs have been added for automatic root
13203 partition discovery, for 32-bit and 64-bit ARM. This is not
13204 particularly useful for discovering the root directory on
13205 these architectures during bare-metal boots (since UEFI is
13206 not common there), but still very useful to allow booting of
13207 ARM disk images in nspawn with the -i option.
13209 * MAC addresses of interfaces created with nspawn's
13210 --network-interface= switch will now be generated from the
13211 machine name, and thus be stable between multiple invocations
13214 * logind will now automatically remove all IPC objects owned
13215 by a user if she or he fully logs out. This makes sure that
13216 users who are logged out cannot continue to consume IPC
13217 resources. This covers SysV memory, semaphores and message
13218 queues as well as POSIX shared memory and message
13219 queues. Traditionally, SysV and POSIX IPC had no lifecycle
13220 limits. With this functionality, that is corrected. This may
13221 be turned off by using the RemoveIPC= switch of logind.conf.
13223 * The systemd-machine-id-setup and tmpfiles tools gained a
13224 --root= switch to operate on a specific root directory,
13227 * journald can now forward logged messages to the TTYs of all
13228 logged in users ("wall"). This is the default for all
13229 emergency messages now.
13231 * A new tool systemd-journal-remote has been added to stream
13232 journal log messages across the network.
13234 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
13235 controller trees are mounted into it. Note that the
13236 directories mounted beneath it are not read-only. This is a
13237 security measure and is particularly useful because glibc
13238 actually includes a search logic to pick any tmpfs it can
13239 find to implement shm_open() if /dev/shm is not available
13240 (which it might very well be in namespaced setups).
13242 * machinectl gained a new "poweroff" command to cleanly power
13243 down a local OS container.
13245 * The PrivateDevices= unit file setting will now also drop the
13246 CAP_MKNOD capability from the capability bound set, and
13247 imply DevicePolicy=closed.
13249 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
13250 comprehensively on all long-running systemd services where
13251 this is appropriate.
13253 * systemd-udevd will now run in a disassociated mount
13254 namespace. To mount directories from udev rules, make sure to
13255 pull in mount units via SYSTEMD_WANTS properties.
13257 * The kdbus support gained support for uploading policy into
13258 the kernel. sd-bus gained support for creating "monitoring"
13259 connections that can eavesdrop into all bus communication
13260 for debugging purposes.
13262 * Timestamps may now be specified in seconds since the UNIX
13263 epoch Jan 1st, 1970 by specifying "@" followed by the value
13266 * Native tcpwrap support in systemd has been removed. tcpwrap
13267 is old code, not really maintained anymore and has serious
13268 shortcomings, and better options such as firewalls
13269 exist. For setups that require tcpwrap usage, please
13270 consider invoking your socket-activated service via tcpd,
13271 like on traditional inetd.
13273 * A new system.conf configuration option
13274 DefaultTimerAccuracySec= has been added that controls the
13275 default AccuracySec= setting of .timer units.
13277 * Timer units gained a new WakeSystem= switch. If enabled,
13278 timers configured this way will cause the system to resume
13279 from system suspend (if the system supports that, which most
13282 * Timer units gained a new Persistent= switch. If enabled,
13283 timers configured this way will save to disk when they have
13284 been last triggered. This information is then used on next
13285 reboot to possible execute overdue timer events, that
13286 could not take place because the system was powered off.
13287 This enables simple anacron-like behaviour for timer units.
13289 * systemctl's "list-timers" will now also list the time a
13290 timer unit was last triggered in addition to the next time
13291 it will be triggered.
13293 * systemd-networkd will now assign predictable IPv4LL
13294 addresses to its local interfaces.
13296 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
13297 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
13298 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
13299 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
13300 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
13301 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
13302 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
13303 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
13306 — Berlin, 2014-03-25
13310 * A new unit file setting RestrictAddressFamilies= has been
13311 added to restrict which socket address families unit
13312 processes gain access to. This takes address family names
13313 like "AF_INET" or "AF_UNIX", and is useful to minimize the
13314 attack surface of services via exotic protocol stacks. This
13315 is built on seccomp system call filters.
13317 * Two new unit file settings RuntimeDirectory= and
13318 RuntimeDirectoryMode= have been added that may be used to
13319 manage a per-daemon runtime directories below /run. This is
13320 an alternative for setting up directory permissions with
13321 tmpfiles snippets, and has the advantage that the runtime
13322 directory's lifetime is bound to the daemon runtime and that
13323 the daemon starts up with an empty directory each time. This
13324 is particularly useful when writing services that drop
13325 privileges using the User= or Group= setting.
13327 * The DeviceAllow= unit setting now supports globbing for
13328 matching against device group names.
13330 * The systemd configuration file system.conf gained new
13331 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
13332 DefaultMemoryAccounting= to globally turn on/off accounting
13333 for specific resources (cgroups) for all units. These
13334 settings may still be overridden individually in each unit
13337 * systemd-gpt-auto-generator is now able to discover /srv and
13338 root partitions in addition to /home and swap partitions. It
13339 also supports LUKS-encrypted partitions now. With this in
13340 place, automatic discovery of partitions to mount following
13341 the Discoverable Partitions Specification
13342 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
13343 is now a lot more complete. This allows booting without
13344 /etc/fstab and without root= on the kernel command line on
13345 systems prepared appropriately.
13347 * systemd-nspawn gained a new --image= switch which allows
13348 booting up disk images and Linux installations on any block
13349 device that follow the Discoverable Partitions Specification
13350 (see above). This means that installations made with
13351 appropriately updated installers may now be started and
13352 deployed using container managers, completely
13353 unmodified. (We hope that libvirt-lxc will add support for
13354 this feature soon, too.)
13356 * systemd-nspawn gained a new --network-macvlan= setting to
13357 set up a private macvlan interface for the
13358 container. Similarly, systemd-networkd gained a new
13359 Kind=macvlan setting in .netdev files.
13361 * systemd-networkd now supports configuring local addresses
13364 * A new tool systemd-network-wait-online has been added to
13365 synchronously wait for network connectivity using
13368 * The sd-bus.h bus API gained a new sd_bus_track object for
13369 tracking the lifecycle of bus peers. Note that sd-bus.h is
13370 still not a public API though (unless you specify
13371 --enable-kdbus on the configure command line, which however
13372 voids your warranty and you get no API stability guarantee).
13374 * The $XDG_RUNTIME_DIR runtime directories for each user are
13375 now individual tmpfs instances, which has the benefit of
13376 introducing separate pools for each user, with individual
13377 size limits, and thus making sure that unprivileged clients
13378 can no longer negatively impact the system or other users by
13379 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
13380 RuntimeDirectorySize= has been introduced that allows
13381 controlling the default size limit for all users. It
13382 defaults to 10% of the available physical memory. This is no
13383 replacement for quotas on tmpfs though (which the kernel
13384 still does not support), as /dev/shm and /tmp are still
13385 shared resources used by both the system and unprivileged
13388 * logind will now automatically turn off automatic suspending
13389 on laptop lid close when more than one display is
13390 connected. This was previously expected to be implemented
13391 individually in desktop environments (such as GNOME),
13392 however has been added to logind now, in order to fix a
13393 boot-time race where a desktop environment might not have
13394 been started yet and thus not been able to take an inhibitor
13395 lock at the time where logind already suspends the system
13396 due to a closed lid.
13398 * logind will now wait at least 30s after each system
13399 suspend/resume cycle, and 3min after system boot before
13400 suspending the system due to a closed laptop lid. This
13401 should give USB docking stations and similar enough time to
13402 be probed and configured after system resume and boot in
13403 order to then act as suspend blocker.
13405 * systemd-run gained a new --property= setting which allows
13406 initialization of resource control properties (and others)
13407 for the created scope or service unit. Example: "systemd-run
13408 --property=BlockIOWeight=10 updatedb" may be used to run
13409 updatedb at a low block IO scheduling weight.
13411 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
13412 now also work in --scope mode.
13414 * When systemd is compiled with kdbus support, basic support
13415 for enforced policies is now in place. (Note that enabling
13416 kdbus still voids your warranty and no API compatibility
13417 promises are made.)
13419 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
13420 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
13421 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
13422 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
13423 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
13424 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
13425 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
13426 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
13427 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
13428 Zbigniew Jędrzejewski-Szmek
13430 — Berlin, 2014-03-12
13434 * systemd will now relabel /dev after loading the SMACK policy
13435 according to SMACK rules.
13437 * A new unit file option AppArmorProfile= has been added to
13438 set the AppArmor profile for the processes of a unit.
13440 * A new condition check ConditionArchitecture= has been added
13441 to conditionalize units based on the system architecture, as
13442 reported by uname()'s "machine" field.
13444 * systemd-networkd now supports matching on the system
13445 virtualization, architecture, kernel command line, hostname
13448 * logind is now a lot more aggressive when suspending the
13449 machine due to a closed laptop lid. Instead of acting only
13450 on the lid close action, it will continuously watch the lid
13451 status and act on it. This is useful for laptops where the
13452 power button is on the outside of the chassis so that it can
13453 be reached without opening the lid (such as the Lenovo
13454 Yoga). On those machines, logind will now immediately
13455 re-suspend the machine if the power button has been
13456 accidentally pressed while the laptop was suspended and in a
13457 backpack or similar.
13459 * logind will now watch SW_DOCK switches and inhibit reaction
13460 to the lid switch if it is pressed. This means that logind
13461 will not suspend the machine anymore if the lid is closed
13462 and the system is docked, if the laptop supports SW_DOCK
13463 notifications via the input layer. Note that ACPI docking
13464 stations do not generate this currently. Also note that this
13465 logic is usually not fully sufficient and Desktop
13466 Environments should take a lid switch inhibitor lock when an
13467 external display is connected, as systemd will not watch
13470 * nspawn will now make use of the devices cgroup controller by
13471 default, and only permit creation of and access to the usual
13472 API device nodes like /dev/null or /dev/random, as well as
13473 access to (but not creation of) the pty devices.
13475 * We will now ship a default .network file for
13476 systemd-networkd that automatically configures DHCP for
13477 network interfaces created by nspawn's --network-veth or
13478 --network-bridge= switches.
13480 * systemd will now understand the usual M, K, G, T suffixes
13481 according to SI conventions (i.e. to the base 1000) when
13482 referring to throughput and hardware metrics. It will stay
13483 with IEC conventions (i.e. to the base 1024) for software
13484 metrics, according to what is customary according to
13485 Wikipedia. We explicitly document which base applies for
13486 each configuration option.
13488 * The DeviceAllow= setting in unit files now supports a syntax to
13489 allow-list an entire group of devices node majors at once, based on
13490 the /proc/devices listing. For example, with the string "char-pts",
13491 it is now possible to allow-list all current and future pseudo-TTYs
13494 * sd-event learned a new "post" event source. Event sources of
13495 this type are triggered by the dispatching of any event
13496 source of a type that is not "post". This is useful for
13497 implementing clean-up and check event sources that are
13498 triggered by other work being done in the program.
13500 * systemd-networkd is no longer statically enabled, but uses
13501 the usual [Install] sections so that it can be
13502 enabled/disabled using systemctl. It still is enabled by
13505 * When creating a veth interface pair with systemd-nspawn, the
13506 host side will now be prefixed with "vb-" if
13507 --network-bridge= is used, and with "ve-" if --network-veth
13508 is used. This way, it is easy to distinguish these cases on
13509 the host, for example to apply different configuration to
13510 them with systemd-networkd.
13512 * The compatibility libraries for libsystemd-journal.so,
13513 libsystem-id128.so, libsystemd-login.so and
13514 libsystemd-daemon.so do not make use of IFUNC
13515 anymore. Instead, we now build libsystemd.so multiple times
13516 under these alternative names. This means that the footprint
13517 is drastically increased, but given that these are
13518 transitional compatibility libraries, this should not matter
13519 much. This change has been made necessary to support the ARM
13520 platform for these compatibility libraries, as the ARM
13521 toolchain is not really at the same level as the toolchain
13522 for other architectures like x86 and does not support
13523 IFUNC. Please make sure to use --enable-compat-libs only
13524 during a transitional period!
13526 * The .include syntax has been deprecated and is not documented
13527 anymore. Drop-in files in .d directories should be used instead.
13529 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
13530 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
13531 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
13532 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
13533 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
13534 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
13535 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
13536 Zbigniew Jędrzejewski-Szmek
13538 — Berlin, 2014-02-24
13542 * A new component "systemd-networkd" has been added that can
13543 be used to configure local network interfaces statically or
13544 via DHCP. It is capable of bringing up bridges, VLANs, and
13545 bonding. Currently, no hook-ups for interactive network
13546 configuration are provided. Use this for your initrd,
13547 container, embedded, or server setup if you need a simple,
13548 yet powerful, network configuration solution. This
13549 configuration subsystem is quite nifty, as it allows wildcard
13550 hotplug matching in interfaces. For example, with a single
13551 configuration snippet, you can configure that all Ethernet
13552 interfaces showing up are automatically added to a bridge,
13553 or similar. It supports link-sensing and more.
13555 * A new tool "systemd-socket-proxyd" has been added which can
13556 act as a bidirectional proxy for TCP sockets. This is
13557 useful for adding socket activation support to services that
13558 do not actually support socket activation, including virtual
13559 machines and the like.
13561 * Add a new tool to save/restore rfkill state on
13564 * Save/restore state of keyboard backlights in addition to
13565 display backlights on shutdown/boot.
13567 * udev learned a new SECLABEL{} construct to label device
13568 nodes with a specific security label when they appear. For
13569 now, only SECLABEL{selinux} is supported, but the syntax is
13570 prepared for additional security frameworks.
13572 * udev gained a new scheme to configure link-level attributes
13573 from files in /etc/systemd/network/*.link. These files can
13574 match against MAC address, device path, driver name and type,
13575 and will apply attributes like the naming policy, link speed,
13576 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
13577 address assignment policy (randomized, …).
13579 * The configuration of network interface naming rules for
13580 "permanent interface names" has changed: a new NamePolicy=
13581 setting in the [Link] section of .link files determines the
13582 priority of possible naming schemes (onboard, slot, MAC,
13583 path). The default value of this setting is determined by
13584 /usr/lib/net/links/99-default.link. Old
13585 80-net-name-slot.rules udev configuration file has been
13586 removed, so local configuration overriding this file should
13587 be adapted to override 99-default.link instead.
13589 * When the User= switch is used in a unit file, also
13590 initialize $SHELL= based on the user database entry.
13592 * systemd no longer depends on libdbus. All communication is
13593 now done with sd-bus, systemd's low-level bus library
13596 * kdbus support has been added to PID 1 itself. When kdbus is
13597 enabled, this causes PID 1 to set up the system bus and
13598 enable support for a new ".busname" unit type that
13599 encapsulates bus name activation on kdbus. It works a little
13600 bit like ".socket" units, except for bus names. A new
13601 generator has been added that converts classic dbus1 service
13602 activation files automatically into native systemd .busname
13603 and .service units.
13605 * sd-bus: add a light-weight vtable implementation that allows
13606 defining objects on the bus with a simple static const
13607 vtable array of its methods, signals and properties.
13609 * systemd will not generate or install static dbus
13610 introspection data anymore to /usr/share/dbus-1/interfaces,
13611 as the precise format of these files is unclear, and
13612 nothing makes use of it.
13614 * A proxy daemon is now provided to proxy clients connecting
13615 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
13616 compatibility with classic D-Bus.
13618 * A bus driver implementation has been added that supports the
13619 classic D-Bus bus driver calls on kdbus, also for
13620 compatibility purposes.
13622 * A new API "sd-event.h" has been added that implements a
13623 minimal event loop API built around epoll. It provides a
13624 couple of features that direct epoll usage is lacking:
13625 prioritization of events, scales to large numbers of timer
13626 events, per-event timer slack (accuracy), system-wide
13627 coalescing of timer events, exit handlers, watchdog
13628 supervision support using systemd's sd_notify() API, child
13631 * A new API "sd-rntl.h" has been added that provides an API
13632 around the route netlink interface of the kernel, similar in
13633 style to "sd-bus.h".
13635 * A new API "sd-dhcp-client.h" has been added that provides a
13636 small DHCPv4 client-side implementation. This is used by
13637 "systemd-networkd".
13639 * There is a new kernel command line option
13640 "systemd.restore_state=0|1". When set to "0", none of the
13641 systemd tools will restore saved runtime state to hardware
13642 devices. More specifically, the rfkill and backlight states
13645 * The FsckPassNo= compatibility option in mount/service units
13646 has been removed. The fstab generator will now add the
13647 necessary dependencies automatically, and does not require
13648 PID1's support for that anymore.
13650 * journalctl gained a new switch, --list-boots, that lists
13651 recent boots with their times and boot IDs.
13653 * The various tools like systemctl, loginctl, timedatectl,
13654 busctl, systemd-run, … have gained a new switch "-M" to
13655 connect to a specific, local OS container (as direct
13656 connection, without requiring SSH). This works on any
13657 container that is registered with machined, such as those
13658 created by libvirt-lxc or nspawn.
13660 * systemd-run and systemd-analyze also gained support for "-H"
13661 to connect to remote hosts via SSH. This is particularly
13662 useful for systemd-run because it enables queuing of jobs
13663 onto remote systems.
13665 * machinectl gained a new command "login" to open a getty
13666 login in any local container. This works with any container
13667 that is registered with machined (such as those created by
13668 libvirt-lxc or nspawn), and which runs systemd inside.
13670 * machinectl gained a new "reboot" command that may be used to
13671 trigger a reboot on a specific container that is registered
13672 with machined. This works on any container that runs an init
13673 system of some kind.
13675 * systemctl gained a new "list-timers" command to print a nice
13676 listing of installed timer units with the times they elapse
13679 * Alternative reboot() parameters may now be specified on the
13680 "systemctl reboot" command line and are passed to the
13681 reboot() system call.
13683 * systemctl gained a new --job-mode= switch to configure the
13684 mode to queue a job with. This is a more generic version of
13685 --fail, --irreversible, and --ignore-dependencies, which are
13686 still available but not advertised anymore.
13688 * /etc/systemd/system.conf gained new settings to configure
13689 various default timeouts of units, as well as the default
13690 start limit interval and burst. These may still be overridden
13693 * PID1 will now export on the bus profile data of the security
13694 policy upload process (such as the SELinux policy upload to
13697 * journald: when forwarding logs to the console, include
13698 timestamps (following the setting in
13699 /sys/module/printk/parameters/time).
13701 * OnCalendar= in timer units now understands the special
13702 strings "yearly" and "annually". (Both are equivalent)
13704 * The accuracy of timer units is now configurable with the new
13705 AccuracySec= setting. It defaults to 1min.
13707 * A new dependency type JoinsNamespaceOf= has been added that
13708 allows running two services within the same /tmp and network
13709 namespace, if PrivateNetwork= or PrivateTmp= are used.
13711 * A new command "cat" has been added to systemctl. It outputs
13712 the original unit file of a unit, and concatenates the
13713 contents of additional "drop-in" unit file snippets, so that
13714 the full configuration is shown.
13716 * systemctl now supports globbing on the various "list-xyz"
13717 commands, like "list-units" or "list-sockets", as well as on
13718 those commands which take multiple unit names.
13720 * journalctl's --unit= switch gained support for globbing.
13722 * All systemd daemons now make use of the watchdog logic so
13723 that systemd automatically notices when they hang.
13725 * If the $container_ttys environment variable is set,
13726 getty-generator will automatically spawn a getty for each
13727 listed tty. This is useful for container managers to request
13728 login gettys to be spawned on as many ttys as needed.
13730 * %h, %s, %U specifier support is not available anymore when
13731 used in unit files for PID 1. This is because NSS calls are
13732 not safe from PID 1. They stay available for --user
13733 instances of systemd, and as special case for the root user.
13735 * loginctl gained a new "--no-legend" switch to turn off output
13736 of the legend text.
13738 * The "sd-login.h" API gained three new calls:
13739 sd_session_is_remote(), sd_session_get_remote_user(),
13740 sd_session_get_remote_host() to query information about
13743 * The udev hardware database now also carries vendor/product
13744 information of SDIO devices.
13746 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
13747 determine whether watchdog notifications are requested by
13748 the system manager.
13750 * Socket-activated per-connection services now include a
13751 short description of the connection parameters in the
13754 * tmpfiles gained a new "--boot" option. When this is not used,
13755 only lines where the command character is not suffixed with
13756 "!" are executed. When this option is specified, those
13757 options are executed too. This partitions tmpfiles
13758 directives into those that can be safely executed at any
13759 time, and those which should be run only at boot (for
13760 example, a line that creates /run/nologin).
13762 * A new API "sd-resolve.h" has been added which provides a simple
13763 asynchronous wrapper around glibc NSS hostname resolution
13764 calls, such as getaddrinfo(). In contrast to glibc's
13765 getaddrinfo_a(), it does not use signals. In contrast to most
13766 other asynchronous name resolution libraries, this one does
13767 not reimplement DNS, but reuses NSS, so that alternate
13768 hostname resolution systems continue to work, such as mDNS,
13769 LDAP, etc. This API is based on libasyncns, but it has been
13770 cleaned up for inclusion in systemd.
13772 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
13773 "sd-daemon.h" are no longer found in individual libraries
13774 libsystemd-journal.so, libsystemd-login.so,
13775 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
13776 merged them into a single library, libsystemd.so, which
13777 provides all symbols. The reason for this is cyclic
13778 dependencies, as these libraries tend to use each other's
13779 symbols. So far, we have managed to workaround that by linking
13780 a copy of a good part of our code into each of these
13781 libraries again and again, which, however, makes certain
13782 things hard to do, like sharing static variables. Also, it
13783 substantially increases footprint. With this change, there
13784 is only one library for the basic APIs systemd
13785 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
13786 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
13787 library as well, however are subject to the --enable-kdbus
13788 switch (see below). Note that "sd-dhcp-client.h" is not part
13789 of this library (this is because it only consumes, never
13790 provides, services of/to other APIs). To make the transition
13791 easy from the separate libraries to the unified one, we
13792 provide the --enable-compat-libs compile-time switch which
13793 will generate stub libraries that are compatible with the
13794 old ones but redirect all calls to the new one.
13796 * All of the kdbus logic and the new APIs "sd-bus.h",
13797 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
13798 and "sd-utf8.h" are compile-time optional via the
13799 "--enable-kdbus" switch, and they are not compiled in by
13800 default. To make use of kdbus, you have to explicitly enable
13801 the switch. Note however, that neither the kernel nor the
13802 userspace API for all of this is considered stable yet. We
13803 want to maintain the freedom to still change the APIs for
13804 now. By specifying this build-time switch, you acknowledge
13805 that you are aware of the instability of the current
13808 * Also, note that while kdbus is pretty much complete,
13809 it lacks one thing: proper policy support. This means you
13810 can build a fully working system with all features; however,
13811 it will be highly insecure. Policy support will be added in
13812 one of the next releases, at the same time that we will
13813 declare the APIs stable.
13815 * When the kernel command line argument "kdbus" is specified,
13816 systemd will automatically load the kdbus.ko kernel module. At
13817 this stage of development, it is only useful for testing kdbus
13818 and should not be used in production. Note: if "--enable-kdbus"
13819 is specified, and the kdbus.ko kernel module is available, and
13820 "kdbus" is added to the kernel command line, the entire system
13821 runs with kdbus instead of dbus-daemon, with the above mentioned
13822 problem of missing the system policy enforcement. Also a future
13823 version of kdbus.ko or a newer systemd will not be compatible with
13824 each other, and will unlikely be able to boot the machine if only
13825 one of them is updated.
13827 * systemctl gained a new "import-environment" command which
13828 uploads the caller's environment (or parts thereof) into the
13829 service manager so that it is inherited by services started
13830 by the manager. This is useful to upload variables like
13831 $DISPLAY into the user service manager.
13833 * A new PrivateDevices= switch has been added to service units
13834 which allows running a service with a namespaced /dev
13835 directory that does not contain any device nodes for
13836 physical devices. More specifically, it only includes devices
13837 such as /dev/null, /dev/urandom, and /dev/zero which are API
13840 * logind has been extended to support behaviour like VT
13841 switching on seats that do not support a VT. This makes
13842 multi-session available on seats that are not the first seat
13843 (seat0), and on systems where kernel support for VTs has
13844 been disabled at compile-time.
13846 * If a process holds a delay lock for system sleep or shutdown
13847 and fails to release it in time, we will now log its
13848 identity. This makes it easier to identify processes that
13849 cause slow suspends or power-offs.
13851 * When parsing /etc/crypttab, support for a new key-slot=
13852 option as supported by Debian is added. It allows indicating
13853 which LUKS slot to use on disk, speeding up key loading.
13855 * The sd_journal_sendv() API call has been checked and
13856 officially declared to be async-signal-safe so that it may
13857 be invoked from signal handlers for logging purposes.
13859 * Boot-time status output is now enabled automatically after a
13860 short timeout if boot does not progress, in order to give
13861 the user an indication what she or he is waiting for.
13863 * The boot-time output has been improved to show how much time
13864 remains until jobs expire.
13866 * The KillMode= switch in service units gained a new possible
13867 value "mixed". If set, and the unit is shut down, then the
13868 initial SIGTERM signal is sent only to the main daemon
13869 process, while the following SIGKILL signal is sent to
13870 all remaining processes of the service.
13872 * When a scope unit is registered, a new property "Controller"
13873 may be set. If set to a valid bus name, systemd will send a
13874 RequestStop() signal to this name when it would like to shut
13875 down the scope. This may be used to hook manager logic into
13876 the shutdown logic of scope units. Also, scope units may now
13877 be put in a special "abandoned" state, in which case the
13878 manager process which created them takes no further
13879 responsibilities for it.
13881 * When reading unit files, systemd will now verify
13882 the access mode of these files, and warn about certain
13883 suspicious combinations. This has been added to make it
13884 easier to track down packaging bugs where unit files are
13885 marked executable or world-writable.
13887 * systemd-nspawn gained a new "--setenv=" switch to set
13888 container-wide environment variables. The similar option in
13889 systemd-activate was renamed from "--environment=" to
13890 "--setenv=" for consistency.
13892 * systemd-nspawn has been updated to create a new kdbus domain
13893 for each container that is invoked, thus allowing each
13894 container to have its own set of system and user buses,
13895 independent of the host.
13897 * systemd-nspawn gained a new --drop-capability= switch to run
13898 the container with less capabilities than the default. Both
13899 --drop-capability= and --capability= now take the special
13900 string "all" for dropping or keeping all capabilities.
13902 * systemd-nspawn gained new switches for executing containers
13903 with specific SELinux labels set.
13905 * systemd-nspawn gained a new --quiet switch to not generate
13906 any additional output but the container's own console
13909 * systemd-nspawn gained a new --share-system switch to run a
13910 container without PID namespacing enabled.
13912 * systemd-nspawn gained a new --register= switch to control
13913 whether the container is registered with systemd-machined or
13914 not. This is useful for containers that do not run full
13915 OS images, but only specific apps.
13917 * systemd-nspawn gained a new --keep-unit which may be used
13918 when invoked as the only program from a service unit, and
13919 results in registration of the unit service itself in
13920 systemd-machined, instead of a newly opened scope unit.
13922 * systemd-nspawn gained a new --network-interface= switch for
13923 moving arbitrary interfaces to the container. The new
13924 --network-veth switch creates a virtual Ethernet connection
13925 between host and container. The new --network-bridge=
13926 switch then allows assigning the host side of this virtual
13927 Ethernet connection to a bridge device.
13929 * systemd-nspawn gained a new --personality= switch for
13930 setting the kernel personality for the container. This is
13931 useful when running a 32-bit container on a 64-bit host. A
13932 similar option Personality= is now also available for service
13935 * logind will now also track a "Desktop" identifier for each
13936 session which encodes the desktop environment of it. This is
13937 useful for desktop environments that want to identify
13938 multiple running sessions of itself easily.
13940 * A new SELinuxContext= setting for service units has been
13941 added that allows setting a specific SELinux execution
13942 context for a service.
13944 * Most systemd client tools will now honour $SYSTEMD_LESS for
13945 settings of the "less" pager. By default, these tools will
13946 override $LESS to allow certain operations to work, such as
13947 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
13948 influence this logic.
13950 * systemd's "seccomp" hook-up has been changed to make use of
13951 the libseccomp library instead of using its own
13952 implementation. This has benefits for portability among
13955 * For usage together with SystemCallFilter=, a new
13956 SystemCallErrorNumber= setting has been introduced that
13957 allows configuration of a system error number to be returned
13958 on filtered system calls, instead of immediately killing the
13959 process. Also, SystemCallArchitectures= has been added to
13960 limit access to system calls of a particular architecture
13961 (in order to turn off support for unused secondary
13962 architectures). There is also a global
13963 SystemCallArchitectures= setting in system.conf now to turn
13964 off support for non-native system calls system-wide.
13966 * systemd requires a kernel with a working name_to_handle_at(),
13967 please see the kernel config requirements in the README file.
13969 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
13970 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
13971 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
13972 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
13973 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
13974 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
13975 Elia Pinto, Florian Weimer, George McCollister, Goffredo
13976 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
13977 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
13978 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
13979 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
13980 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
13981 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
13982 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
13983 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
13984 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
13985 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
13986 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
13987 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
13988 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
13989 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
13990 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
13991 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
13992 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
13994 — Berlin, 2014-02-20
13998 * logind has gained support for facilitating privileged input
13999 and drm device access for unprivileged clients. This work is
14000 useful to allow Wayland display servers (and similar
14001 programs, such as kmscon) to run under the user's ID and
14002 access input and drm devices which are normally
14003 protected. When this is used (and the kernel is new enough)
14004 logind will "mute" IO on the file descriptors passed to
14005 Wayland as long as it is in the background and "unmute" it
14006 if it returns into the foreground. This allows secure
14007 session switching without allowing background sessions to
14008 eavesdrop on input and display data. This also introduces
14009 session switching support if VT support is turned off in the
14010 kernel, and on seats that are not seat0.
14012 * A new kernel command line option luks.options= is understood
14013 now which allows specifying LUKS options for usage for LUKS
14014 encrypted partitions specified with luks.uuid=.
14016 * tmpfiles.d(5) snippets may now use specifier expansion in
14017 path names. More specifically %m, %b, %H, %v, are now
14018 replaced by the local machine id, boot id, hostname, and
14019 kernel version number.
14021 * A new tmpfiles.d(5) command "m" has been introduced which
14022 may be used to change the owner/group/access mode of a file
14023 or directory if it exists, but do nothing if it does not.
14025 * This release removes high-level support for the
14026 MemorySoftLimit= cgroup setting. The underlying kernel
14027 cgroup attribute memory.soft_limit= is currently badly
14028 designed and likely to be removed from the kernel API in its
14029 current form, hence we should not expose it for now.
14031 * The memory.use_hierarchy cgroup attribute is now enabled for
14032 all cgroups systemd creates in the memory cgroup
14033 hierarchy. This option is likely to be come the built-in
14034 default in the kernel anyway, and the non-hierarchical mode
14035 never made much sense in the intrinsically hierarchical
14038 * A new field _SYSTEMD_SLICE= is logged along with all journal
14039 messages containing the slice a message was generated
14040 from. This is useful to allow easy per-customer filtering of
14041 logs among other things.
14043 * systemd-journald will no longer adjust the group of journal
14044 files it creates to the "systemd-journal" group. Instead we
14045 rely on the journal directory to be owned by the
14046 "systemd-journal" group, and its setgid bit set, so that the
14047 kernel file system layer will automatically enforce that
14048 journal files inherit this group assignment. The reason for
14049 this change is that we cannot allow NSS look-ups from
14050 journald which would be necessary to resolve
14051 "systemd-journal" to a numeric GID, because this might
14052 create deadlocks if NSS involves synchronous queries to
14053 other daemons (such as nscd, or sssd) which in turn are
14054 logging clients of journald and might block on it, which
14055 would then dead lock. A tmpfiles.d(5) snippet included in
14056 systemd will make sure the setgid bit and group are
14057 properly set on the journal directory if it exists on every
14058 boot. However, we recommend adjusting it manually after
14059 upgrades too (or from RPM scriptlets), so that the change is
14060 not delayed until next reboot.
14062 * Backlight and random seed files in /var/lib/ have moved into
14063 the /var/lib/systemd/ directory, in order to centralize all
14064 systemd generated files in one directory.
14066 * Boot time performance measurements (as displayed by
14067 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
14068 performance information if that's available to determine how
14069 much time BIOS and boot loader initialization required. With
14070 a sufficiently new BIOS you hence no longer need to boot
14071 with Gummiboot to get access to such information.
14073 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
14074 Cristian Rodríguez, Dave Reisner, David Herrmann, David
14075 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
14076 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
14077 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
14078 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
14079 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
14081 — Berlin, 2013-10-02
14085 * The Restart= option for services now understands a new
14086 on-watchdog setting, which will restart the service
14087 automatically if the service stops sending out watchdog keep
14088 alive messages (as configured with WatchdogSec=).
14090 * The getty generator (which is responsible for bringing up a
14091 getty on configured serial consoles) will no longer only
14092 start a getty on the primary kernel console but on all
14093 others, too. This makes the order in which console= is
14094 specified on the kernel command line less important.
14096 * libsystemd-logind gained a new sd_session_get_vt() call to
14097 retrieve the VT number of a session.
14099 * If the option "tries=0" is set for an entry of /etc/crypttab
14100 its passphrase is queried indefinitely instead of any
14101 maximum number of tries.
14103 * If a service with a configure PID file terminates its PID
14104 file will now be removed automatically if it still exists
14105 afterwards. This should put an end to stale PID files.
14107 * systemd-run will now also take relative binary path names
14108 for execution and no longer insists on absolute paths.
14110 * InaccessibleDirectories= and ReadOnlyDirectories= now take
14111 paths that are optionally prefixed with "-" to indicate that
14112 it should not be considered a failure if they do not exist.
14114 * journalctl -o (and similar commands) now understands a new
14115 output mode "short-precise", it is similar to "short" but
14116 shows timestamps with usec accuracy.
14118 * The option "discard" (as known from Debian) is now
14119 synonymous to "allow-discards" in /etc/crypttab. In fact,
14120 "discard" is preferred now (since it is easier to remember
14123 * Some licensing clean-ups were made, so that more code is now
14124 LGPL-2.1 licensed than before.
14126 * A minimal tool to save/restore the display backlight
14127 brightness across reboots has been added. It will store the
14128 backlight setting as late as possible at shutdown, and
14129 restore it as early as possible during reboot.
14131 * A logic to automatically discover and enable home and swap
14132 partitions on GPT disks has been added. With this in place
14133 /etc/fstab becomes optional for many setups as systemd can
14134 discover certain partitions located on the root disk
14135 automatically. Home partitions are recognized under their
14136 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
14137 partitions are recognized under their GPT type ID
14138 0657fd6da4ab43c484e50933c84b4f4f.
14140 * systemd will no longer pass any environment from the kernel
14141 or initrd to system services. If you want to set an
14142 environment for all services, do so via the kernel command
14143 line systemd.setenv= assignment.
14145 * The systemd-sysctl tool no longer natively reads the file
14146 /etc/sysctl.conf. If desired, the file should be symlinked
14147 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
14148 legacy support by a symlink rather than built-in code, it
14149 also makes the otherwise hidden order of application of the
14150 different files visible. (Note that this partly reverts to a
14151 pre-198 application order of sysctl knobs!)
14153 * The "systemctl set-log-level" and "systemctl dump" commands
14154 have been moved to systemd-analyze.
14156 * systemd-run learned the new --remain-after-exit switch,
14157 which causes the scope unit not to be cleaned up
14158 automatically after the process terminated.
14160 * tmpfiles learned a new --exclude-prefix= switch to exclude
14161 certain paths from operation.
14163 * journald will now automatically flush all messages to disk
14164 as soon as a message at the log level CRIT, ALERT or EMERG
14167 Contributions from: Andrew Cook, Brandon Philips, Christian
14168 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
14169 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
14170 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
14171 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
14172 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
14173 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
14174 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
14175 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
14176 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
14177 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
14178 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
14179 William Giokas, Zbigniew Jędrzejewski-Szmek
14181 — Berlin, 2013-09-13
14185 * The documentation has been updated to cover the various new
14186 concepts introduced with 205.
14188 * Unit files now understand the new %v specifier which
14189 resolves to the kernel version string as returned by "uname
14192 * systemctl now supports filtering the unit list output by
14193 load state, active state and sub state, using the new
14194 --state= parameter.
14196 * "systemctl status" will now show the results of the
14197 condition checks (like ConditionPathExists= and similar) of
14198 the last start attempts of the unit. They are also logged to
14201 * "journalctl -b" may now be used to look for boot output of a
14202 specific boot. Try "journalctl -b -1" for the previous boot,
14203 but the syntax is substantially more powerful.
14205 * "journalctl --show-cursor" has been added which prints the
14206 cursor string the last shown log line. This may then be used
14207 with the new "journalctl --after-cursor=" switch to continue
14208 browsing logs from that point on.
14210 * "journalctl --force" may now be used to force regeneration
14213 * Creation of "dead" device nodes has been moved from udev
14214 into kmod and tmpfiles. Previously, udev would read the kmod
14215 databases to pre-generate dead device nodes based on meta
14216 information contained in kernel modules, so that these would
14217 be auto-loaded on access rather then at boot. As this
14218 does not really have much to do with the exposing actual
14219 kernel devices to userspace this has always been slightly
14220 alien in the udev codebase. Following the new scheme kmod
14221 will now generate a runtime snippet for tmpfiles from the
14222 module meta information and it now is tmpfiles' job to the
14223 create the nodes. This also allows overriding access and
14224 other parameters for the nodes using the usual tmpfiles
14225 facilities. As side effect this allows us to remove the
14226 CAP_SYS_MKNOD capability bit from udevd entirely.
14228 * logind's device ACLs may now be applied to these "dead"
14229 devices nodes too, thus finally allowing managed access to
14230 devices such as /dev/snd/sequencer without loading the
14231 backing module right-away.
14233 * A new RPM macro has been added that may be used to apply
14234 tmpfiles configuration during package installation.
14236 * systemd-detect-virt and ConditionVirtualization= now can
14237 detect User-Mode-Linux machines (UML).
14239 * journald will now implicitly log the effective capabilities
14240 set of processes in the message metadata.
14242 * systemd-cryptsetup has gained support for TrueCrypt volumes.
14244 * The initrd interface has been simplified (more specifically,
14245 support for passing performance data via environment
14246 variables and fsck results via files in /run has been
14247 removed). These features were non-essential, and are
14248 nowadays available in a much nicer way by having systemd in
14249 the initrd serialize its state and have the hosts systemd
14250 deserialize it again.
14252 * The udev "keymap" data files and tools to apply keyboard
14253 specific mappings of scan to key codes, and force-release
14254 scan code lists have been entirely replaced by a udev
14255 "keyboard" builtin and a hwdb data file.
14257 * systemd will now honour the kernel's "quiet" command line
14258 argument also during late shutdown, resulting in a
14259 completely silent shutdown when used.
14261 * There's now an option to control the SO_REUSEPORT socket
14262 option in .socket units.
14264 * Instance units will now automatically get a per-template
14265 subslice of system.slice unless something else is explicitly
14266 configured. For example, instances of sshd@.service will now
14267 implicitly be placed in system-sshd.slice rather than
14268 system.slice as before.
14270 * Test coverage support may now be enabled at build time.
14272 Contributions from: Dave Reisner, Frederic Crozat, Harald
14273 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
14274 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
14275 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
14276 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
14277 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
14278 Giokas, Zbigniew Jędrzejewski-Szmek
14280 — Berlin, 2013-07-23
14284 * Two new unit types have been introduced:
14286 Scope units are very similar to service units, however, are
14287 created out of pre-existing processes — instead of PID 1
14288 forking off the processes. By using scope units it is
14289 possible for system services and applications to group their
14290 own child processes (worker processes) in a powerful way
14291 which then maybe used to organize them, or kill them
14292 together, or apply resource limits on them.
14294 Slice units may be used to partition system resources in an
14295 hierarchical fashion and then assign other units to them. By
14296 default there are now three slices: system.slice (for all
14297 system services), user.slice (for all user sessions),
14298 machine.slice (for VMs and containers).
14300 Slices and scopes have been introduced primarily in
14301 context of the work to move cgroup handling to a
14302 single-writer scheme, where only PID 1
14303 creates/removes/manages cgroups.
14305 * There's a new concept of "transient" units. In contrast to
14306 normal units these units are created via an API at runtime,
14307 not from configuration from disk. More specifically this
14308 means it is now possible to run arbitrary programs as
14309 independent services, with all execution parameters passed
14310 in via bus APIs rather than read from disk. Transient units
14311 make systemd substantially more dynamic then it ever was,
14312 and useful as a general batch manager.
14314 * logind has been updated to make use of scope and slice units
14315 for managing user sessions. As a user logs in he will get
14316 his own private slice unit, to which all sessions are added
14317 as scope units. We also added support for automatically
14318 adding an instance of user@.service for the user into the
14319 slice. Effectively logind will no longer create cgroup
14320 hierarchies on its own now, it will defer entirely to PID 1
14321 for this by means of scope, service and slice units. Since
14322 user sessions this way become entities managed by PID 1
14323 the output of "systemctl" is now a lot more comprehensive.
14325 * A new mini-daemon "systemd-machined" has been added which
14326 may be used by virtualization managers to register local
14327 VMs/containers. nspawn has been updated accordingly, and
14328 libvirt will be updated shortly. machined will collect a bit
14329 of meta information about the VMs/containers, and assign
14330 them their own scope unit (see above). The collected
14331 meta-data is then made available via the "machinectl" tool,
14332 and exposed in "ps" and similar tools. machined/machinectl
14333 is compile-time optional.
14335 * As discussed earlier, the low-level cgroup configuration
14336 options ControlGroup=, ControlGroupModify=,
14337 ControlGroupPersistent=, ControlGroupAttribute= have been
14338 removed. Please use high-level attribute settings instead as
14339 well as slice units.
14341 * A new bus call SetUnitProperties() has been added to alter
14342 various runtime parameters of a unit. This is primarily
14343 useful to alter cgroup parameters dynamically in a nice way,
14344 but will be extended later on to make more properties
14345 modifiable at runtime. systemctl gained a new set-properties
14346 command that wraps this call.
14348 * A new tool "systemd-run" has been added which can be used to
14349 run arbitrary command lines as transient services or scopes,
14350 while configuring a number of settings via the command
14351 line. This tool is currently very basic, however already
14352 very useful. We plan to extend this tool to even allow
14353 queuing of execution jobs with time triggers from the
14354 command line, similar in fashion to "at".
14356 * nspawn will now inform the user explicitly that kernels with
14357 audit enabled break containers, and suggest the user to turn
14360 * Support for detecting the IMA and AppArmor security
14361 frameworks with ConditionSecurity= has been added.
14363 * journalctl gained a new "-k" switch for showing only kernel
14364 messages, mimicking dmesg output; in addition to "--user"
14365 and "--system" switches for showing only user's own logs
14368 * systemd-delta can now show information about drop-in
14369 snippets extending unit files.
14371 * libsystemd-bus has been substantially updated but is still
14372 not available as public API.
14374 * systemd will now look for the "debug" argument on the kernel
14375 command line and enable debug logging, similar to what
14376 "systemd.log_level=debug" already did before.
14378 * "systemctl set-default", "systemctl get-default" has been
14379 added to configure the default.target symlink, which
14380 controls what to boot into by default.
14382 * "systemctl set-log-level" has been added as a convenient
14383 way to raise and lower systemd logging threshold.
14385 * "systemd-analyze plot" will now show the time the various
14386 generators needed for execution, as well as information
14387 about the unit file loading.
14389 * libsystemd-journal gained a new sd_journal_open_files() call
14390 for opening specific journal files. journactl also gained a
14391 new switch to expose this new functionality. Previously we
14392 only supported opening all files from a directory, or all
14393 files from the system, as opening individual files only is
14394 racy due to journal file rotation.
14396 * systemd gained the new DefaultEnvironment= setting in
14397 /etc/systemd/system.conf to set environment variables for
14400 * If a privileged process logs a journal message with the
14401 OBJECT_PID= field set, then journald will automatically
14402 augment this with additional OBJECT_UID=, OBJECT_GID=,
14403 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
14404 system services want to log events about specific client
14405 processes. journactl/systemctl has been updated to make use
14406 of this information if all log messages regarding a specific
14409 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
14410 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
14411 Reisner, David Coppa, David King, David Strauss, Eelco
14412 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
14413 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
14414 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
14415 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
14416 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
14417 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
14418 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
14419 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
14420 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
14421 Łukasz Stelmach, 장동준
14425 * The Python bindings gained some minimal support for the APIs
14426 exposed by libsystemd-logind.
14428 * ConditionSecurity= gained support for detecting SMACK. Since
14429 this condition already supports SELinux and AppArmor we only
14430 miss IMA for this. Patches welcome!
14432 Contributions from: Karol Lewandowski, Lennart Poettering,
14433 Zbigniew Jędrzejewski-Szmek
14437 * systemd-nspawn will now create /etc/resolv.conf if
14438 necessary, before bind-mounting the host's file onto it.
14440 * systemd-nspawn will now store meta information about a
14441 container on the container's cgroup as extended attribute
14442 fields, including the root directory.
14444 * The cgroup hierarchy has been reworked in many ways. All
14445 objects any of the components systemd creates in the cgroup
14446 tree are now suffixed. More specifically, user sessions are
14447 now placed in cgroups suffixed with ".session", users in
14448 cgroups suffixed with ".user", and nspawn containers in
14449 cgroups suffixed with ".nspawn". Furthermore, all cgroup
14450 names are now escaped in a simple scheme to avoid collision
14451 of userspace object names with kernel filenames. This work
14452 is preparation for making these objects relocatable in the
14453 cgroup tree, in order to allow easy resource partitioning of
14454 these objects without causing naming conflicts.
14456 * systemctl list-dependencies gained the new switches
14457 --plain, --reverse, --after and --before.
14459 * systemd-inhibit now shows the process name of processes that
14460 have taken an inhibitor lock.
14462 * nss-myhostname will now also resolve "localhost"
14463 implicitly. This makes /etc/hosts an optional file and
14464 nicely handles that on IPv6 ::1 maps to both "localhost" and
14465 the local hostname.
14467 * libsystemd-logind.so gained a new call
14468 sd_get_machine_names() to enumerate running containers and
14469 VMs (currently only supported by very new libvirt and
14470 nspawn). sd_login_monitor can now be used to watch
14471 VMs/containers coming and going.
14473 * .include is not allowed recursively anymore, and only in
14474 unit files. Usually it is better to use drop-in snippets in
14475 .d/*.conf anyway, as introduced with systemd 198.
14477 * systemd-analyze gained a new "critical-chain" command that
14478 determines the slowest chain of units run during system
14479 boot-up. It is very useful for tracking down where
14480 optimizing boot time is the most beneficial.
14482 * systemd will no longer allow manipulating service paths in
14483 the name=systemd:/system cgroup tree using ControlGroup= in
14484 units. (But is still fine with it in all other dirs.)
14486 * There's a new systemd-nspawn@.service service file that may
14487 be used to easily run nspawn containers as system
14488 services. With the container's root directory in
14489 /var/lib/container/foobar it is now sufficient to run
14490 "systemctl start systemd-nspawn@foobar.service" to boot it.
14492 * systemd-cgls gained a new parameter "--machine" to list only
14493 the processes within a certain container.
14495 * ConditionSecurity= now can check for "apparmor". We still
14496 are lacking checks for SMACK and IMA for this condition
14497 check though. Patches welcome!
14499 * A new configuration file /etc/systemd/sleep.conf has been
14500 added that may be used to configure which kernel operation
14501 systemd is supposed to execute when "suspend", "hibernate"
14502 or "hybrid-sleep" is requested. This makes the new kernel
14503 "freeze" state accessible to the user.
14505 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
14506 the passed argument if applicable.
14508 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
14509 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
14510 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
14511 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
14512 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
14513 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
14514 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
14519 * The output of 'systemctl list-jobs' got some polishing. The
14520 '--type=' argument may now be passed more than once. A new
14521 command 'systemctl list-sockets' has been added which shows
14522 a list of kernel sockets systemd is listening on with the
14523 socket units they belong to, plus the units these socket
14526 * The experimental libsystemd-bus library got substantial
14527 updates to work in conjunction with the (also experimental)
14528 kdbus kernel project. It works well enough to exchange
14529 messages with some sophistication. Note that kdbus is not
14530 ready yet, and the library is mostly an elaborate test case
14531 for now, and not installable.
14533 * systemd gained a new unit 'systemd-static-nodes.service'
14534 that generates static device nodes earlier during boot, and
14535 can run in conjunction with udev.
14537 * libsystemd-login gained a new call sd_pid_get_user_unit()
14538 to retrieve the user systemd unit a process is running
14539 in. This is useful for systems where systemd is used as
14542 * systemd-nspawn now places all containers in the new /machine
14543 top-level cgroup directory in the name=systemd
14544 hierarchy. libvirt will soon do the same, so that we get a
14545 uniform separation of /system, /user and /machine for system
14546 services, user processes and containers/virtual
14547 machines. This new cgroup hierarchy is also useful to stick
14548 stable names to specific container instances, which can be
14549 recognized later this way (this name may be controlled
14550 via systemd-nspawn's new -M switch). libsystemd-login also
14551 gained a new call sd_pid_get_machine_name() to retrieve the
14552 name of the container/VM a specific process belongs to.
14554 * bootchart can now store its data in the journal.
14556 * libsystemd-journal gained a new call
14557 sd_journal_add_conjunction() for AND expressions to the
14558 matching logic. This can be used to express more complex
14559 logical expressions.
14561 * journactl can now take multiple --unit= and --user-unit=
14564 * The cryptsetup logic now understands the "luks.key=" kernel
14565 command line switch for specifying a file to read the
14566 decryption key from. Also, if a configured key file is not
14567 found the tool will now automatically fall back to prompting
14570 * Python systemd.journal module was updated to wrap recently
14571 added functions from libsystemd-journal. The interface was
14572 changed to bring the low level interface in s.j._Reader
14573 closer to the C API, and the high level interface in
14574 s.j.Reader was updated to wrap and convert all data about
14577 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
14578 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
14579 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
14580 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
14581 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
14582 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
14586 * journalctl --update-catalog now understands a new --root=
14587 option to operate on catalogs found in a different root
14590 * During shutdown after systemd has terminated all running
14591 services a final killing loop kills all remaining left-over
14592 processes. We will now print the name of these processes
14593 when we send SIGKILL to them, since this usually indicates a
14596 * If /etc/crypttab refers to password files stored on
14597 configured mount points automatic dependencies will now be
14598 generated to ensure the specific mount is established first
14599 before the key file is attempted to be read.
14601 * 'systemctl status' will now show information about the
14602 network sockets a socket unit is listening on.
14604 * 'systemctl status' will also shown information about any
14605 drop-in configuration file for units. (Drop-In configuration
14606 files in this context are files such as
14607 /etc/systemd/system/foobar.service.d/*.conf)
14609 * systemd-cgtop now optionally shows summed up CPU times of
14610 cgroups. Press '%' while running cgtop to switch between
14611 percentage and absolute mode. This is useful to determine
14612 which cgroups use up the most CPU time over the entire
14613 runtime of the system. systemd-cgtop has also been updated
14614 to be 'pipeable' for processing with further shell tools.
14616 * 'hostnamectl set-hostname' will now allow setting of FQDN
14619 * The formatting and parsing of time span values has been
14620 changed. The parser now understands fractional expressions
14621 such as "5.5h". The formatter will now output fractional
14622 expressions for all time spans under 1min, i.e. "5.123456s"
14623 rather than "5s 123ms 456us". For time spans under 1s
14624 millisecond values are shown, for those under 1ms
14625 microsecond values are shown. This should greatly improve
14626 all time-related output of systemd.
14628 * libsystemd-login and libsystemd-journal gained new
14629 functions for querying the poll() events mask and poll()
14630 timeout value for integration into arbitrary event
14633 * localectl gained the ability to list available X11 keymaps
14634 (models, layouts, variants, options).
14636 * 'systemd-analyze dot' gained the ability to filter for
14637 specific units via shell-style globs, to create smaller,
14638 more useful graphs. I.e. it is now possible to create simple
14639 graphs of all the dependencies between only target units, or
14640 of all units that Avahi has dependencies with.
14642 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
14643 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
14644 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
14645 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
14646 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
14647 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
14648 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
14652 * The boot-time readahead implementation for rotating media
14653 will now read the read-ahead data in multiple passes which
14654 consist of all read requests made in equidistant time
14655 intervals. This means instead of strictly reading read-ahead
14656 data in its physical order on disk we now try to find a
14657 middle ground between physical and access time order.
14659 * /etc/os-release files gained a new BUILD_ID= field for usage
14660 on operating systems that provide continuous builds of OS
14663 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
14664 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
14665 William Douglas, Zbigniew Jędrzejewski-Szmek
14669 * systemd-python gained an API exposing libsystemd-daemon.
14671 * The SMACK setup logic gained support for uploading CIPSO
14674 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
14675 ReadOnlyDirectories= and InaccessibleDirectories= has
14676 changed. The private /tmp and /var/tmp directories are now
14677 shared by all processes of a service (which means
14678 ExecStartPre= may now leave data in /tmp that ExecStart= of
14679 the same service can still access). When a service is
14680 stopped its temporary directories are immediately deleted
14681 (normal clean-up with tmpfiles is still done in addition to
14684 * By default, systemd will now set a couple of sysctl
14685 variables in the kernel: the safe sysrq options are turned
14686 on, IP route verification is turned on, and source routing
14687 disabled. The recently added hardlink and softlink
14688 protection of the kernel is turned on. These settings should
14689 be reasonably safe, and good defaults for all new systems.
14691 * The predictable network naming logic may now be turned off
14692 with a new kernel command line switch: net.ifnames=0.
14694 * A new libsystemd-bus module has been added that implements a
14695 pretty complete D-Bus client library. For details see:
14697 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
14699 * journald will now explicitly flush the journal files to disk
14700 at the latest 5min after each write. The file will then also
14701 be marked offline until the next write. This should increase
14702 reliability in case of a crash. The synchronization delay
14703 can be configured via SyncIntervalSec= in journald.conf.
14705 * There's a new remote-fs-setup.target unit that can be used
14706 to pull in specific services when at least one remote file
14707 system is to be mounted.
14709 * There are new targets timers.target and paths.target as
14710 canonical targets to pull user timer and path units in
14711 from. This complements sockets.target with a similar
14712 purpose for socket units.
14714 * libudev gained a new call udev_device_set_attribute_value()
14715 to set sysfs attributes of a device.
14717 * The udev daemon now sets the default number of worker
14718 processes executed in parallel based on the number of available
14719 CPUs instead of the amount of available RAM. This is supposed
14720 to provide a more reliable default and limit a too aggressive
14721 parallelism for setups with 1000s of devices connected.
14723 Contributions from: Auke Kok, Colin Walters, Cristian
14724 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
14725 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
14726 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
14727 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
14728 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
14729 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
14730 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
14731 Zbigniew Jędrzejewski-Szmek
14735 * Configuration of unit files may now be extended via drop-in
14736 files without having to edit/override the unit files
14737 themselves. More specifically, if the administrator wants to
14738 change one value for a service file foobar.service he can
14739 now do so by dropping in a configuration snippet into
14740 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
14741 will load all these snippets and apply them on top of the
14742 main unit configuration file, possibly extending or
14743 overriding its settings. Using these drop-in snippets is
14744 generally nicer than the two earlier options for changing
14745 unit files locally: copying the files from
14746 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
14747 them there; or creating a new file in /etc/systemd/system/
14748 that incorporates the original one via ".include". Drop-in
14749 snippets into these .d/ directories can be placed in any
14750 directory systemd looks for units in, and the usual
14751 overriding semantics between /usr/lib, /etc and /run apply
14754 * Most unit file settings which take lists of items can now be
14755 reset by assigning the empty string to them. For example,
14756 normally, settings such as Environment=FOO=BAR append a new
14757 environment variable assignment to the environment block,
14758 each time they are used. By assigning Environment= the empty
14759 string the environment block can be reset to empty. This is
14760 particularly useful with the .d/*.conf drop-in snippets
14761 mentioned above, since this adds the ability to reset list
14762 settings from vendor unit files via these drop-ins.
14764 * systemctl gained a new "list-dependencies" command for
14765 listing the dependencies of a unit recursively.
14767 * Inhibitors are now honored and listed by "systemctl
14768 suspend", "systemctl poweroff" (and similar) too, not only
14769 GNOME. These commands will also list active sessions by
14772 * Resource limits (as exposed by the various control group
14773 controllers) can now be controlled dynamically at runtime
14774 for all units. More specifically, you can now use a command
14775 like "systemctl set-cgroup-attr foobar.service cpu.shares
14776 2000" to alter the CPU shares a specific service gets. These
14777 settings are stored persistently on disk, and thus allow the
14778 administrator to easily adjust the resource usage of
14779 services with a few simple commands. This dynamic resource
14780 management logic is also available to other programs via the
14781 bus. Almost any kernel cgroup attribute and controller is
14784 * systemd-vconsole-setup will now copy all font settings to
14785 all allocated VTs, where it previously applied them only to
14788 * libsystemd-login gained the new sd_session_get_tty() API
14791 * This release drops support for a few legacy or
14792 distribution-specific LSB facility names when parsing init
14793 scripts: $x-display-manager, $mail-transfer-agent,
14794 $mail-transport-agent, $mail-transfer-agent, $smtp,
14795 $null. Also, the mail-transfer-agent.target unit backing
14796 this has been removed. Distributions which want to retain
14797 compatibility with this should carry the burden for
14798 supporting this themselves and patch support for these back
14799 in, if they really need to. Also, the facilities $syslog and
14800 $local_fs are now ignored, since systemd does not support
14801 early-boot LSB init scripts anymore, and these facilities
14802 are implied anyway for normal services. syslog.target has
14805 * There are new bus calls on PID1's Manager object for
14806 cancelling jobs, and removing snapshot units. Previously,
14807 both calls were only available on the Job and Snapshot
14808 objects themselves.
14810 * systemd-journal-gatewayd gained SSL support.
14812 * The various "environment" files, such as /etc/locale.conf
14813 now support continuation lines with a backslash ("\") as
14814 last character in the line, similarly in style (but different)
14815 to how this is supported in shells.
14817 * For normal user processes the _SYSTEMD_USER_UNIT= field is
14818 now implicitly appended to every log entry logged. systemctl
14819 has been updated to filter by this field when operating on a
14820 user systemd instance.
14822 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
14823 CAP_AUDIT_CONTROL capabilities to the capabilities set for
14824 the container. This makes it easier to boot unmodified
14825 Fedora systems in a container, which however still requires
14826 audit=0 to be passed on the kernel command line. Auditing in
14827 kernel and userspace is unfortunately still too broken in
14828 context of containers, hence we recommend compiling it out
14829 of the kernel or using audit=0. Hopefully this will be fixed
14830 one day for good in the kernel.
14832 * nspawn gained the new --bind= and --bind-ro= parameters to
14833 bind mount specific directories from the host into the
14836 * nspawn will now mount its own devpts file system instance
14837 into the container, in order not to leak pty devices from
14838 the host into the container.
14840 * systemd will now read the firmware boot time performance
14841 information from the EFI variables, if the used boot loader
14842 supports this, and takes it into account for boot performance
14843 analysis via "systemd-analyze". This is currently supported
14844 only in conjunction with Gummiboot, but could be supported
14845 by other boot loaders too. For details see:
14847 https://systemd.io/BOOT_LOADER_INTERFACE
14849 * A new generator has been added that automatically mounts the
14850 EFI System Partition (ESP) to /boot, if that directory
14851 exists, is empty, and no other file system has been
14852 configured to be mounted there.
14854 * logind will now send out PrepareForSleep(false) out
14855 unconditionally, after coming back from suspend. This may be
14856 used by applications as asynchronous notification for
14857 system resume events.
14859 * "systemctl unlock-sessions" has been added, that allows
14860 unlocking the screens of all user sessions at once, similar
14861 to how "systemctl lock-sessions" already locked all users
14862 sessions. This is backed by a new D-Bus call UnlockSessions().
14864 * "loginctl seat-status" will now show the master device of a
14865 seat. (i.e. the device of a seat that needs to be around for
14866 the seat to be considered available, usually the graphics
14869 * tmpfiles gained a new "X" line type, that allows
14870 configuration of files and directories (with wildcards) that
14871 shall be excluded from automatic cleanup ("aging").
14873 * udev default rules set the device node permissions now only
14874 at "add" events, and do not change them any longer with a
14875 later "change" event.
14877 * The log messages for lid events and power/sleep keypresses
14878 now carry a message ID.
14880 * We now have a substantially larger unit test suite, but this
14881 continues to be work in progress.
14883 * udevadm hwdb gained a new --root= parameter to change the
14884 root directory to operate relative to.
14886 * logind will now issue a background sync() request to the kernel
14887 early at shutdown, so that dirty buffers are flushed to disk early
14888 instead of at the last moment, in order to optimize shutdown
14891 * A new bootctl tool has been added that is an interface for
14892 certain boot loader operations. This is currently a preview
14893 and is likely to be extended into a small mechanism daemon
14894 like timedated, localed, hostnamed, and can be used by
14895 graphical UIs to enumerate available boot options, and
14896 request boot into firmware operations.
14898 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
14899 the rest of the package. It also has been updated to work
14900 correctly in initrds.
14902 * polkit previously has been runtime optional, and is now also
14903 compile time optional via a configure switch.
14905 * systemd-analyze has been reimplemented in C. Also "systemctl
14906 dot" has moved into systemd-analyze.
14908 * "systemctl status" with no further parameters will now print
14909 the status of all active or failed units.
14911 * Operations such as "systemctl start" can now be executed
14912 with a new mode "--irreversible" which may be used to queue
14913 operations that cannot accidentally be reversed by a later
14914 job queuing. This is by default used to make shutdown
14915 requests more robust.
14917 * The Python API of systemd now gained a new module for
14918 reading journal files.
14920 * A new tool kernel-install has been added that can install
14921 kernel images according to the Boot Loader Specification:
14923 https://systemd.io/BOOT_LOADER_SPECIFICATION
14925 * Boot time console output has been improved to provide
14926 animated boot time output for hanging jobs.
14928 * A new tool systemd-activate has been added which can be used
14929 to test socket activation with, directly from the command
14930 line. This should make it much easier to test and debug
14931 socket activation in daemons.
14933 * journalctl gained a new "--reverse" (or -r) option to show
14934 journal output in reverse order (i.e. newest line first).
14936 * journalctl gained a new "--pager-end" (or -e) option to jump
14937 to immediately jump to the end of the journal in the
14938 pager. This is only supported in conjunction with "less".
14940 * journalctl gained a new "--user-unit=" option, that works
14941 similarly to "--unit=" but filters for user units rather than
14944 * A number of unit files to ease adoption of systemd in
14945 initrds has been added. This moves some minimal logic from
14946 the various initrd implementations into systemd proper.
14948 * The journal files are now owned by a new group
14949 "systemd-journal", which exists specifically to allow access
14950 to the journal, and nothing else. Previously, we used the
14951 "adm" group for that, which however possibly covers more
14952 than just journal/log file access. This new group is now
14953 already used by systemd-journal-gatewayd to ensure this
14954 daemon gets access to the journal files and as little else
14955 as possible. Note that "make install" will also set FS ACLs
14956 up for /var/log/journal to give "adm" and "wheel" read
14957 access to it, in addition to "systemd-journal" which owns
14958 the journal files. We recommend that packaging scripts also
14959 add read access to "adm" + "wheel" to /var/log/journal, and
14960 all existing/future journal files. To normal users and
14961 administrators little changes, however packagers need to
14962 ensure to create the "systemd-journal" system group at
14963 package installation time.
14965 * The systemd-journal-gatewayd now runs as unprivileged user
14966 systemd-journal-gateway:systemd-journal-gateway. Packaging
14967 scripts need to create these system user/group at
14970 * timedated now exposes a new boolean property CanNTP that
14971 indicates whether a local NTP service is available or not.
14973 * systemd-detect-virt will now also detect xen PVs
14975 * The pstore file system is now mounted by default, if it is
14978 * In addition to the SELinux and IMA policies we will now also
14979 load SMACK policies at early boot.
14981 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
14982 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
14983 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
14984 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
14985 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
14986 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
14987 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
14988 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
14989 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
14990 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
14991 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
14992 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
14993 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
14994 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
14998 * Timer units now support calendar time events in addition to
14999 monotonic time events. That means you can now trigger a unit
15000 based on a calendar time specification such as "Thu,Fri
15001 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
15002 or fifth day of any month of the year 2013, given that it is
15003 a Thursday or a Friday. This brings timer event support
15004 considerably closer to cron's capabilities. For details on
15005 the supported calendar time specification language see
15008 * udev now supports a number of different naming policies for
15009 network interfaces for predictable names, and a combination
15010 of these policies is now the default. Please see this wiki
15011 document for details:
15013 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
15015 * Auke Kok's bootchart implementation has been added to the
15016 systemd tree. It is an optional component that can graph the
15017 boot in quite some detail. It is one of the best bootchart
15018 implementations around and minimal in its code and
15021 * nss-myhostname has been integrated into the systemd source
15022 tree. nss-myhostname guarantees that the local hostname
15023 always stays resolvable via NSS. It has been a weak
15024 requirement of systemd-hostnamed since a long time, and
15025 since its code is actually trivial we decided to just
15026 include it in systemd's source tree. It can be turned off
15027 with a configure switch.
15029 * The read-ahead logic is now capable of properly detecting
15030 whether a btrfs file system is on SSD or rotating media, in
15031 order to optimize the read-ahead scheme. Previously, it was
15032 only capable of detecting this on traditional file systems
15035 * In udev, additional device properties are now read from the
15036 IAB in addition to the OUI database. Also, Bluetooth company
15037 identities are attached to the devices as well.
15039 * In service files %U may be used as specifier that is
15040 replaced by the configured user name of the service.
15042 * nspawn may now be invoked without a controlling TTY. This
15043 makes it suitable for invocation as its own service. This
15044 may be used to set up a simple containerized server system
15045 using only core OS tools.
15047 * systemd and nspawn can now accept socket file descriptors
15048 when they are started for socket activation. This enables
15049 implementation of socket activated nspawn
15050 containers. i.e. think about autospawning an entire OS image
15051 when the first SSH or HTTP connection is received. We expect
15052 that similar functionality will also be added to libvirt-lxc
15055 * journalctl will now suppress ANSI color codes when
15056 presenting log data.
15058 * systemctl will no longer show control group information for
15059 a unit if the control group is empty anyway.
15061 * logind can now automatically suspend/hibernate/shutdown the
15064 * /etc/machine-info and hostnamed now also expose the chassis
15065 type of the system. This can be used to determine whether
15066 the local system is a laptop, desktop, handset or
15067 tablet. This information may either be configured by the
15068 user/vendor or is automatically determined from ACPI and DMI
15069 information if possible.
15071 * A number of polkit actions are now bound together with "imply"
15072 rules. This should simplify creating UIs because many actions
15073 will now authenticate similar ones as well.
15075 * Unit files learnt a new condition ConditionACPower= which
15076 may be used to conditionalize a unit depending on whether an
15077 AC power source is connected or not, of whether the system
15078 is running on battery power.
15080 * systemctl gained a new "is-failed" verb that may be used in
15081 shell scripts and suchlike to check whether a specific unit
15082 is in the "failed" state.
15084 * The EnvironmentFile= setting in unit files now supports file
15085 globbing, and can hence be used to easily read a number of
15086 environment files at once.
15088 * systemd will no longer detect and recognize specific
15089 distributions. All distribution-specific #ifdeffery has been
15090 removed, systemd is now fully generic and
15091 distribution-agnostic. Effectively, not too much is lost as
15092 a lot of the code is still accessible via explicit configure
15093 switches. However, support for some distribution specific
15094 legacy configuration file formats has been dropped. We
15095 recommend distributions to simply adopt the configuration
15096 files everybody else uses now and convert the old
15097 configuration from packaging scripts. Most distributions
15098 already did that. If that's not possible or desirable,
15099 distributions are welcome to forward port the specific
15100 pieces of code locally from the git history.
15102 * When logging a message about a unit systemd will now always
15103 log the unit name in the message meta data.
15105 * localectl will now also discover system locale data that is
15106 not stored in locale archives, but directly unpacked.
15108 * logind will no longer unconditionally use framebuffer
15109 devices as seat masters, i.e. as devices that are required
15110 to be existing before a seat is considered preset. Instead,
15111 it will now look for all devices that are tagged as
15112 "seat-master" in udev. By default, framebuffer devices will
15113 be marked as such, but depending on local systems, other
15114 devices might be marked as well. This may be used to
15115 integrate graphics cards using closed source drivers (such
15116 as NVidia ones) more nicely into logind. Note however, that
15117 we recommend using the open source NVidia drivers instead,
15118 and no udev rules for the closed-source drivers will be
15119 shipped from us upstream.
15121 Contributions from: Adam Williamson, Alessandro Crismani, Auke
15122 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
15123 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
15124 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
15125 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
15126 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
15127 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
15128 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
15129 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
15130 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
15131 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
15132 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
15137 * udev gained support for loading additional device properties
15138 from an indexed database that is keyed by vendor/product IDs
15139 and similar device identifiers. For the beginning this
15140 "hwdb" is populated with data from the well-known PCI and
15141 USB database, but also includes PNP, ACPI and OID data. In
15142 the longer run this indexed database shall grow into
15143 becoming the one central database for non-essential
15144 userspace device metadata. Previously, data from the PCI/USB
15145 database was only attached to select devices, since the
15146 lookup was a relatively expensive operation due to O(n) time
15147 complexity (with n being the number of entries in the
15148 database). Since this is now O(1), we decided to add in this
15149 data for all devices where this is available, by
15150 default. Note that the indexed database needs to be rebuilt
15151 when new data files are installed. To achieve this you need
15152 to update your packaging scripts to invoke "udevadm hwdb
15153 --update" after installation of hwdb data files. For
15154 RPM-based distributions we introduced the new
15155 %udev_hwdb_update macro for this purpose.
15157 * The Journal gained support for the "Message Catalog", an
15158 indexed database to link up additional information with
15159 journal entries. For further details please check:
15161 https://www.freedesktop.org/wiki/Software/systemd/catalog
15163 The indexed message catalog database also needs to be
15164 rebuilt after installation of message catalog files. Use
15165 "journalctl --update-catalog" for this. For RPM-based
15166 distributions we introduced the %journal_catalog_update
15167 macro for this purpose.
15169 * The Python Journal bindings gained support for the standard
15170 Python logging framework.
15172 * The Journal API gained new functions for checking whether
15173 the underlying file system of a journal file is capable of
15174 properly reporting file change notifications, or whether
15175 applications that want to reflect journal changes "live"
15176 need to recheck journal files continuously in appropriate
15179 * It is now possible to set the "age" field for tmpfiles
15180 entries to 0, indicating that files matching this entry
15181 shall always be removed when the directories are cleaned up.
15183 * coredumpctl gained a new "gdb" verb which invokes gdb
15184 right-away on the selected coredump.
15186 * There's now support for "hybrid sleep" on kernels that
15187 support this, in addition to "suspend" and "hibernate". Use
15188 "systemctl hybrid-sleep" to make use of this.
15190 * logind's HandleSuspendKey= setting (and related settings)
15191 now gained support for a new "lock" setting to simply
15192 request the screen lock on all local sessions, instead of
15193 actually executing a suspend or hibernation.
15195 * systemd will now mount the EFI variables file system by
15198 * Socket units now gained support for configuration of the
15199 SMACK security label.
15201 * timedatectl will now output the time of the last and next
15202 daylight saving change.
15204 * We dropped support for various legacy and distro-specific
15205 concepts, such as insserv, early-boot SysV services
15206 (i.e. those for non-standard runlevels such as 'b' or 'S')
15207 or ArchLinux /etc/rc.conf support. We recommend the
15208 distributions who still need support this to either continue
15209 to maintain the necessary patches downstream, or find a
15210 different solution. (Talk to us if you have questions!)
15212 * Various systemd components will now bypass polkit checks for
15213 root and otherwise handle properly if polkit is not found to
15214 be around. This should fix most issues for polkit-less
15215 systems. Quite frankly this should have been this way since
15216 day one. It is absolutely our intention to make systemd work
15217 fine on polkit-less systems, and we consider it a bug if
15218 something does not work as it should if polkit is not around.
15220 * For embedded systems it is now possible to build udev and
15221 systemd without blkid and/or kmod support.
15223 * "systemctl switch-root" is now capable of switching root
15224 more than once. I.e. in addition to transitions from the
15225 initrd to the host OS it is now possible to transition to
15226 further OS images from the host. This is useful to implement
15227 offline updating tools.
15229 * Various other additions have been made to the RPM macros
15230 shipped with systemd. Use %udev_rules_update() after
15231 installing new udev rules files. %_udevhwdbdir,
15232 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
15233 %_sysctldir are now available which resolve to the right
15234 directories for packages to place various data files in.
15236 * journalctl gained the new --full switch (in addition to
15237 --all, to disable ellipsation for long messages.
15239 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
15240 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
15241 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
15242 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
15243 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
15244 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
15245 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
15246 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
15247 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
15251 * journalctl gained new --since= and --until= switches to
15252 filter by time. It also now supports nice filtering for
15253 units via --unit=/-u.
15255 * Type=oneshot services may use ExecReload= and do the
15258 * The journal daemon now supports time-based rotation and
15259 vacuuming, in addition to the usual disk-space based
15262 * The journal will now index the available field values for
15263 each field name. This enables clients to show pretty drop
15264 downs of available match values when filtering. The bash
15265 completion of journalctl has been updated
15266 accordingly. journalctl gained a new switch -F to list all
15267 values a certain field takes in the journal database.
15269 * More service events are now written as structured messages
15270 to the journal, and made recognizable via message IDs.
15272 * The timedated, localed and hostnamed mini-services which
15273 previously only provided support for changing time, locale
15274 and hostname settings from graphical DEs such as GNOME now
15275 also have a minimal (but very useful) text-based client
15276 utility each. This is probably the nicest way to changing
15277 these settings from the command line now, especially since
15278 it lists available options and is fully integrated with bash
15281 * There's now a new tool "systemd-coredumpctl" to list and
15282 extract coredumps from the journal.
15284 * We now install a README each in /var/log/ and
15285 /etc/rc.d/init.d explaining where the system logs and init
15286 scripts went. This hopefully should help folks who go to
15287 that dirs and look into the otherwise now empty void and
15288 scratch their heads.
15290 * When user-services are invoked (by systemd --user) the
15291 $MANAGERPID env var is set to the PID of systemd.
15293 * SIGRTMIN+24 when sent to a --user instance will now result
15294 in immediate termination of systemd.
15296 * gatewayd received numerous feature additions such as a
15297 "follow" mode, for live syncing and filtering.
15299 * browse.html now allows filtering and showing detailed
15300 information on specific entries. Keyboard navigation and
15301 mouse screen support has been added.
15303 * gatewayd/journalctl now supports HTML5/JSON
15304 Server-Sent-Events as output.
15306 * The SysV init script compatibility logic will now
15307 heuristically determine whether a script supports the
15308 "reload" verb, and only then make this available as
15309 "systemctl reload".
15311 * "systemctl status --follow" has been removed, use "journalctl
15314 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
15315 have been removed since they are hardly useful to be
15318 * And I'd like to take the opportunity to specifically mention
15319 Zbigniew for his great contributions. Zbigniew, you rock!
15321 Contributions from: Andrew Eikum, Christian Hesse, Colin
15322 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
15323 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
15324 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
15325 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
15326 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
15327 Jędrzejewski-Szmek, Сковорода Никита Андреевич
15331 * If /etc/vconsole.conf is non-existent or empty we will no
15332 longer load any console font or key map at boot by
15333 default. Instead the kernel defaults will be left
15334 intact. This is definitely the right thing to do, as no
15335 configuration should mean no configuration, and hard-coding
15336 font names that are different on all archs is probably a bad
15337 idea. Also, the kernel default key map and font should be
15338 good enough for most cases anyway, and mostly identical to
15339 the userspace fonts/key maps we previously overloaded them
15340 with. If distributions want to continue to default to a
15341 non-kernel font or key map they should ship a default
15342 /etc/vconsole.conf with the appropriate contents.
15344 Contributions from: Colin Walters, Daniel J Walsh, Dave
15345 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
15346 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15350 * journalctl gained a new --cursor= switch to show entries
15351 starting from the specified location in the journal.
15353 * We now enforce a size limit on journal entry fields exported
15354 with "-o json" in journalctl. Fields larger than 4K will be
15355 assigned null. This can be turned off with --all.
15357 * An (optional) journal gateway daemon is now available as
15358 "systemd-journal-gatewayd.service". This service provides
15359 access to the journal via HTTP and JSON. This functionality
15360 will be used to implement live log synchronization in both
15361 pull and push modes, but has various other users too, such
15362 as easy log access for debugging of embedded devices. Right
15363 now it is already useful to retrieve the journal via HTTP:
15365 # systemctl start systemd-journal-gatewayd.service
15366 # wget http://localhost:19531/entries
15368 This will download the journal contents in a
15369 /var/log/messages compatible format. The same as JSON:
15371 # curl -H"Accept: application/json" http://localhost:19531/entries
15373 This service is also accessible via a web browser where a
15374 single static HTML5 app is served that uses the JSON logic
15375 to enable the user to do some basic browsing of the
15376 journal. This will be extended later on. Here's an example
15377 screenshot of this app in its current state:
15379 https://0pointer.de/public/journal-gatewayd
15381 Contributions from: Kay Sievers, Lennart Poettering, Robert
15382 Milasan, Tom Gundersen
15386 * The bash completion logic is now available for journalctl
15389 * We do not mount the "cpuset" controller anymore together with
15390 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
15391 started if no parameters are assigned to it. "cpuset" hence
15392 broke code that assumed it could create "cpu" groups and
15395 * journalctl -f will now subscribe to terminal size changes,
15396 and line break accordingly.
15398 Contributions from: Dave Reisner, Kay Sievers, Lennart
15399 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
15403 * nspawn will now create a symlink /etc/localtime in the
15404 container environment, copying the host's timezone
15405 setting. Previously this has been done via a bind mount, but
15406 since symlinks cannot be bind mounted this has now been
15407 changed to create/update the appropriate symlink.
15409 * journalctl -n's line number argument is now optional, and
15410 will default to 10 if omitted.
15412 * journald will now log the maximum size the journal files may
15413 take up on disk. This is particularly useful if the default
15414 built-in logic of determining this parameter from the file
15415 system size is used. Use "systemctl status
15416 systemd-journald.service" to see this information.
15418 * The multi-seat X wrapper tool has been stripped down. As X
15419 is now capable of enumerating graphics devices via udev in a
15420 seat-aware way the wrapper is not strictly necessary
15421 anymore. A stripped down temporary stop-gap is still shipped
15422 until the upstream display managers have been updated to
15423 fully support the new X logic. Expect this wrapper to be
15424 removed entirely in one of the next releases.
15426 * HandleSleepKey= in logind.conf has been split up into
15427 HandleSuspendKey= and HandleHibernateKey=. The old setting
15428 is not available anymore. X11 and the kernel are
15429 distinguishing between these keys and we should too. This
15430 also means the inhibition lock for these keys has been split
15433 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
15434 Poettering, Lukas Nykryn, Václav Pavlín
15438 * Whenever a unit changes state we will now log this to the
15439 journal and show along the unit's own log output in
15440 "systemctl status".
15442 * ConditionPathIsMountPoint= can now properly detect bind
15443 mount points too. (Previously, a bind mount of one file
15444 system to another place in the same file system could not be
15445 detected as mount, since they shared struct stat's st_dev
15448 * We will now mount the cgroup controllers cpu, cpuacct,
15449 cpuset and the controllers net_cls, net_prio together by
15452 * nspawn containers will now have a virtualized boot
15453 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
15454 over with a randomized ID at container initialization). This
15455 has the effect of making "journalctl -b" do the right thing
15458 * The JSON output journal serialization has been updated not
15459 to generate "endless" list objects anymore, but rather one
15460 JSON object per line. This is more in line how most JSON
15461 parsers expect JSON objects. The new output mode
15462 "json-pretty" has been added to provide similar output, but
15463 neatly aligned for readability by humans.
15465 * We dropped all explicit sync() invocations in the shutdown
15466 code. The kernel does this implicitly anyway in the kernel
15467 reboot() syscall. halt(8)'s -n option is now a compatibility
15470 * We now support virtualized reboot() in containers, as
15471 supported by newer kernels. We will fall back to exit() if
15472 CAP_SYS_REBOOT is not available to the container. Also,
15473 nspawn makes use of this now and will actually reboot the
15474 container if the containerized OS asks for that.
15476 * journalctl will only show local log output by default
15477 now. Use --merge (-m) to show remote log output, too.
15479 * libsystemd-journal gained the new sd_journal_get_usage()
15480 call to determine the current disk usage of all journal
15481 files. This is exposed in the new "journalctl --disk-usage"
15484 * journald gained a new configuration setting SplitMode= in
15485 journald.conf which may be used to control how user journals
15486 are split off. See journald.conf(5) for details.
15488 * A new condition type ConditionFileNotEmpty= has been added.
15490 * tmpfiles' "w" lines now support file globbing, to write
15491 multiple files at once.
15493 * We added Python bindings for the journal submission
15494 APIs. More Python APIs for a number of selected APIs will
15495 likely follow. Note that we intend to add native bindings
15496 only for the Python language, as we consider it common
15497 enough to deserve bindings shipped within systemd. There are
15498 various projects outside of systemd that provide bindings
15499 for languages such as PHP or Lua.
15501 * Many conditions will now resolve specifiers such as %i. In
15502 addition, PathChanged= and related directives of .path units
15503 now support specifiers as well.
15505 * There's now a new RPM macro definition for the system preset
15508 * journald will now warn if it ca not forward a message to the
15509 syslog daemon because its socket is full.
15511 * timedated will no longer write or process /etc/timezone,
15512 except on Debian. As we do not support late mounted /usr
15513 anymore /etc/localtime always being a symlink is now safe,
15514 and hence the information in /etc/timezone is not necessary
15517 * logind will now always reserve one VT for a text getty (VT6
15518 by default). Previously if more than 6 X sessions where
15519 started they took up all the VTs with auto-spawned gettys,
15520 so that no text gettys were available anymore.
15522 * udev will now automatically inform the btrfs kernel logic
15523 about btrfs RAID components showing up. This should make
15524 simple hotplug based btrfs RAID assembly work.
15526 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
15527 (but not for its children which will stay at the kernel
15528 default). This should allow setups with a lot more listening
15531 * systemd will now always pass the configured timezone to the
15532 kernel at boot. timedated will do the same when the timezone
15535 * logind's inhibition logic has been updated. By default,
15536 logind will now handle the lid switch, the power and sleep
15537 keys all the time, even in graphical sessions. If DEs want
15538 to handle these events on their own they should take the new
15539 handle-power-key, handle-sleep-key and handle-lid-switch
15540 inhibitors during their runtime. A simple way to achieve
15541 that is to invoke the DE wrapped in an invocation of:
15543 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
15545 * Access to unit operations is now checked via SELinux taking
15546 the unit file label and client process label into account.
15548 * systemd will now notify the administrator in the journal
15549 when he over-mounts a non-empty directory.
15551 * There are new specifiers that are resolved in unit files,
15552 for the hostname (%H), the machine ID (%m) and the boot ID
15555 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
15556 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
15557 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
15558 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
15559 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
15560 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
15561 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
15565 * Support for reading structured kernel messages from
15566 /dev/kmsg has now been added and is enabled by default.
15568 * Support for reading kernel messages from /proc/kmsg has now
15569 been removed. If you want kernel messages in the journal
15570 make sure to run a recent kernel (>= 3.5) that supports
15571 reading structured messages from /dev/kmsg (see
15572 above). /proc/kmsg is now exclusive property of classic
15573 syslog daemons again.
15575 * The libudev API gained the new
15576 udev_device_new_from_device_id() call.
15578 * The logic for file system namespace (ReadOnlyDirectory=,
15579 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
15580 require pivot_root() anymore. This means fewer temporary
15581 directories are created below /tmp for this feature.
15583 * nspawn containers will now see and receive all submounts
15584 made on the host OS below the root file system of the
15587 * Forward Secure Sealing is now supported for Journal files,
15588 which provide cryptographical sealing of journal files so
15589 that attackers cannot alter log history anymore without this
15590 being detectable. Lennart will soon post a blog story about
15591 this explaining it in more detail.
15593 * There are two new service settings RestartPreventExitStatus=
15594 and SuccessExitStatus= which allow configuration of exit
15595 status (exit code or signal) which will be excepted from the
15596 restart logic, resp. consider successful.
15598 * journalctl gained the new --verify switch that can be used
15599 to check the integrity of the structure of journal files and
15600 (if Forward Secure Sealing is enabled) the contents of
15603 * nspawn containers will now be run with /dev/stdin, /dev/fd/
15604 and similar symlinks pre-created. This makes running shells
15605 as container init process a lot more fun.
15607 * The fstab support can now handle PARTUUID= and PARTLABEL=
15610 * A new ConditionHost= condition has been added to match
15611 against the hostname (with globs) and machine ID. This is
15612 useful for clusters where a single OS image is used to
15613 provision a large number of hosts which shall run slightly
15614 different sets of services.
15616 * Services which hit the restart limit will now be placed in a
15619 Contributions from: Bertram Poettering, Dave Reisner, Huang
15620 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
15621 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
15625 * When running in --user mode systemd will now become a
15626 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
15627 tree a lot more organized.
15629 * A new PartOf= unit dependency type has been introduced that
15630 may be used to group services in a natural way.
15632 * "systemctl enable" may now be used to enable instances of
15635 * journalctl now prints error log levels in red, and
15636 warning/notice log levels in bright white. It also supports
15637 filtering by log level now.
15639 * cgtop gained a new -n switch (similar to top), to configure
15640 the maximum number of iterations to run for. It also gained
15641 -b, to run in batch mode (accepting no input).
15643 * The suffix ".service" may now be omitted on most systemctl
15644 command lines involving service unit names.
15646 * There's a new bus call in logind to lock all sessions, as
15647 well as a loginctl verb for it "lock-sessions".
15649 * libsystemd-logind.so gained a new call sd_journal_perror()
15650 that works similar to libc perror() but logs to the journal
15651 and encodes structured information about the error number.
15653 * /etc/crypttab entries now understand the new keyfile-size=
15656 * shutdown(8) now can send a (configurable) wall message when
15657 a shutdown is cancelled.
15659 * The mount propagation mode for the root file system will now
15660 default to "shared", which is useful to make containers work
15661 nicely out-of-the-box so that they receive new mounts from
15662 the host. This can be undone locally by running "mount
15663 --make-rprivate /" if needed.
15665 * The prefdm.service file has been removed. Distributions
15666 should maintain this unit downstream if they intend to keep
15667 it around. However, we recommend writing normal unit files
15668 for display managers instead.
15670 * Since systemd is a crucial part of the OS we will now
15671 default to a number of compiler switches that improve
15672 security (hardening) such as read-only relocations, stack
15673 protection, and suchlike.
15675 * The TimeoutSec= setting for services is now split into
15676 TimeoutStartSec= and TimeoutStopSec= to allow configuration
15677 of individual time outs for the start and the stop phase of
15680 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
15681 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
15682 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
15683 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
15684 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
15685 Gundersen, Zbigniew Jędrzejewski-Szmek
15689 * The journal and id128 C APIs are now fully documented as man
15692 * Extra safety checks have been added when transitioning from
15693 the initial RAM disk to the main system to avoid accidental
15696 * /etc/crypttab entries now understand the new keyfile-offset=
15699 * systemctl -t can now be used to filter by unit load state.
15701 * The journal C API gained the new sd_journal_wait() call to
15702 make writing synchronous journal clients easier.
15704 * journalctl gained the new -D switch to show journals from a
15705 specific directory.
15707 * journalctl now displays a special marker between log
15708 messages of two different boots.
15710 * The journal is now explicitly flushed to /var via a service
15711 systemd-journal-flush.service, rather than implicitly simply
15712 by seeing /var/log/journal to be writable.
15714 * journalctl (and the journal C APIs) can now match for much
15715 more complex expressions, with alternatives and
15718 * When transitioning from the initial RAM disk to the main
15719 system we will now kill all processes in a killing spree to
15720 ensure no processes stay around by accident.
15722 * Three new specifiers may be used in unit files: %u, %h, %s
15723 resolve to the user name, user home directory resp. user
15724 shell. This is useful for running systemd user instances.
15726 * We now automatically rotate journal files if their data
15727 object hash table gets a fill level > 75%. We also size the
15728 hash table based on the configured maximum file size. This
15729 together should lower hash collisions drastically and thus
15730 speed things up a bit.
15732 * journalctl gained the new "--header" switch to introspect
15733 header data of journal files.
15735 * A new setting SystemCallFilters= has been added to services which may
15736 be used to apply deny lists or allow lists to system calls. This is
15737 based on SECCOMP Mode 2 of Linux 3.5.
15739 * nspawn gained a new --link-journal= switch (and quicker: -j)
15740 to link the container journal with the host. This makes it
15741 very easy to centralize log viewing on the host for all
15742 guests while still keeping the journal files separated.
15744 * Many bugfixes and optimizations
15746 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
15747 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
15748 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
15753 * Several tools now understand kernel command line arguments,
15754 which are only read when run in an initial RAM disk. They
15755 usually follow closely their normal counterparts, but are
15758 * There's a new tool to analyze the readahead files that are
15759 automatically generated at boot. Use:
15761 /usr/lib/systemd/systemd-readahead analyze /.readahead
15763 * We now provide an early debug shell on tty9 if this enabled. Use:
15765 systemctl enable debug-shell.service
15767 * All plymouth related units have been moved into the Plymouth
15768 package. Please make sure to upgrade your Plymouth version
15771 * systemd-tmpfiles now supports getting passed the basename of
15772 a configuration file only, in which case it will look for it
15773 in all appropriate directories automatically.
15775 * udevadm info now takes a /dev or /sys path as argument, and
15776 does the right thing. Example:
15778 udevadm info /dev/sda
15779 udevadm info /sys/class/block/sda
15781 * systemctl now prints a warning if a unit is stopped but a
15782 unit that might trigger it continues to run. Example: a
15783 service is stopped but the socket that activates it is left
15786 * "systemctl status" will now mention if the log output was
15787 shortened due to rotation since a service has been started.
15789 * The journal API now exposes functions to determine the
15790 "cutoff" times due to rotation.
15792 * journald now understands SIGUSR1 and SIGUSR2 for triggering
15793 immediately flushing of runtime logs to /var if possible,
15794 resp. for triggering immediate rotation of the journal
15797 * It is now considered an error if a service is attempted to
15798 be stopped that is not loaded.
15800 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
15802 * systemd-analyze now supports Python 3
15804 * tmpfiles now supports cleaning up directories via aging
15805 where the first level dirs are always kept around but
15806 directories beneath it automatically aged. This is enabled
15807 by prefixing the age field with '~'.
15809 * Seat objects now expose CanGraphical, CanTTY properties
15810 which is required to deal with very fast bootups where the
15811 display manager might be running before the graphics drivers
15812 completed initialization.
15814 * Seat objects now expose a State property.
15816 * We now include RPM macros for service enabling/disabling
15817 based on the preset logic. We recommend RPM based
15818 distributions to make use of these macros if possible. This
15819 makes it simpler to reuse RPM spec files across
15822 * We now make sure that the collected systemd unit name is
15823 always valid when services log to the journal via
15826 * There's a new man page kernel-command-line(7) detailing all
15827 command line options we understand.
15829 * The fstab generator may now be disabled at boot by passing
15830 fstab=0 on the kernel command line.
15832 * A new kernel command line option modules-load= is now understood
15833 to load a specific kernel module statically, early at boot.
15835 * Unit names specified on the systemctl command line are now
15836 automatically escaped as needed. Also, if file system or
15837 device paths are specified they are automatically turned
15838 into the appropriate mount or device unit names. Example:
15840 systemctl status /home
15841 systemctl status /dev/sda
15843 * The SysVConsole= configuration option has been removed from
15844 system.conf parsing.
15846 * The SysV search path is no longer exported on the D-Bus
15849 * The Names= option has been removed from unit file parsing.
15851 * There's a new man page bootup(7) detailing the boot process.
15853 * Every unit and every generator we ship with systemd now
15854 comes with full documentation. The self-explanatory boot is
15857 * A couple of services gained "systemd-" prefixes in their
15858 name if they wrap systemd code, rather than only external
15859 code. Among them fsck@.service which is now
15860 systemd-fsck@.service.
15862 * The HaveWatchdog property has been removed from the D-Bus
15865 * systemd.confirm_spawn= on the kernel command line should now
15868 * There's a new man page crypttab(5) which details all options
15869 we actually understand.
15871 * systemd-nspawn gained a new --capability= switch to pass
15872 additional capabilities to the container.
15874 * timedated will now read known NTP implementation unit names
15875 from /usr/lib/systemd/ntp-units.d/*.list,
15876 systemd-timedated-ntp.target has been removed.
15878 * journalctl gained a new switch "-b" that lists log data of
15879 the current boot only.
15881 * The notify socket is in the abstract namespace again, in
15882 order to support daemons which chroot() at start-up.
15884 * There is a new Storage= configuration option for journald
15885 which allows configuration of where log data should go. This
15886 also provides a way to disable journal logging entirely, so
15887 that data collected is only forwarded to the console, the
15888 kernel log buffer or another syslog implementation.
15890 * Many bugfixes and optimizations
15892 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
15893 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
15894 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
15895 Shawn Landden, Tom Gundersen
15899 * "systemctl help <unit>" now shows the man page if one is
15902 * Several new man pages have been added.
15904 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
15905 MaxLevelConsole= can now be specified in
15906 journald.conf. These options allow reducing the amount of
15907 data stored on disk or forwarded by the log level.
15909 * TimerSlackNSec= can now be specified in system.conf for
15910 PID1. This allows system-wide power savings.
15912 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
15913 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
15918 * logind is now capable of (optionally) handling power and
15919 sleep keys as well as the lid switch.
15921 * journalctl now understands the syntax "journalctl
15922 /usr/bin/avahi-daemon" to get all log output of a specific
15925 * CapabilityBoundingSet= in system.conf now also influences
15926 the capability bound set of usermode helpers of the kernel.
15928 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
15929 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
15930 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
15931 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
15935 * Note that we skipped 139 releases here in order to set the
15936 new version to something that is greater than both udev's
15937 and systemd's most recent version number.
15939 * udev: all udev sources are merged into the systemd source tree now.
15940 All future udev development will happen in the systemd tree. It
15941 is still fully supported to use the udev daemon and tools without
15942 systemd running, like in initramfs or other init systems. Building
15943 udev though, will require the *build* of the systemd tree, but
15944 udev can be properly *run* without systemd.
15946 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
15947 should be used to create dead device nodes as workarounds for broken
15950 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
15951 no longer supported. udev_monitor_new_from_netlink() needs to be
15952 used to subscribe to events.
15954 * udev: when udevd is started by systemd, processes which are left
15955 behind by forking them off of udev rules, are unconditionally cleaned
15956 up and killed now after the event handling has finished. Services or
15957 daemons must be started as systemd services. Services can be
15958 pulled-in by udev to get started, but they can no longer be directly
15959 forked by udev rules.
15961 * udev: the daemon binary is called systemd-udevd now and installed
15962 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
15963 to adapt to that, create symlink, or rename the binary after building
15966 * libudev no longer provides these symbols:
15967 udev_monitor_from_socket()
15968 udev_queue_get_failed_list_entry()
15969 udev_get_{dev,sys,run}_path()
15970 The versions number was bumped and symbol versioning introduced.
15972 * systemd-loginctl and systemd-journalctl have been renamed
15973 to loginctl and journalctl to match systemctl.
15975 * The config files: /etc/systemd/systemd-logind.conf and
15976 /etc/systemd/systemd-journald.conf have been renamed to
15977 logind.conf and journald.conf. Package updates should rename
15978 the files to the new names on upgrade.
15980 * For almost all files the license is now LGPL2.1+, changed
15981 from the previous GPL2.0+. Exceptions are some minor stuff
15982 of udev (which will be changed to LGPL2.1 eventually, too),
15983 and the MIT licensed sd-daemon.[ch] library that is suitable
15984 to be used as drop-in files.
15986 * systemd and logind now handle system sleep states, in
15987 particular suspending and hibernating.
15989 * logind now implements a sleep/shutdown/idle inhibiting logic
15990 suitable for a variety of uses. Soonishly Lennart will blog
15991 about this in more detail.
15993 * var-run.mount and var-lock.mount are no longer provided
15994 (which previously bind mounted these directories to their new
15995 places). Distributions which have not converted these
15996 directories to symlinks should consider stealing these files
15997 from git history and add them downstream.
15999 * We introduced the Documentation= field for units and added
16000 this to all our shipped units. This is useful to make it
16001 easier to explore the boot and the purpose of the various
16004 * All smaller setup units (such as
16005 systemd-vconsole-setup.service) now detect properly if they
16006 are run in a container and are skipped when
16007 appropriate. This guarantees an entirely noise-free boot in
16008 Linux container environments such as systemd-nspawn.
16010 * A framework for implementing offline system updates is now
16011 integrated, for details see:
16012 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
16014 * A new service type Type=idle is available now which helps us
16015 avoiding ugly interleaving of getty output and boot status
16018 * There's now a system-wide CapabilityBoundingSet= option to
16019 globally reduce the set of capabilities for the
16020 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
16021 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
16022 even CAP_NET_ADMIN system-wide for secure systems.
16024 * There are now system-wide DefaultLimitXXX= options to
16025 globally change the defaults of the various resource limits
16026 for all units started by PID 1.
16028 * Harald Hoyer's systemd test suite has been integrated into
16029 systemd which allows easy testing of systemd builds in qemu
16030 and nspawn. (This is really awesome! Ask us for details!)
16032 * The fstab parser is now implemented as generator, not inside
16035 * systemctl will now warn you if .mount units generated from
16036 /etc/fstab are out of date due to changes in fstab that
16037 have not been read by systemd yet.
16039 * systemd is now suitable for usage in initrds. Dracut has
16040 already been updated to make use of this. With this in place
16041 initrds get a slight bit faster but primarily are much
16042 easier to introspect and debug since "systemctl status" in
16043 the host system can be used to introspect initrd services,
16044 and the journal from the initrd is kept around too.
16046 * systemd-delta has been added, a tool to explore differences
16047 between user/admin configuration and vendor defaults.
16049 * PrivateTmp= now affects both /tmp and /var/tmp.
16051 * Boot time status messages are now much prettier and feature
16052 proper english language. Booting up systemd has never been
16055 * Read-ahead pack files now include the inode number of all
16056 files to pre-cache. When the inode changes the pre-caching
16057 is not attempted. This should be nicer to deal with updated
16058 packages which might result in changes of read-ahead
16061 * We now temporaritly lower the kernel's read_ahead_kb variable
16062 when collecting read-ahead data to ensure the kernel's
16063 built-in read-ahead does not add noise to our measurements
16064 of necessary blocks to pre-cache.
16066 * There's now RequiresMountsFor= to add automatic dependencies
16067 for all mounts necessary for a specific file system path.
16069 * MountAuto= and SwapAuto= have been removed from
16070 system.conf. Mounting file systems at boot has to take place
16073 * nspawn now learned a new switch --uuid= to set the machine
16074 ID on the command line.
16076 * nspawn now learned the -b switch to automatically search
16077 for an init system.
16079 * vt102 is now the default TERM for serial TTYs, upgraded from
16082 * systemd-logind now works on VT-less systems.
16084 * The build tree has been reorganized. The individual
16085 components now have directories of their own.
16087 * A new condition type ConditionPathIsReadWrite= is now available.
16089 * nspawn learned the new -C switch to create cgroups for the
16090 container in other hierarchies.
16092 * We now have support for hardware watchdogs, configurable in
16095 * The scheduled shutdown logic now has a public API.
16097 * We now mount /tmp as tmpfs by default, but this can be
16098 masked and /etc/fstab can override it.
16100 * Since udisks does not make use of /media anymore we are not
16101 mounting a tmpfs on it anymore.
16103 * journalctl gained a new --local switch to only interleave
16104 locally generated journal files.
16106 * We can now load the IMA policy at boot automatically.
16108 * The GTK tools have been split off into a systemd-ui.
16110 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
16111 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
16112 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
16113 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
16114 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
16115 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
16116 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
16117 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
16118 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
16123 * This is mostly a bugfix release
16125 * Support optional initialization of the machine ID from the
16126 KVM or container configured UUID.
16128 * Support immediate reboots with "systemctl reboot -ff"
16130 * Show /etc/os-release data in systemd-analyze output
16132 * Many bugfixes for the journal, including endianness fixes and
16133 ensuring that disk space enforcement works
16135 * sd-login.h is C++ compatible again
16137 * Extend the /etc/os-release format on request of the Debian
16140 * We now refuse non-UTF8 strings used in various configuration
16141 and unit files. This is done to ensure we do not pass invalid
16142 data over D-Bus or expose it elsewhere.
16144 * Register Mimo USB Screens as suitable for automatic seat
16147 * Read SELinux client context from journal clients in a race
16150 * Reorder configuration file lookup order. /etc now always
16151 overrides /run in order to allow the administrator to always
16152 and unconditionally override vendor-supplied or
16153 automatically generated data.
16155 * The various user visible bits of the journal now have man
16156 pages. We still lack man pages for the journal API calls
16159 * We now ship all man pages in HTML format again in the
16162 Contributions from: Dave Reisner, Dirk Eibach, Frederic
16163 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
16164 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
16169 * This is mostly a bugfix release
16171 * systems lacking /etc/os-release are no longer supported.
16173 * Various functionality updates to libsystemd-login.so
16175 * Track class of PAM logins to distinguish greeters from
16176 normal user logins.
16178 Contributions from: Kay Sievers, Lennart Poettering, Michael
16183 * This is an important bugfix release for v41.
16185 * Building man pages is now optional which should be useful
16186 for those building systemd from git but unwilling to install
16189 * Watchdog support for supervising services is now usable. In
16190 a future release support for hardware watchdogs
16191 (i.e. /dev/watchdog) will be added building on this.
16193 * Service start rate limiting is now configurable and can be
16194 turned off per service. When a start rate limit is hit a
16195 reboot can automatically be triggered.
16197 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
16199 Contributions from: Benjamin Franzke, Bill Nottingham,
16200 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
16201 Schmidt, Michał Górny, Piotr Drąg
16205 * The systemd binary is installed /usr/lib/systemd/systemd now;
16206 An existing /sbin/init symlink needs to be adapted with the
16209 * The code that loads kernel modules has been ported to invoke
16210 libkmod directly, instead of modprobe. This means we do not
16211 support systems with module-init-tools anymore.
16213 * Watchdog support is now already useful, but still not
16216 * A new kernel command line option systemd.setenv= is
16217 understood to set system wide environment variables
16218 dynamically at boot.
16220 * We now limit the set of capabilities of systemd-journald.
16222 * We now set SIGPIPE to ignore by default, since it only is
16223 useful in shell pipelines, and has little use in general
16224 code. This can be disabled with IgnoreSIPIPE=no in unit
16227 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
16228 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
16233 * This is mostly a bugfix release
16235 * We now expose the reason why a service failed in the
16236 "Result" D-Bus property.
16238 * Rudimentary service watchdog support (will be completed over
16239 the next few releases.)
16241 * When systemd forks off in order execute some service we will
16242 now immediately changes its argv[0] to reflect which process
16243 it will execute. This is useful to minimize the time window
16244 with a generic argv[0], which makes bootcharts more useful
16246 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
16247 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
16248 Mike Kazantsev, Ray Strode
16252 * This is mostly a test release, but incorporates many
16255 * New systemd-cgtop tool to show control groups by their
16258 * Linking against libacl for ACLs is optional again. If
16259 disabled, support tracking device access for active logins
16260 goes becomes unavailable, and so does access to the user
16261 journals by the respective users.
16263 * If a group "adm" exists, journal files are automatically
16264 owned by them, thus allow members of this group full access
16265 to the system journal as well as all user journals.
16267 * The journal now stores the SELinux context of the logging
16268 client for all entries.
16270 * Add C++ inclusion guards to all public headers
16272 * New output mode "cat" in the journal to print only text
16273 messages, without any meta data like date or time.
16275 * Include tiny X server wrapper as a temporary stop-gap to
16276 teach XOrg udev display enumeration. This is used by display
16277 managers such as gdm, and will go away as soon as XOrg
16278 learned native udev hotplugging for display devices.
16280 * Add new systemd-cat tool for executing arbitrary programs
16281 with STDERR/STDOUT connected to the journal. Can also act as
16282 BSD logger replacement, and does so by default.
16284 * Optionally store all locally generated coredumps in the
16285 journal along with meta data.
16287 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
16288 writing short strings to files (for usage for /sys), and for
16289 creating symlinks, character and block device nodes.
16291 * New unit file option ControlGroupPersistent= to make cgroups
16292 persistent, following the mechanisms outlined in
16293 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
16295 * Support multiple local RTCs in a sane way
16297 * No longer monopolize IO when replaying readahead data on
16298 rotating disks, since we might starve non-file-system IO to
16299 death, since fanotify() will not see accesses done by blkid,
16302 * Do not show kernel threads in systemd-cgls anymore, unless
16303 requested with new -k switch.
16305 Contributions from: Dan Horák, Kay Sievers, Lennart
16306 Poettering, Michal Schmidt
16310 * This is mostly a test release, but incorporates many
16313 * The git repository moved to:
16314 git://anongit.freedesktop.org/systemd/systemd
16315 ssh://git.freedesktop.org/git/systemd/systemd
16317 * First release with the journal
16318 https://0pointer.de/blog/projects/the-journal.html
16320 * The journal replaces both systemd-kmsg-syslogd and
16321 systemd-stdout-bridge.
16323 * New sd_pid_get_unit() API call in libsystemd-logind
16325 * Many systemadm clean-ups
16327 * Introduce remote-fs-pre.target which is ordered before all
16328 remote mounts and may be used to start services before all
16331 * Added Mageia support
16333 * Add bash completion for systemd-loginctl
16335 * Actively monitor PID file creation for daemons which exit in
16336 the parent process before having finished writing the PID
16337 file in the daemon process. Daemons which do this need to be
16338 fixed (i.e. PID file creation must have finished before the
16339 parent exits), but we now react a bit more gracefully to them.
16341 * Add colourful boot output, mimicking the well-known output
16342 of existing distributions.
16344 * New option PassCredentials= for socket units, for
16345 compatibility with a recent kernel ABI breakage.
16347 * /etc/rc.local is now hooked in via a generator binary, and
16348 thus will no longer act as synchronization point during
16351 * systemctl list-unit-files now supports --root=.
16353 * systemd-tmpfiles now understands two new commands: z, Z for
16354 relabelling files according to the SELinux database. This is
16355 useful to apply SELinux labels to specific files in /sys,
16356 among other things.
16358 * Output of SysV services is now forwarded to both the console
16359 and the journal by default, not only just the console.
16361 * New man pages for all APIs from libsystemd-login.
16363 * The build tree got reorganized and the build system is a
16364 lot more modular allowing embedded setups to specifically
16365 select the components of systemd they are interested in.
16367 * Support for Linux systems lacking the kernel VT subsystem is
16370 * configure's --with-rootdir= got renamed to
16371 --with-rootprefix= to follow the naming used by udev and
16374 * Unless specified otherwise we will now install to /usr instead
16375 of /usr/local by default.
16377 * Processes with '@' in argv[0][0] are now excluded from the
16378 final shut-down killing spree, following the logic explained
16380 https://systemd.io/ROOT_STORAGE_DAEMONS/
16382 * All processes remaining in a service cgroup when we enter
16383 the START or START_PRE states are now killed with
16384 SIGKILL. That means it is no longer possible to spawn
16385 background processes from ExecStart= lines (which was never
16386 supported anyway, and bad style).
16388 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
16389 reloading of units together.
16391 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
16392 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
16393 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
16394 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
16395 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek