1 systemd System and Service Manager
3 CHANGES WITH 255 in spe:
5 Announcements of Future Feature Removals and Incompatible Changes:
7 * Support for split-usr (/usr/ mounted separately during late boot,
8 instead of being mounted by the initrd before switching to the rootfs)
9 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
10 /usr/lib/, …) has been removed. For more details, see:
11 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
13 * We intend to remove cgroup v1 support from a systemd release after
14 the end of 2023. If you run services that make explicit use of
15 cgroup v1 features (i.e. the "legacy hierarchy" with separate
16 hierarchies for each controller), please implement compatibility with
17 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
18 Most of Linux userspace has been ported over already.
20 * Support for System V service scripts is now deprecated and will be
21 removed in a future release. Please make sure to update your software
22 *now* to include a native systemd unit file instead of a legacy
23 System V script to retain compatibility with future systemd releases.
25 * Support for the SystemdOptions EFI variable is deprecated.
26 'bootctl systemd-efi-options' will emit a warning when used. It seems
27 that this feature is little-used and it is better to use alternative
28 approaches like credentials and confexts. The plan is to drop support
29 altogether at a later point, but this might be revisited based on
32 * systemd-run's switch --expand-environment= which currently is disabled
33 by default when combined with --scope, will be changed in a future
34 release to be enabled by default.
36 * "systemctl switch-root" is now restricted to initrd transitions only.
37 Transitions between real systems should be done with "systemctl soft-reboot"
40 * The ip=off and ip=none kernel command line options interpreted by
41 systemd-network-generator will now result in IPv6RA + link-local
42 addressing to be disabled, too. Previously DHCP was turned off, but
43 IPv6RA and IPv6 link-local addressing was left enabled.
47 * udev will now create symlinks to loopback block devices in the
48 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
49 string field selected during allocation. The systemd-dissect tool and
50 the util-linux losetup command now supports a complementing new
51 switch --loop-ref= for selecting the string. This means a loopback
52 block device may now be allocated under a caller-chosen reference and
53 can subsequently be referenced by that without first having to look
54 up the block device name the caller ended up with.
56 * udev also creates symlinks to loopback block devices in the
57 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
58 of the inode attached to the loopback block device. This means that
59 attaching a file to a loopback device will implicitly make a handle
60 available to be found via that file's inode information.
64 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
65 is now dropped, as it never worked, hence it should not be used by
68 * The predictable network interface naming logic is extended to include
69 the SR-IOV-R "representor" information in network interface names.
70 This feature was intended for v254, but even though the code was
71 merged, the part that actually enabled the feature was forgotten.
72 It is now enabled by default and is part of the new "v255" naming
75 * The 'prefixstable' ipv6 address generation mode now considers the
76 SSID when generating stable addresses, so that a different stable
77 address is used when roaming between wireless networks. If you
78 already use 'prefixstable' addresses with wireless networks, the
79 stable address chosen will be changed by the update.
81 * The DHCPv4 client gained a RapidCommit option, default true, which
82 enables RFC4039 Rapid Commit behavior to obtain a lease in a
83 simplified 2-message exchange instead of the typical 4-message
84 exchange if also supported by the DHCP server.
86 * The SendHostname and Hostname options are now available for the
87 DHCPv6 client, independent of the DHCPv4 option, so that these
88 configuration values can be set independently for each client.
90 Changes in systemd-analyze:
92 * "systemd-analyze plot" has gained tooltips on each unit name with
93 related-unit information in its svg output, such as Before=,
94 Requires=, and similar properties.
98 Announcements of Future Feature Removals and Incompatible Changes:
100 * The next release (v255) will remove support for split-usr (/usr/
101 mounted separately during late boot, instead of being mounted by the
102 initrd before switching to the rootfs) and unmerged-usr (parallel
103 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
105 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
107 * We intend to remove cgroup v1 support from a systemd release after
108 the end of 2023. If you run services that make explicit use of
109 cgroup v1 features (i.e. the "legacy hierarchy" with separate
110 hierarchies for each controller), please implement compatibility with
111 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
112 Most of Linux userspace has been ported over already.
114 * Support for System V service scripts is now deprecated and will be
115 removed in a future release. Please make sure to update your software
116 *now* to include a native systemd unit file instead of a legacy
117 System V script to retain compatibility with future systemd releases.
119 * Support for the SystemdOptions EFI variable is deprecated.
120 'bootctl systemd-efi-options' will emit a warning when used. It seems
121 that this feature is little-used and it is better to use alternative
122 approaches like credentials and confexts. The plan is to drop support
123 altogether at a later point, but this might be revisited based on
126 * EnvironmentFile= now treats the line following a comment line
127 trailing with escape as a non comment line. For details, see:
128 https://github.com/systemd/systemd/issues/27975
130 * PrivateNetwork=yes and NetworkNamespacePath= now imply
131 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
133 * Behaviour of sandboxing options for the per-user service manager
134 units has changed. They now imply PrivateUsers=yes, which means user
135 namespaces will be implicitly enabled when a sandboxing option is
136 enabled in a user unit. Enabling user namespaces has the drawback
137 that system users will no longer be visible (and processes/files will
138 appear as owned by 'nobody') in the user unit.
140 By definition a sandboxed user unit should run with reduced
141 privileges, so impact should be small. This will remove a great
142 source of confusion that has been reported by users over the years,
143 due to how these options require an extra setting to be manually
144 enabled when used in the per-user service manager, which is not
145 needed in the system service manager. For more details, see:
146 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
148 * systemd-run's switch --expand-environment= which currently is disabled
149 by default when combined with --scope, will be changed in a future
150 release to be enabled by default.
152 Security Relevant Changes:
154 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
155 process capability to invoked session processes of regular users on
156 local seats (as well as to systemd --user), unless configured
157 otherwise via data from JSON user records, or via the PAM module's
158 parameter list. This is useful in order allow desktop tools such as
159 GNOME's Alarm Clock application to set a timer for
160 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
161 per-user service unit file may thus use AmbientCapability= to pass
162 the capability to invoked processes. Note that this capability is
163 relatively narrow in focus (in particular compared to other process
164 capabilities such as CAP_SYS_ADMIN) and we already — by default —
165 permit more impactful operations such as system suspend to local
170 * Memory limits that apply while the unit is activating are now
171 supported. Previously IO and CPU settings were already supported via
172 StartupCPUWeight= and similar. The same logic has been added for the
173 various manager and unit memory settings (DefaultStartupMemoryLow=,
174 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
175 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
177 * The service manager gained support for enqueuing POSIX signals to
178 services that carry an additional integer value, exposing the
179 sigqueue() system call. This is accessible via new D-Bus calls
180 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
181 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
182 via the new --kill-value= option.
184 * systemctl gained a new "list-paths" verb, which shows all currently
185 active .path units, similarly to how "systemctl list-timers" shows
186 active timers, and "systemctl list-sockets" shows active sockets.
188 * systemctl gained a new --when= switch which is honoured by the various
189 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
190 scheduling these operations by time, similar in fashion to how this
191 has been supported by SysV shutdown.
193 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
194 supports the new PR_SET_MDWE prctl() call, it is used instead of the
195 seccomp()-based system call filter to achieve the same effect.
197 * A new set of kernel command line options is now understood:
198 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
199 systemd.tty.columns.<name>= allow configuring the TTY type and
200 dimensions for the tty specified via <name>. When systemd invokes a
201 service on a tty (via TTYName=) it will look for these and configure
202 the TTY accordingly. This is particularly useful in VM environments
203 to propagate host terminal settings into the appropriate TTYs of the
206 * A new RootEphemeral= setting is now understood in service units. It
207 takes a boolean argument. If enabled for services that use RootImage=
208 or RootDirectory= an ephemeral copy of the disk image or directory
209 tree is made when the service is started. It is removed automatically
210 when the service is stopped. That ephemeral copy is made using
211 btrfs/xfs reflinks or btrfs snapshots, if available.
213 * The service activation logic gained new settings RestartSteps= and
214 RestartMaxDelaySec= which allow exponentially-growing restart
215 intervals for Restart=.
217 * The service activation logic gained a new setting RestartMode= which
218 can be set to 'direct' to skip the inactive/failed states when
219 restarting, so that dependent units are not notified until the service
220 converges to a final (successful or failed) state. For example, this
221 means that OnSuccess=/OnFailure= units will not be triggered until the
222 service state has converged.
224 * PID 1 will now automatically load the virtio_console kernel module
225 during early initialization if running in a suitable VM. This is done
226 so that early-boot logging can be written to the console if available.
228 * Similarly, virtio-vsock support is loaded early in suitable VM
229 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
230 to the VMM if configured, thus loading this early is beneficial.
232 * A new verb "fdstore" has been added to systemd-analyze to show the
233 current contents of the file descriptor store of a unit. This is
234 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
237 * The service manager will now set a new $FDSTORE environment variable
238 when invoking processes for services that have the file descriptor
241 * A new service option FileDescriptorStorePreserve= has been added that
242 allows tuning the life-cycle of the per-service file descriptor
243 store. If set to "yes", the entries in the fd store are retained even
244 after the service has been fully stopped.
246 * The "systemctl clean" command may now be used to clear the fdstore of
249 * Unit *.preset files gained a new directive "ignore", in addition to
250 the existing "enable" and "disable". As the name suggests, matching
251 units are left unchanged, i.e. neither enabled nor disabled.
253 * Service units gained a new setting DelegateSubgroup=. It takes the
254 name of a sub-cgroup to place any processes the service manager forks
255 off in. Previously, the service manager would place all service
256 processes directly in the top-level cgroup it created for the
257 service. This usually meant that main process in a service with
258 delegation enabled would first have to create a subgroup and move
259 itself down into it, in order to not conflict with the "no processes
260 in inner cgroups" rule of cgroup v2. With this option, this step is
261 now handled by PID 1.
263 * The service manager will now look for .upholds/ directories,
264 similarly to the existing support for .wants/ and .requires/
265 directories. Symlinks in this directory result in Upholds=
268 The [Install] section of unit files gained support for a new
269 UpheldBy= directive to generate .upholds/ symlinks automatically when
272 * The service manager now supports a new kernel command line option
273 systemd.default_device_timeout_sec=, which may be used to override
274 the default timeout for .device units.
276 * A new "soft-reboot" mechanism has been added to the service manager.
277 A "soft reboot" is similar to a regular reboot, except that it
278 affects userspace only: the service manager shuts down any running
279 services and other units, then optionally switches into a new root
280 file system (mounted to /run/nextroot/), and then passes control to a
281 systemd instance in the new file system which then starts the system
282 up again. The kernel is not rebooted and neither is the hardware,
283 firmware or boot loader. This provides a fast, lightweight mechanism
284 to quickly reset or update userspace, without the latency that a full
285 system reset involves. Moreover, open file descriptors may be passed
286 across the soft reboot into the new system where they will be passed
287 back to the originating services. This allows pinning resources
288 across the reboot, thus minimizing grey-out time further. This new
289 reboot mechanism is accessible via the new "systemctl soft-reboot"
292 * Services using RootDirectory= or RootImage= will now have read-only
293 access to a copy of the host's os-release file under
294 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
295 This was already the case for Portable Services, and the feature has
296 now been extended to all services that do not run off the host's
299 * A new service setting MemoryKSM= has been added to enable kernel
300 same-page merging individually for services.
302 * A new service setting ImportCredentials= has been added that augments
303 LoadCredential= and LoadCredentialEncrypted= and searches for
304 credentials to import from the system, and supports globbing.
306 * A new job mode "restart-dependencies" has been added to the service
307 manager (exposed via systemctl --job-mode=). It is only valid when
308 used with "start" jobs, and has the effect that the "start" job will
309 be propagated as "restart" jobs to currently running units that have
310 a BindsTo= or Requires= dependency on the started unit.
312 * A new verb "whoami" has been added to "systemctl" which determines as
313 part of which unit the command is being invoked. It writes the unit
314 name to standard output. If one or more PIDs are specified reports
315 the unit names the processes referenced by the PIDs belong to.
317 * The system and service credential logic has been improved: there's
318 now a clearly defined place where system provisioning tools running
319 in the initrd can place credentials that will be imported into the
320 system's set of credentials during the initrd → host transition: the
321 /run/credentials/@initrd/ directory. Once the credentials placed
322 there are imported into the system credential set they are deleted
323 from this directory, and the directory itself is deleted afterwards
326 * A new kernel command line option systemd.set_credential_binary= has
327 been added, that is similar to the pre-existing
328 systemd.set_credential= but accepts arbitrary binary credential data,
329 encoded in Base64. Note that the kernel command line is not a
330 recommend way to transfer credentials into a system, since it is
331 world-readable from userspace.
333 * The default machine ID to use may now be configured via the
334 system.machine_id system credential. It will only be used if no
335 machine ID was set yet on the host.
337 * On Linux kernel 6.4 and newer system and service credentials will now
338 be placed in a tmpfs instance that has the "noswap" mount option
339 set. Previously, a "ramfs" instance was used. By switching to tmpfs
340 ACL support and overall size limits can now be enforced, without
341 compromising on security, as the memory is never paged out either
344 * The service manager now can detect when it is running in a
345 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
346 accepted by ConditionSecurity= for units that want to conditionalize
347 themselves on this. systemd-detect-virt gained new 'cvm' and
348 '--list-cvm' switches to respectively perform the detection or list
349 all known flavours of confidential VM, depending on the vendor. The
350 manager will publish a 'ConfidentialVirtualization' D-Bus property,
351 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
352 variable for unit generators. Finally, udev rules can match on a new
353 'cvm' key that will be set when in a confidential VM.
354 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
355 strings and QEMU's fw_cfg protocol will not be used to import
356 credentials and kernel command line parameters by the system manager,
357 systemd-boot and systemd-stub, because the hypervisor is considered
358 untrusted in this particular setting.
362 * The sd-journal API gained a new call sd_journal_get_seqnum() to
363 retrieve the current log record's sequence number and sequence number
364 ID, which allows applications to order records the same way as
365 journal does internally. The sequence number is now also exported in
366 the JSON and "export" output of the journal.
368 * journalctl gained a new switch --truncate-newline. If specified
369 multi-line log records will be truncated at the first newline,
370 i.e. only the first line of each log message will be shown.
372 * systemd-journal-upload gained support for --namespace=, similar to
373 the switch of the same name of journalctl.
377 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
378 may be used to exclude certain files from the effect of CopyFiles=.
380 * systemd-repart's Verity support now implements the Minimize= setting
381 to minimize the size of the resulting partition.
383 * systemd-repart gained a new --offline= switch, which may be used to
384 control whether images shall be built "online" or "offline",
385 i.e. whether to make use of kernel facilities such as loopback block
386 devices and device mapper or not.
388 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
389 partition with some files, it will now default to VFAT rather than
392 * systemd-repart gained a new --architecture= switch. If specified, the
393 per-architecture GPT partition types (i.e. the root and /usr/
394 partitions) configured in the partition drop-in files are
395 automatically adjusted to match the specified CPU architecture, in
396 order to simplify cross-architecture DDI building.
398 * systemd-repart will now default to a minimum size of 300MB for XFS
399 filesystems if no size parameter is specified. This matches what the
400 XFS tools (xfsprogs) can support.
402 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
404 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
405 Instead, pyelftools is now needed, and it will be used to perform the
406 ELF -> PE relocations at build time.
408 * bootctl gained a new switch --print-root-device/-R that prints the
409 block device the root file system is backed by. If specified twice,
410 it returns the whole disk block device (as opposed to partition block
411 device) the root file system is on. It's useful for invocations such
412 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
415 * systemd-stub will now look for the SMBIOS Type 1 field
416 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
417 kernel command line it invokes. This is useful for VMMs such as qemu
418 to pass additional kernel command lines into the system even when
419 booting via full UEFI. The contents of the field are measured into
422 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
423 value "auto". With this value, a kernel will be automatically
424 analyzed, and if it qualifies as UKI, it will be installed as if the
425 setting was to set to "uki", otherwise as "bls".
427 * systemd-stub can now optionally load UEFI PE "add-on" images that may
428 contain additional kernel command line information. These "add-ons"
429 superficially look like a regular UEFI executable, and are expected
430 to be signed via SecureBoot/shim. However, they do not actually
431 contain code, but instead a subset of the PE sections that UKIs
432 support. They are supposed to provide a way to extend UKIs with
433 additional resources in a secure and authenticated way. Currently,
434 only the .cmdline PE section may be used in add-ons, in which case
435 any specified string is appended to the command line embedded into
436 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
437 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
438 the future we expect other sections to be made extensible like this as
441 * ukify has been updated to allow building these UEFI PE "add-on"
442 images, using the new 'addon<EFI-ARCH>.efi.stub'.
444 * ukify gained a new "genkey" verb for generating a set of of key pairs
445 to sign UKIs and their PCR data with.
447 * ukify now accepts SBAT information to place in the .sbat PE section
448 of UKIs and addons. If a UKI is built the SBAT information from the
449 inner kernel is merged with any SBAT information associated with
450 systemd-stub and the SBAT data specified on the ukify command line.
452 * The kernel-install script has been rewritten in C, and reuses much of
453 the infrastructure of existing tools such as bootctl. It also gained
454 --esp-path= and --boot-path= options to override the path to the ESP,
455 and the $BOOT partition. Options --make-entry-directory= and
456 --entry-token= have been added as well, similar to bootctl's options
459 * A new kernel-install plugin 60-ukify has been added which will
460 combine kernel/initrd locally into a UKI and optionally sign them
461 with a local key. This may be used to switch to UKI mode even on
462 systems where a local kernel or initrd is used. (Typically UKIs are
463 built and signed by the vendor.)
465 * The ukify tool now supports "pesign" in addition to the pre-existing
466 "sbsign" for signing UKIs.
468 * systemd-measure and systemd-stub now look for the .uname PE section
469 that should contain the kernel's "uname -r" string.
471 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
472 "offline", i.e. without access to a TPM (physical or
475 Memory Pressure & Control:
477 * The sd-event API gained new calls sd_event_add_memory_pressure(),
478 sd_event_source_set_memory_pressure_type(),
479 sd_event_source_set_memory_pressure_period() to create and configure
480 an event source that is called whenever the OS signals memory
481 pressure. Another call sd_event_trim_memory() is provided that
482 compacts the process' memory use by releasing allocated but unused
483 malloc() memory back to the kernel. Services can also provide their
484 own custom callback to do memory trimming. This should improve system
485 behaviour under memory pressure, as on Linux traditionally provided
486 no mechanism to return process memory back to the kernel if the
487 kernel was under memory pressure. This makes use of the kernel's PSI
488 interface. Most long-running services in systemd have been hooked up
489 with this, and in particular systems with low memory should benefit
492 * Service units gained new settings MemoryPressureWatch= and
493 MemoryPressureThresholdSec= to configure the PSI memory pressure
494 logic individually. If these options are used, the
495 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
496 variables will be set for the invoked processes to inform them about
497 the requested memory pressure behaviour. (This is used by the
498 aforementioned sd-events API additions, if set.)
500 * systemd-analyze gained a new "malloc" verb that shows the output
501 generated by glibc's malloc_info() on services that support it. Right
502 now, only the service manager has been updated accordingly. This
503 call requires privileges.
505 User & Session Management:
507 * The sd-login API gained a new call sd_session_get_username() to
508 return the user name of the owner of a login session. It also gained
509 a new call sd_session_get_start_time() to retrieve the time the login
510 session started. A new call sd_session_get_leader() has been added to
511 return the PID of the "leader" process of a session. A new call
512 sd_uid_get_login_time() returns the time since the specified user has
513 most recently been continuously logged in with at least one session.
515 * JSON user records gained a new set of fields capabilityAmbientSet and
516 capabilityBoundingSet which contain a list of POSIX capabilities to
517 set for the logged in users in the ambient and bounding sets,
518 respectively. homectl gained the ability to configure these two sets
519 for users via --capability-bounding-set=/--capability-ambient-set=.
521 * pam_systemd learnt two new module options
522 default-capability-bounding-set= and default-capability-ambient-set=,
523 which configure the default bounding sets for users as they are
524 logging in, if the JSON user record doesn't specify this explicitly
525 (see above). The built-in default for the ambient set now contains
526 the CAP_WAKE_ALARM, thus allowing regular users who may log in
527 locally to resume from a system suspend via a timer.
529 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
530 call to update the TTY of a session after it has been allocated. This
531 is useful for SSH sessions which are typically allocated first, and
532 for which a TTY is added later.
534 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
535 combines the various other sd_pid_notify() flavours into one: takes a
536 format string, an overriding PID, and a set of file descriptors to
537 send. It also gained a new call sd_pid_notify_barrier() call which is
538 equivalent to sd_notify_barrier() but allows the originating PID to
541 * "loginctl list-users" and "loginctl list-sessions" will now show the
542 state of each logged in user/session in their tabular output. It will
543 also show the current idle state of sessions.
547 * systemd-dissect will now show the intended CPU architecture of an
550 * systemd-dissect will now install itself as mount helper for the "ddi"
551 pseudo-file system type. This means you may now mount DDIs directly
552 via /bin/mount or /etc/fstab, making full use of embedded Verity
553 information and all other DDI features.
555 Example: mount -t ddi myimage.raw /some/where
557 * The systemd-dissect tool gained the new switches --attach/--detach to
558 attach/detach a DDI to a loopback block device without mounting it.
559 It will automatically derive the right sector size from the image
560 and set up Verity and similar, but not mount the file systems in it.
562 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
563 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
564 implied. Given that these file systems are typically untrusted, this
565 should make mounting them automatically have less of a security
568 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
569 systemd-tmpfiles, …) now understand a new switch --image-policy= which
570 takes a string encoding image dissection policy. With this mechanism
571 automatic discovery and use of specific partition types and the
572 cryptographic requirements on the partitions (Verity, LUKS, …) can be
573 restricted, permitting better control of the exposed attack surfaces
574 when mounting disk images. systemd-gpt-auto-generator will honour such
575 an image policy too, configurable via the systemd.image_policy= kernel
576 command line option. Unit files gained the RootImagePolicy=,
577 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
578 disk images a service runs off.
580 * systemd-analyze gained a new verb "image-policy" to validate and
581 parse image policy strings.
583 * systemd-dissect gained support for a new --validate switch to
584 superficially validate DDI structure, and check whether a specific
585 image policy allows the DDI.
587 * systemd-dissect gained support for a new --mtree-hash switch to
588 optionally disable calculating mtree hashes, which can be slow on
591 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
592 are now able to operate on directories too, other than images.
596 * networkd's GENEVE support as gained a new .network option
597 InheritInnerProtocol=.
599 * The [Tunnel] section in .netdev files has gained a new setting
600 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
602 * A new global IPv6PrivacyExtensions= setting has been added that
603 selects the default value of the per-network setting of the same
606 * The predictable network interface naming logic was extended to
607 include SR-IOV-R "representor" information in network interface
608 names. Unfortunately, this feature was not enabled by default and can
609 only be enabled at compilation time by setting
610 -Ddefault-net-naming-scheme=v254.
612 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
613 the RFC8910 captive portal option.
617 * udevadm gained the new "verify" verb for validating udev rules files
620 * udev gained a new tool "iocost" that can be used to configure QoS IO
621 cost data based on hwdb information onto suitable block devices. Also
622 see https://github.com/iocost-benchmark/iocost-benchmarks.
624 TPM2 Support + Disk Encryption & Authentication:
626 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
627 ("Storage Root Key") as first step in the TPM2, and then use that
628 for binding FDE to, if TPM2 support is used. This matches
629 recommendations of TCG (see
630 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
632 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
633 understand textual identifiers for these PCRs.
635 * systemd-veritysetup + /etc/veritytab gained support for a series of
636 new options: hash-offset=, superblock=, format=, data-block-size=,
637 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
638 fec-offset=, fec-roots= to configure various aspects of a Verity
641 * systemd-cryptsetup + /etc/crypttab gained support for a new
642 veracrypt-pim= option for setting the Personal Iteration Multiplier
643 of veracrypt volumes.
645 * systemd-integritysetup + /etc/integritytab gained support for a new
646 mode= setting for controlling the dm-integrity mode (journal, bitmap,
647 direct) for the volume.
649 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
650 registers, their symbolic names and current values.
654 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
655 access right is specified this is equivalent to "x" but only if the
656 inode in question already has the executable bit set for at least
657 some user/group. Otherwise the "x" bit will be turned off.
659 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
660 with C+ will result in a "merge" copy, i.e. all files of the source
661 tree are copied into the target tree, even if that tree already
662 exists, resulting in a combined tree of files already present in the
663 target tree and those copied in.
665 * systemd-tmpfiles gained a new --graceful switch. If specified lines
666 with unknown users/groups will silently be skipped.
670 * systemd-notify gained two new options --fd= and --fdname= for sending
671 arbitrary file descriptors to the service manager (while specifying an
672 explicit name for it).
674 * systemd-notify gained a new --exec switch, which makes it execute the
675 specified command line after sending the requested messages. This is
676 useful for sending out READY=1 first, and then continuing invocation
677 without changing process ID, so that the tool can be nicely used
678 within an ExecStart= line of a unit file that uses Type=notify.
680 sd-event + sd-bus APIs:
682 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
683 which may be used to explicitly end a rate-limit state an event
684 source might be in, resetting all rate limiting counters.
686 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
687 sockets, it will now encode the "description" set via
688 sd_bus_set_description() into the source socket address. It will also
689 look for this information when accepting a connection. This is useful
690 to track individual D-Bus connections on a D-Bus broker for debug
695 * systemd-resolved gained a new resolved.conf setting
696 StateRetentionSec= which may be used to retain cached DNS records
697 even after their nominal TTL, and use them in case upstream DNS
698 servers cannot be reached. This can be used to make name resolution
699 more resilient in case of network problems.
701 * resolvectl gained a new verb "show-cache" to show the current cache
702 contents of systemd-resolved. This verb communicates with the
703 systemd-resolved daemon and requires privileges.
707 * Meson >= 0.60.0 is now required to build systemd.
709 * The default keymap to apply may now be chosen at build-time via the
710 new -Ddefault-keymap= meson option.
712 * Most of systemd's long-running services now have a generic handler of
713 the SIGRTMIN+18 signal handler which executes various operations
714 depending on the sigqueue() parameter sent along. For example, values
715 0x100…0x107 allow changing the maximum log level of such
716 services. 0x200…0x203 allow changing the log target of such
717 services. 0x300 make the services trim their memory similarly to the
718 automatic PSI-triggered action, see above. 0x301 make the services
719 output their malloc_info() data to the logs.
721 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
722 files, inspired by systemctl's verbs of the same name which edit unit
723 files. Similarly, networkctl gained the same verbs for editing
724 .network, .netdev, .link files.
726 * A new syscall filter group "@sandbox" has been added that contains
727 syscalls for sandboxing system calls such as those for seccomp and
730 * New documentation has been added:
732 https://systemd.io/COREDUMP
733 https://systemd.io/MEMORY_PRESSURE
736 * systemd-firstboot gained a new --reset option. If specified, the
737 settings in /etc/ it knows how to initialize are reset.
739 * systemd-sysext is now a multi-call binary and is also installed under
740 the systemd-confext alias name (via a symlink). When invoked that way
741 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
742 powerful, atomic, secure configuration management of sorts, that
743 locally can merge configuration from multiple confext configuration
744 images into a single immutable tree.
746 * The --network-macvlan=, --network-ipvlan=, --network-interface=
747 switches of systemd-nspawn may now optionally take the intended
748 network interface inside the container.
750 * All our programs will now send an sd_notify() message with their exit
751 status in the EXIT_STATUS= field when exiting, using the usual
752 protocol, including PID 1. This is useful for VMMs and container
753 managers to collect an exit status from a system as it shuts down, as
754 set via "systemctl exit …". This is particularly useful in test cases
755 and similar, as invocations via a VM can now nicely propagate an exit
756 status to the host, similar to local processes.
758 * systemd-run gained a new switch --expand-environment=no to disable
759 server-side environment variable expansion in specified command
760 lines. Expansion defaults to enabled for all execution types except
761 --scope, where it defaults to off (and prints a warning) for backward
762 compatibility reasons. --scope will be flipped to enabled by default
763 too in a future release. If you are using --scope and passing a '$'
764 character in the payload you should start explicitly using
765 --expand-environment=yes/no according to the use case.
767 * The systemd-system-update-generator has been updated to also look for
768 the special flag file /etc/system-update in addition to the existing
769 support for /system-update to decide whether to enter system update
772 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
773 mount options by default.
775 * systemd-fstab-generator now understands two new kernel command line
776 options systemd.mount-extra= and systemd.swap-extra=, which configure
777 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
778 will be ran on these block devices, like it already happens for
779 'root='. It also now supports the new fstab.extra and
780 fstab.extra.initrd credentials that may contain additional /etc/fstab
781 lines to apply at boot.
783 * systemd-getty-generator now understands two new credentials
784 getty.ttys.container and getty.ttys.serial. These credentials may
785 contain a list of TTY devices – one per line – to instantiate
786 container-getty@.service and serial-getty@.service on.
788 * The getty/serial-getty/container-getty units now import the 'agetty.*'
789 and 'login.*' credentials, which are consumed by the 'login' and
790 'agetty' programs starting from util-linux v2.40.
792 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
793 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
794 which case the Path= setting is taken relative to the ESP or XBOOTLDR
795 partitions, rather than the system's root directory /. The relevant
796 directories are automatically discovered.
798 * The systemd-ac-power tool gained a new switch --low, which reports
799 whether the battery charge is considered "low", similar to how the
800 s2h suspend logic checks this state to decide whether to enter system
801 suspend or hibernation.
803 * The /etc/os-release file can now have two new optional fields
804 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
807 * When the system hibernates, information about the device and offset
808 used is now written to a non-volatile EFI variable. On next boot the
809 system will attempt to resume from the location indicated in this EFI
810 variable. This should make hibernation a lot more robust, while
811 requiring no manual configuration of the resume location.
813 * The $XDG_STATE_HOME environment variable (added in more recent
814 versions of the XDG basedir specification) is now honoured to
815 implement the StateDirectory= setting in user services.
817 * A new component "systemd-battery-check" has been added. It may run
818 during early boot (usually in the initrd), and checks the battery
819 charge level of the system. In case the charge level is very low the
820 user is notified (graphically via Plymouth – if available – as well
821 as in text form on the console), and the system is turned off after a
822 10s delay. The feature can be disabled by passing
823 systemd.battery-check=0 through the kernel command line.
825 * The 'passwdqc' library is now supported as an alternative to the
826 'pwquality' library and can be selected at build time.
828 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
829 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
830 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
831 Arian van Putten, Arthur Shau, A S Alam,
832 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
833 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
834 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
835 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
836 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
837 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
838 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
839 David Schroeder, David Tardon, dependabot[bot],
840 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
841 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
842 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
843 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
844 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
845 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
846 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
847 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
848 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
849 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
850 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
851 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
852 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
853 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
854 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
855 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
856 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
857 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
858 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
859 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
860 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
861 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
862 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
863 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
864 saikat0511, Samanta Navarro, Sam James, Sam Morris,
865 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
866 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
867 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
868 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
869 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
870 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
871 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
872 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
873 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
874 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
875 Дамјан Георгиевски, наб
877 — Edinburgh, 2023-07-28
881 Announcements of Future Feature Removals and Incompatible Changes:
883 * We intend to remove cgroup v1 support from systemd release after the
884 end of 2023. If you run services that make explicit use of cgroup v1
885 features (i.e. the "legacy hierarchy" with separate hierarchies for
886 each controller), please implement compatibility with cgroup v2 (i.e.
887 the "unified hierarchy") sooner rather than later. Most of Linux
888 userspace has been ported over already.
890 * We intend to remove support for split-usr (/usr mounted separately
891 during boot) and unmerged-usr (parallel directories /bin and
892 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
893 half of 2023, in the first release that falls into that time window.
894 For more details, see:
895 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
897 * We intend to change behaviour w.r.t. units of the per-user service
898 manager and sandboxing options, so that they work without having to
899 manually enable PrivateUsers= as well, which is not required for
900 system units. To make this work, we will implicitly enable user
901 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
902 user unit. The drawback is that system users will no longer be visible
903 (and appear as 'nobody') to the user unit when a sandboxing option is
904 enabled. By definition a sandboxed user unit should run with reduced
905 privileges, so impact should be small. This will remove a great source
906 of confusion that has been reported by users over the years, due to
907 how these options require an extra setting to be manually enabled when
908 used in the per-user service manager, as opposed as to the system
909 service manager. We plan to enable this change in the next release
910 later this year. For more details, see:
911 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
913 Deprecations and incompatible changes:
915 * systemctl will now warn when invoked without /proc/ mounted
916 (e.g. when invoked after chroot() into an directory tree without the
917 API mount points like /proc/ being set up.) Operation in such an
918 environment is not fully supported.
920 * The return value of 'systemctl is-active|is-enabled|is-failed' for
921 unknown units is changed: previously 1 or 3 were returned, but now 4
922 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
924 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
925 systemd-hwdb (added in 2014) should be used instead.
927 * 'bootctl --json' now outputs a single JSON array, instead of a stream
928 of newline-separated JSON objects.
930 * Udev rules in 60-evdev.rules have been changed to load hwdb
931 properties for all modalias patterns. Previously only the first
932 matching pattern was used. This could change what properties are
933 assigned if the user has more and less specific patterns that could
934 match the same device, but it is expected that the change will have
935 no effect for most users.
937 * systemd-networkd-wait-online exits successfully when all interfaces
938 are ready or unmanaged. Previously, if neither '--any' nor
939 '--interface=' options were used, at least one interface had to be in
940 configured state. This change allows the case where systemd-networkd
941 is enabled, but no interfaces are configured, to be handled
942 gracefully. It may occur in particular when a different network
943 manager is also enabled and used.
945 * Some compatibility helpers were dropped: EmergencyAction= in the user
946 manager, as well as measuring kernel command line into PCR 8 in
947 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
950 * The '-Dupdate-helper-user-timeout=' build-time option has been
951 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
952 integer as parameter instead of a string.
954 * The DDI image dissection logic (which backs RootImage= in service
955 unit files, the --image= switch in various tools such as
956 systemd-nspawn, as well as systemd-dissect) will now only mount file
957 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
958 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
959 variable. These file systems are fairly well supported and maintained
960 in current kernels, while others are usually more niche, exotic or
961 legacy and thus typically do not receive the same level of security
964 * The default per-link multicast DNS mode is changed to "yes"
965 (that was previously "no"). As the default global multicast DNS mode
966 has been "yes" (but can be changed by the build option), now the
967 multicast DNS is enabled on all links by default. You can disable the
968 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
969 or on an interface by calling "resolvectl mdns INTERFACE no".
973 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
974 (UKIs) has been added. This replaces functionality provided by
975 'dracut --uefi' and extends it with automatic calculation of PE file
976 offsets, insertion of signed PCR policies generated by
977 systemd-measure, support for initrd concatenation, signing of the
978 embedded Linux image and the combined image with sbsign, and
979 heuristics to autodetect the kernel uname and verify the splash
982 Changes in systemd and units:
984 * A new service type Type=notify-reload is defined. When such a unit is
985 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
986 service process. The manager will then wait until it receives a
987 "RELOADING=1" followed by a "READY=1" notification from the unit as
988 response (via sd_notify()). Otherwise, this type is the same as
989 Type=notify. A new setting ReloadSignal= may be used to change the
990 signal to send from the default of SIGHUP.
992 user@.service, systemd-networkd.service, systemd-udevd.service, and
993 systemd-logind have been updated to this type.
995 * Initrd environments which are not on a pure memory file system (e.g.
996 overlayfs combination as opposed to tmpfs) are now supported. With
997 this change, during the initrd → host transition ("switch root")
998 systemd will erase all files of the initrd only when the initrd is
999 backed by a memory file system such as tmpfs.
1001 * New per-unit MemoryZSwapMax= option has been added to configure
1002 memory.zswap.max cgroup properties (the maximum amount of zswap
1005 * A new LogFilterPatterns= option has been added for units. It may be
1006 used to specify accept/deny regular expressions for log messages
1007 generated by the unit, that shall be enforced by systemd-journald.
1008 Rejected messages are neither stored in the journal nor forwarded.
1009 This option may be used to suppress noisy or uninteresting messages
1012 * The manager has a new
1013 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
1014 query process ownership via a PIDFD, which is more resilient against
1015 PID recycling issues.
1017 * Scope units now support OOMPolicy=. Login session scopes default to
1018 OOMPolicy=continue, allowing login scopes to survive the OOM killer
1019 terminating some processes in the scope.
1021 * systemd-fstab-generator now supports x-systemd.makefs option for
1022 /sysroot/ (in the initrd).
1024 * The maximum rate at which daemon reloads are executed can now be
1025 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
1026 options. (Or the equivalent on the kernel command line:
1027 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
1028 addition, systemd now logs the originating unit and PID when a reload
1029 request is received over D-Bus.
1031 * When enabling a swap device systemd will now reinitialize the device
1032 when the page size of the swap space does not match the page size of
1033 the running kernel. Note that this requires the 'swapon' utility to
1034 provide the '--fixpgsz' option, as implemented by util-linux, and it
1035 is not supported by busybox at the time of writing.
1037 * systemd now executes generator programs in a mount namespace
1038 "sandbox" with most of the file system read-only and write access
1039 restricted to the output directories, and with a temporary /tmp/
1040 mount provided. This provides a safeguard against programming errors
1041 in the generators, but also fixes here-docs in shells, which
1042 previously didn't work in early boot when /tmp/ wasn't available
1043 yet. (This feature has no security implications, because the code is
1044 still privileged and can trivially exit the sandbox.)
1046 * The system manager will now parse a new "vmm.notify_socket"
1047 system credential, which may be supplied to a VM via SMBIOS. If
1048 found, the manager will send a "READY=1" notification on the
1049 specified socket after boot is complete. This allows readiness
1050 notification to be sent from a VM guest to the VM host over a VSOCK
1053 * The sample PAM configuration file for systemd-user@.service now
1054 includes a call to pam_namespace. This puts children of user@.service
1055 in the expected namespace. (Many distributions replace their file
1056 with something custom, so this change has limited effect.)
1058 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
1059 can be used to override the mount units burst late limit for
1060 parsing '/proc/self/mountinfo', which was introduced in v249.
1063 * Drop-ins for init.scope changing control group resource limits are
1064 now applied, while they were previously ignored.
1066 * New build-time configuration options '-Ddefault-timeout-sec=' and
1067 '-Ddefault-user-timeout-sec=' have been added, to let distributions
1068 choose the default timeout for starting/stopping/aborting system and
1069 user units respectively.
1071 * Service units gained a new setting OpenFile= which may be used to
1072 open arbitrary files in the file system (or connect to arbitrary
1073 AF_UNIX sockets in the file system), and pass the open file
1074 descriptor to the invoked process via the usual file descriptor
1075 passing protocol. This is useful to give unprivileged services access
1076 to select files which have restrictive access modes that would
1077 normally not allow this. It's also useful in case RootDirectory= or
1078 RootImage= is used to allow access to files from the host environment
1079 (which is after all not visible from the service if these two options
1084 * The new net naming scheme "v253" has been introduced. In the new
1085 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
1086 a PCI bus. This extends the coverage of predictable interface names
1087 in some embedded systems.
1089 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
1090 a more informative path on some embedded systems.
1092 * Partition block devices will now also get symlinks in
1093 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
1094 block device nodes via the kernel's "diskseq" value. Previously those
1095 symlinks were only created for the main block device.
1097 * A new operator '-=' is supported for SYMLINK variables. This allows
1098 symlinks to be unconfigured even if an earlier rule added them.
1100 * 'udevadm --trigger --settle' now also works for network devices
1101 that are being renamed.
1103 Changes in sd-boot, bootctl, and the Boot Loader Specification:
1105 * systemd-boot now passes its random seed directly to the kernel's RNG
1106 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
1107 means the RNG gets seeded very early in boot before userspace has
1110 * systemd-boot will pass a disk-backed random seed – even when secure
1111 boot is enabled – if it can additionally get a random seed from EFI
1112 itself (via EFI's RNG protocol), or a prior seed in
1113 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
1115 * systemd-boot-system-token.service was renamed to
1116 systemd-boot-random-seed.service and extended to always save a random
1117 seed to ESP on every boot when a compatible boot loader is used. This
1118 allows a refreshed random seed to be used in the boot loader.
1120 * systemd-boot handles various seed inputs using a domain- and
1121 field-separated hashing scheme.
1123 * systemd-boot's 'random-seed-mode' option has been removed. A system
1124 token is now always required to be present for random seeds to be
1127 * systemd-boot now supports being loaded from other locations than the
1128 ESP, for example for direct kernel boot under QEMU or when embedded
1131 * systemd-boot now parses SMBIOS information to detect
1132 virtualization. This information is used to skip some warnings which
1133 are not useful in a VM and to conditionalize other aspects of
1136 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
1137 Secure Boot automated certificate enrollment from the ESP only if it
1138 is considered 'safe' to do so. At the moment 'safe' means running in
1141 * systemd-stub now processes random seeds in the same way as
1142 systemd-boot already does, in case a unified kernel image is being
1143 used from a different bootloader than systemd-boot, or without any
1146 * bootctl will now generate a system token on all EFI systems, even
1147 virtualized ones, and is activated in the case that the system token
1148 is missing from either sd-boot and sd-stub booted systems.
1150 * bootctl now implements two new verbs: 'kernel-identify' prints the
1151 type of a kernel image file, and 'kernel-inspect' provides
1152 information about the embedded command line and kernel version of
1155 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
1156 as for kernel-install.
1158 * The JSON output of "bootctl list" will now contain two more fields:
1159 isDefault and isSelected are boolean fields set to true on the
1160 default and currently booted boot menu entries.
1162 * bootctl gained a new verb "unlink" for removing a boot loader entry
1163 type #1 file from disk in a safe and robust way.
1165 * bootctl also gained a new verb "cleanup" that automatically removes
1166 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
1167 is not referenced anymore by any installed Type #1 boot loader
1168 specification entry. This is particularly useful in environments where
1169 a large number of entries reference the same or partly the same
1170 resources (for example, for snapshot-based setups).
1172 Changes in kernel-install:
1174 * A new "installation layout" can be configured as layout=uki. With
1175 this setting, a Boot Loader Specification Type#1 entry will not be
1176 created. Instead, a new kernel-install plugin 90-uki-copy.install
1177 will copy any .efi files from the staging area into the boot
1178 partition. A plugin to generate the UKI .efi file must be provided
1181 Changes in systemctl:
1183 * 'systemctl reboot' has dropped support for accepting a positional
1184 argument as the argument to the reboot(2) syscall. Please use the
1185 --reboot-argument= option instead.
1187 * 'systemctl disable' will now warn when called on units without
1188 install information. A new --no-warn option has been added that
1189 silences this warning.
1191 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
1192 of the drop-in to edit. (Previously, 'override.conf' was always
1195 * 'systemctl list-dependencies' now respects --type= and --state=.
1197 * 'systemctl kexec' now supports XEN VMM environments.
1199 * 'systemctl edit' will now tell the invoked editor to jump into the
1200 first line with actual unit file data, skipping over synthesized
1203 Changes in systemd-networkd and related tools:
1205 * The [DHCPv4] section in .network file gained new SocketPriority=
1206 setting that assigns the Linux socket priority used by the DHCPv4 raw
1207 socket. This may be used in conjunction with the
1208 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
1209 desired ethernet 802.1Q frame priority for DHCPv4 initial
1210 packets. This cannot be achieved with netfilter mangle tables because
1211 of the raw socket bypass.
1213 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
1214 new QuickAck= boolean setting that enables the TCP quick ACK mode for
1215 the routes configured by the acquired DHCPv4 lease or received router
1216 advertisements (RAs).
1218 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
1219 routes) now accepts three values, for high, medium, and low preference
1220 of the router (which can be set with the RouterPreference=) setting.
1222 * systemd-networkd-wait-online now supports matching via alternative
1225 * The [DHCPv6] section in .network file gained new SendRelease=
1226 setting which enables the DHCPv6 client to send release when
1227 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
1228 It is enabled by default.
1230 * If the Address= setting in [Network] or [Address] sections in .network
1231 specified without its prefix length, then now systemd-networkd assumes
1232 /32 for IPv4 or /128 for IPv6 addresses.
1234 * networkctl shows network and link file dropins in status output.
1236 Changes in systemd-dissect:
1238 * systemd-dissect gained a new option --list, to print the paths of
1239 all files and directories in a DDI.
1241 * systemd-dissect gained a new option --mtree, to generate a file
1242 manifest compatible with BSD mtree(5) of a DDI
1244 * systemd-dissect gained a new option --with, to execute a command with
1245 the specified DDI temporarily mounted and used as working
1246 directory. This is for example useful to convert a DDI to "tar"
1247 simply by running it within a "systemd-dissect --with" invocation.
1249 * systemd-dissect gained a new option --discover, to search for
1250 Discoverable Disk Images (DDIs) in well-known directories of the
1251 system. This will list machine, portable service and system extension
1254 * systemd-dissect now understands 2nd stage initrd images stored as a
1255 Discoverable Disk Image (DDI).
1257 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
1258 disk UUID stored in the GPT header) among the other data it can show.
1260 * systemd-dissect gained a new --in-memory switch to operate on an
1261 in-memory copy of the specified DDI file. This is useful to access a
1262 DDI with write access without persisting any changes. It's also
1263 useful for accessing a DDI without keeping the originating file
1266 * The DDI dissection logic will now automatically detect the intended
1267 sector size of disk images stored in files, based on the GPT
1268 partition table arrangement. Loopback block devices for such DDIs
1269 will then be configured automatically for the right sector size. This
1270 is useful to make dealing with modern 4K sector size DDIs fully
1271 automatic. The systemd-dissect tool will now show the detected sector
1272 size among the other DDI information in its output.
1274 Changes in systemd-repart:
1276 * systemd-repart gained new options --include-partitions= and
1277 --exclude-partitions= to filter operation on partitions by type UUID.
1278 This allows systemd-repart to be used to build images in which the
1279 type of one partition is set based on the contents of another
1280 partition (for example when the boot partition shall include a verity
1281 hash of the root partition).
1283 * systemd-repart also gained a --defer-partitions= option that is
1284 similar to --exclude-partitions=, but the size of the partition is
1285 still taken into account when sizing partitions, but without
1288 * systemd-repart gained a new --sector-size= option to specify what
1289 sector size should be used when an image is created.
1291 * systemd-repart now supports generating erofs file systems via
1292 CopyFiles= (a read-only file system similar to squashfs).
1294 * The Minimize= option was extended to accept "best" (which means the
1295 most minimal image possible, but may require multiple attempts) and
1296 "guess" (which means a reasonably small image).
1298 * The systemd-growfs binary now comes with a regular unit file template
1299 systemd-growfs@.service which can be instantiated directly for any
1300 desired file system. (Previously, the unit was generated dynamically
1301 by various generators, but no regular unit file template was
1304 Changes in journal tools:
1306 * Various systemd tools will append extra fields to log messages when
1307 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
1308 this includes information about D-Bus messages when sd-bus is used,
1309 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
1310 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
1311 Details of what is logged and when are subject to change.
1313 * The systemd-journald-audit.socket can now be disabled via the usual
1314 "systemctl disable" mechanism to stop collection of audit
1315 messages. Please note that it is not enabled statically anymore and
1316 must be handled by the preset/enablement logic in package
1317 installation scripts.
1319 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
1320 be used to curtail disk use by systemd-journal-remote. This is
1321 similar to the options supported by systemd-journald.
1323 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
1326 * When enrolling new keys systemd-cryptenroll now supports unlocking
1327 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
1328 password was strictly required to be specified.
1330 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
1331 (except for tokens with user verification, UV) to identify tokens
1332 before authentication. Multiple FIDO2 tokens can now be enrolled at
1333 the same time, and systemd-cryptsetup will automatically select one
1334 that corresponds to one of the available LUKS key slots.
1336 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
1337 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
1338 bank and number into which the volume key should be measured. This is
1339 automatically enabled for the encrypted root volume discovered and
1340 activated by systemd-gpt-auto-generator.
1342 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
1343 "noexec,nosuid,nodev".
1345 * systemd-gpt-auto-generator will now honour the rootfstype= and
1346 rootflags= kernel command line switches for root file systems it
1347 discovers, to match behaviour in case an explicit root fs is
1348 specified via root=.
1350 * systemd-pcrphase gained new options --machine-id and --file-system=
1351 to measure the machine-id and mount point information into PCR 15. New
1352 service unit files systemd-pcrmachine.service and
1353 systemd-pcrfs@.service have been added that invoke the tool with
1354 these switches during early boot.
1356 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
1357 with a success exit code even if no TPM device is detected.
1359 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
1360 making it harder to brute-force.
1362 Changes in other tools:
1364 * systemd-homed gained support for luksPbkdfForceIterations (the
1365 intended number of iterations for the PBKDF operation on LUKS).
1367 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
1368 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
1369 may now be used to specify additional arguments for mkfs when
1370 systemd-homed formats a file system.
1372 * systemd-hostnamed now exports the contents of
1373 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
1374 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
1375 unprivileged code to access those values.
1377 systemd-hostnamed also exports the SUPPORT_END= field from
1378 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
1379 this to show the status of the installed system.
1381 * systemd-measure gained an --append= option to sign multiple phase
1382 paths with different signing keys. This allows secrets to be
1383 accessible only in certain parts of the boot sequence. Note that
1384 'ukify' provides similar functionality in a more accessible form.
1386 * systemd-timesyncd will now write a structured log message with
1387 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
1388 on a on-disk timestamp, similarly to what it did when reaching
1389 synchronization via NTP.
1391 * systemd-timesyncd will now update the on-disk timestamp file on each
1392 boot at least once, making it more likely that the system time
1393 increases in subsequent boots.
1395 * systemd-vconsole-setup gained support for system/service credentials:
1396 vconsole.keymap/vconsole.keymap_toggle and
1397 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
1398 the similarly-named options in vconsole.conf.
1400 * systemd-localed will now save the XKB keyboard configuration to
1401 /etc/vconsole.conf, and also read it from there with a higher
1402 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
1403 file. Previously, this information was stored in the former file in
1404 converted form, and only in latter file in the original form. Tools
1405 which want to access keyboard configuration can now do so from a
1408 * systemd-resolved gained support for configuring the nameservers and
1409 search domains via kernel command line (nameserver=, domain=) and
1410 credentials (network.dns, network.search_domains).
1412 * systemd-resolved will now synthesize host names for the DNS stub
1413 addresses it supports. Specifically when "_localdnsstub" is resolved,
1414 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
1415 127.0.0.54 is returned.
1417 * systemd-notify will now send a "RELOADING=1" notification when called
1418 with --reloading, and "STOPPING=1" when called with --stopping. This
1419 can be used to implement notifications from units where it's easier
1420 to call a program than to use the sd-daemon library.
1422 * systemd-analyze's 'plot' command can now output its information in
1423 JSON, controlled via the --json= switch. Also, new --table, and
1424 --no-legend options have been added.
1426 * 'machinectl enable' will now automatically enable machines.target
1427 unit in addition to adding the machine unit to the target.
1429 Similarly, 'machinectl start|stop' gained a --now option to enable or
1430 disable the machine unit when starting or stopping it.
1432 * systemd-sysusers will now create /etc/ if it is missing.
1434 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
1435 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
1436 added to provide the new initial value for the new automated battery
1437 estimation functionality. If 'HibernateDelaySec=' is set to any value,
1438 the automated estimate (and thus the automated hibernation on low
1439 battery to avoid data loss) functionality will be disabled.
1441 * Default tmpfiles.d/ configuration will now automatically create
1442 credentials storage directory '/etc/credstore/' with the appropriate,
1443 secure permissions. If '/run/credstore/' exists, its permissions will
1444 be fixed too in case they are not correct.
1446 Changes in libsystemd and shared code:
1448 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
1449 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
1451 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
1452 128-bit ID in files such as /etc/machine-id has an invalid
1453 format. They also accept NULL as output parameter in more places,
1454 which is useful when the caller only wants to validate the inputs and
1455 does not need the output value.
1457 * sd-login gained new functions sd_pidfd_get_session(),
1458 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
1459 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
1460 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
1461 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
1462 but accept a PIDFD instead of a PID.
1464 * sd-path (and systemd-path) now export four new paths:
1465 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
1466 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
1467 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
1468 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
1470 * sd_notify() now supports AF_VSOCK as transport for notification
1471 messages (in addition to the existing AF_UNIX support). This is
1472 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
1474 * Detection of chroot() environments now works if /proc/ is not
1475 mounted. This affects systemd-detect-virt --chroot, but also means
1476 that systemd tools will silently skip various operations in such an
1479 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
1480 virtualization is now detected.
1482 Changes in the build system:
1484 * Standalone variants of systemd-repart and systemd-shutdown may now be
1485 built (if -Dstandalone=true).
1487 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
1488 example, allow scripts to conditionalize execution on AC power
1491 * The libp11kit library is now loaded through dlopen(3).
1493 Changes in the documentation:
1495 * Specifications that are not closely tied to systemd have moved to
1496 https://uapi-group.org/specifications/: the Boot Loader Specification
1497 and the Discoverable Partitions Specification.
1499 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
1500 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
1501 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
1502 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
1503 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
1504 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
1505 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
1506 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
1507 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
1508 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
1509 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
1510 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
1511 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
1512 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
1513 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
1514 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
1515 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
1516 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
1517 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
1518 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
1519 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
1520 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
1521 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
1522 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
1523 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
1524 reuben olinsky, Richard E. van der Luit, Richard Phibel,
1525 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
1526 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
1527 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
1528 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
1529 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
1530 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
1531 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
1532 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
1535 — Warsaw, 2023-02-15
1539 Announcements of Future Feature Removals:
1541 * We intend to remove cgroup v1 support from systemd release after the
1542 end of 2023. If you run services that make explicit use of cgroup v1
1543 features (i.e. the "legacy hierarchy" with separate hierarchies for
1544 each controller), please implement compatibility with cgroup v2 (i.e.
1545 the "unified hierarchy") sooner rather than later. Most of Linux
1546 userspace has been ported over already.
1548 * We intend to remove support for split-usr (/usr mounted separately
1549 during boot) and unmerged-usr (parallel directories /bin and
1550 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
1551 half of 2023, in the first release that falls into that time window.
1552 For more details, see:
1553 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1555 Compatibility Breaks:
1557 * ConditionKernelVersion= checks that use the '=' or '!=' operators
1558 will now do simple string comparisons (instead of version comparisons
1559 à la stverscmp()). Version comparisons are still done for the
1560 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
1561 specified, a shell-style glob match is now done. This creates a minor
1562 incompatibility compared to older systemd versions when the '*', '?',
1563 '[', ']' characters are used, as these will now match as shell globs
1564 instead of literally. Given that kernel version strings typically do
1565 not include these characters we expect little breakage through this
1568 * The service manager will now read the SELinux label used for SELinux
1569 access checks from the unit file at the time it loads the file.
1570 Previously, the label would be read at the moment of the access
1571 check, which was problematic since at that time the unit file might
1572 already have been updated or removed.
1576 * systemd-measure is a new tool for calculating and signing expected
1577 TPM2 PCR values for a given unified kernel image (UKI) booted via
1578 sd-stub. The public key used for the signature and the signed
1579 expected PCR information can be embedded inside the UKI. This
1580 information can be extracted from the UKI by external tools and code
1581 in the image itself and is made available to userspace in the booted
1584 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
1585 updated to make use of this information if available in the booted
1586 kernel: when locking an encrypted volume/credential to the TPM
1587 systemd-cryptenroll/systemd-creds will use the public key to bind the
1588 volume/credential to any kernel that carries PCR information signed
1589 by the same key pair. When unlocking such volumes/credentials
1590 systemd-cryptsetup/systemd-creds will use the signature embedded in
1591 the booted UKI to gain access.
1593 Binding TPM-based disk encryption to public keys/signatures of PCR
1594 values — instead of literal PCR values — addresses the inherent
1595 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
1596 disks remain accessible even if the UKI is updated, without any TPM
1597 specific preparation during the OS update — as long as each UKI
1598 carries the necessary PCR signature information.
1600 Net effect: if you boot a properly prepared kernel, TPM-bound disk
1601 encryption now defaults to be locked to kernels which carry PCR
1602 signatures from the same key pair. Example: if a hypothetical distro
1603 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
1604 by default – bound to only FooOS kernels, and encrypted volumes bound
1605 to the TPM cannot be unlocked on kernels from other sources. (But do
1606 note this behaviour requires preparation/enabling in the UKI, and of
1607 course users can always enroll non-TPM ways to unlock the volume.)
1609 * systemd-pcrphase is a new tool that is invoked at six places during
1610 system runtime, and measures additional words into TPM2 PCR 11, to
1611 mark milestones of the boot process. This allows binding access to
1612 specific TPM2-encrypted secrets to specific phases of the boot
1613 process. (Example: LUKS2 disk encryption key only accessible in the
1614 initrd, but not later.)
1616 Changes in systemd itself, i.e. the manager and units
1618 * The cpu controller is delegated to user manager units by default, and
1619 CPUWeight= settings are applied to the top-level user slice units
1620 (app.slice, background.slice, session.slice). This provides a degree
1621 of resource isolation between different user services competing for
1624 * Systemd can optionally do a full preset in the "first boot" condition
1625 (instead of just enable-only). This behaviour is controlled by the
1626 compile-time option -Dfirst-boot-full-preset. Right now it defaults
1627 to 'false', but the plan is to switch it to 'true' for the subsequent
1630 * Drop-ins are now allowed for transient units too.
1632 * Systemd will set the taint flag 'support-ended' if it detects that
1633 the OS image is past its end-of-support date. This date is declared
1634 in a new /etc/os-release field SUPPORT_END= described below.
1636 * Two new settings ConditionCredential= and AssertCredential= can be
1637 used to skip or fail units if a certain system credential is not
1640 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
1642 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
1643 specify the SMACK security label to use when not specified in a unit
1646 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
1647 specify the default timeout when waiting for device units to
1650 * C.UTF-8 is used as the default locale if nothing else has been
1653 * [Condition|Assert]Firmware= have been extended to support certain
1654 SMBIOS fields. For example
1656 ConditionFirmware=smbios-field(board_name = "Custom Board")
1658 conditionalizes the unit to run only when
1659 /sys/class/dmi/id/board_name contains "Custom Board" (without the
1662 * ConditionFirstBoot= now correctly evaluates as true only during the
1663 boot phase of the first boot. A unit executed later, after booting
1664 has completed, will no longer evaluate this condition as true.
1666 * Socket units will now create sockets in the SELinuxContext= of the
1667 associated service unit, if any.
1669 * Boot phase transitions (start initrd → exit initrd → boot complete →
1670 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
1671 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
1672 unsealed only in the initrd.
1674 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
1675 also be provided to ExecStartPre= processes.
1677 * Various units are now correctly ordered against
1678 initrd-switch-root.target where previously a conflict without
1679 ordering was configured. A stop job for those units would be queued,
1680 but without the ordering it could be executed only after
1681 initrd-switch-root.service, leading to units not being restarted in
1682 the host system as expected.
1684 * In order to fully support the IPMI watchdog driver, which has not yet
1685 been ported to the new common watchdog device interface,
1686 /dev/watchdog0 will be tried first and systemd will silently fallback
1687 to /dev/watchdog if it is not found.
1689 * New watchdog-related D-Bus properties are now published by systemd:
1690 WatchdogDevice, WatchdogLastPingTimestamp,
1691 WatchdogLastPingTimestampMonotonic.
1693 * At shutdown, API virtual files systems (proc, sys, etc.) will be
1696 * At shutdown, systemd will now log about processes blocking unmounting
1699 * A new meson build option 'clock-valid-range-usec-max' was added to
1700 allow disabling system time correction if RTC returns a timestamp far
1703 * Propagated restart jobs will no longer be discarded while a unit is
1706 * PID 1 will now import system credentials from SMBIOS Type 11 fields
1707 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
1708 simple, fast and generic path for supplying credentials to a VM,
1709 without involving external tools such as cloud-init/ignition.
1711 * The CPUWeight= setting of unit files now accepts a new special value
1712 "idle", which configures "idle" level scheduling for the unit.
1714 * Service processes that are activated due to a .timer or .path unit
1715 triggering will now receive information about this via environment
1716 variables. Note that this is information is lossy, as activation
1717 might be coalesced and only one of the activating triggers will be
1718 reported. This is hence more suited for debugging or tracing rather
1719 than for behaviour decisions.
1721 * The riscv_flush_icache(2) system call has been added to the list of
1722 system calls allowed by default when SystemCallFilter= is used.
1724 * The selinux context derived from the target executable, instead of
1725 'init_t' used for the manager itself, is now used when creating
1726 listening sockets for units that specify SELinuxContextFromNet=yes.
1728 Changes in sd-boot, bootctl, and the Boot Loader Specification:
1730 * The Boot Loader Specification has been cleaned up and clarified.
1731 Various corner cases in version string comparisons have been fixed
1732 (e.g. comparisons for empty strings). Boot counting is now part of
1733 the main specification.
1735 * New PCRs measurements are performed during boot: PCR 11 for the
1736 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
1737 took place this is now reported to userspace via the new
1738 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
1740 * As before, systemd-stub will measure kernel parameters and system
1741 credentials into PCR 12. It will now report this fact via the
1742 StubPcrKernelParameters EFI variable to userspace.
1744 * The UEFI monotonic boot counter is now included in the updated random
1745 seed file maintained by sd-boot, providing some additional entropy.
1747 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
1748 of arranging the image manually and jumping to the kernel entry
1749 point. sd-stub also installs a temporary UEFI SecurityOverride to
1750 allow the (unsigned) nested image to be booted. This is safe because
1751 the outer (signed) stub+kernel binary must have been verified before
1752 the stub was executed.
1754 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
1755 is now supported by sd-boot.
1757 * bootctl gained a bunch of new options: --all-architectures to install
1758 binaries for all supported EFI architectures, --root= and --image=
1759 options to operate on a directory or disk image, and
1760 --install-source= to specify the source for binaries to install,
1761 --efi-boot-option-description= to control the name of the boot entry.
1763 * The sd-boot stub exports a StubFeatures flag, which is used by
1764 bootctl to show features supported by the stub that was used to boot.
1766 * The PE section offsets that are used by tools that assemble unified
1767 kernel images have historically been hard-coded. This may lead to
1768 overlapping PE sections which may break on boot. The UKI will now try
1769 to detect and warn about this.
1771 Any tools that assemble UKIs must update to calculate these offsets
1772 dynamically. Future sd-stub versions may use offsets that will not
1773 work with the currently used set of hard-coded offsets!
1775 * sd-stub now accepts (and passes to the initrd and then to the full
1776 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
1777 signatures of expected PCR values, to allow sealing secrets via the
1778 TPM2 against pre-calculated PCR measurements.
1780 Changes in the hardware database:
1782 * 'systemd-hwdb query' now supports the --root= option.
1784 Changes in systemctl:
1786 * systemctl now supports --state= and --type= options for the 'show'
1789 * systemctl gained a new verb 'list-automounts' to list automount
1792 * systemctl gained support for a new --image= switch to be able to
1793 operate on the specified disk image (similar to the existing --root=
1794 which operates relative to some directory).
1796 Changes in systemd-networkd:
1798 * networkd can set Linux NetLabel labels for integration with the
1799 network control in security modules via a new NetLabel= option.
1801 * The RapidCommit= is (re-)introduced to enable faster configuration
1802 via DHCPv6 (RFC 3315).
1804 * networkd gained a new option TCPCongestionControlAlgorithm= that
1805 allows setting a per-route TCP algorithm.
1807 * networkd gained a new option KeepFileDescriptor= to allow keeping a
1808 reference (file descriptor) open on TUN/TAP interfaces, which is
1809 useful to avoid link flaps while the underlying service providing the
1810 interface is being serviced.
1812 * RouteTable= now also accepts route table names.
1814 Changes in systemd-nspawn:
1816 * The --bind= and --overlay= options now support relative paths.
1818 * The --bind= option now supports a 'rootidmap' value, which will
1819 use id-mapped mounts to map the root user inside the container to the
1820 owner of the mounted directory on the host.
1822 Changes in systemd-resolved:
1824 * systemd-resolved now persists DNSOverTLS in its state file too. This
1825 fixes a problem when used in combination with NetworkManager, which
1826 sends the setting only once, causing it to be lost if resolved was
1827 restarted at any point.
1829 * systemd-resolved now exposes a varlink socket at
1830 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
1831 root. Processed DNS requests in a JSON format will be published to
1832 any clients connected to this socket.
1834 resolvectl gained a 'monitor' verb to make use of this.
1836 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
1837 instead of returning SERVFAIL, as per RFC:
1838 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
1840 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
1841 is still supported.)
1843 Changes in libsystemd and other libraries:
1845 * libsystemd now exports sd_bus_error_setfv() (a convenience function
1846 for setting bus errors), sd_id128_string_equal (a convenience
1847 function for 128-bit ID string comparisons), and
1848 sd_bus_message_read_strv_extend() (a function to incrementally read
1851 * libsystemd now exports sd_device_get_child_first()/_next() as a
1852 high-level interface for enumerating child devices. It also supports
1853 sd_device_new_child() for opening a child device given a device
1856 * libsystemd now exports sd_device_monitor_set()/get_description()
1857 which allow setting a custom description that will be used in log
1858 messages by sd_device_monitor*.
1860 * Private shared libraries (libsystemd-shared-nnn.so,
1861 libsystemd-core-nnn.so) are now installed into arch-specific
1862 directories to allow multi-arch installs.
1864 * A new sd-gpt.h header is now published, listing GUIDs from the
1865 Discoverable Partitions specification. For more details see:
1866 https://systemd.io/DISCOVERABLE_PARTITIONS/
1868 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
1869 database given an explicit path to the file.
1871 * The signal number argument to sd_event_add_signal() now can now be
1872 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
1873 be automatically invoked to block the specified signal. This is
1874 useful to simplify invocations as the caller doesn't have to do this
1877 * A new convenience call sd_event_set_signal_exit() has been added to
1878 sd-event to set up signal handling so that the event loop
1879 automatically terminates cleanly on SIGTERM/SIGINT.
1881 Changes in other components:
1883 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
1884 can now be provided via the credential mechanism.
1886 * systemd-analyze gained a new verb 'compare-versions' that implements
1887 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
1888 'dpkg --compare-versions').
1890 * 'systemd-analyze dump' is extended to accept glob patterns for unit
1891 names to limit the output to matching units.
1893 * tmpfiles.d/ lines can read file contents to write from a credential.
1894 The new modifier char '^' is used to specify that the argument is a
1895 credential name. This mechanism is used to automatically populate
1896 /etc/motd, /etc/issue, and /etc/hosts from credentials.
1898 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
1899 an inode if the specification is prefixed with ':' and the inode
1902 * Default tmpfiles.d/ configuration now carries a line to automatically
1903 use an 'ssh.authorized_keys.root' credential if provided to set up
1904 the SSH authorized_keys file for the root user.
1906 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
1909 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
1910 in base64. This is useful to write arbitrary binary data into files.
1912 * The pkgconfig and rpm macros files now export the directory for user
1913 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
1915 * Detection of Apple Virtualization and detection of Parallels and
1916 KubeVirt virtualization on non-x86 archs have been added.
1918 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
1919 user when their system will become unsupported.
1921 * When performing suspend-then-hibernate, the system will estimate the
1922 discharge rate and use that to set the delay until hibernation and
1923 hibernate immediately instead of suspending when running from a
1924 battery and the capacity is below 5%.
1926 * systemd-sysctl gained a --strict option to fail when a sysctl
1927 setting is unknown to the kernel.
1929 * machinectl supports --force for the 'copy-to' and 'copy-from'
1932 * coredumpctl gained the --root and --image options to look for journal
1933 files under the specified root directory, image, or block device.
1935 * 'journalctl -o' and similar commands now implement a new output mode
1936 "short-delta". It is similar to "short-monotonic", but also shows the
1937 time delta between subsequent messages.
1939 * journalctl now respects the --quiet flag when verifying consistency
1942 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
1943 will indicate whether a message was logged in the 'initrd' phase or
1944 in the 'system' phase of the boot process.
1946 * Journal files gained a new compatibility flag
1947 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
1948 to the storage format that allow reducing size on disk. As with other
1949 compatibility flags, older journalctl versions will not be able to
1950 read journal files using this new format. The environment variable
1951 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
1952 disable this functionality. It is enabled by default.
1954 * systemd-run's --working-directory= switch now works when used in
1955 combination with --scope.
1957 * portablectl gained a --force flag to skip certain sanity checks. This
1958 is implemented using new flags accepted by systemd-portabled for the
1959 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
1960 flag now means that the attach/detach checks whether the units are
1961 already present and running will be skipped. Similarly,
1962 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
1963 image name matches the name declared inside of the image will be
1964 skipped. Callers must be sure to do those checks themselves if
1967 * systemd-portabled will now use the original filename to check
1968 extension-release.NAME for correctness, in case it is passed a
1971 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
1974 * sysext's extension-release files now support '_any' as a special
1975 value for the ID= field, to allow distribution-independent extensions
1976 (e.g.: fully statically compiled binaries, scripts). It also gained
1977 support for a new ARCHITECTURE= field that may be used to explicitly
1978 restrict an image to hosts of a specific architecture.
1980 * systemd-repart now supports creating squashfs partitions. This
1981 requires mksquashfs from squashfs-tools.
1983 * systemd-repart gained a --split flag to also generate split
1984 artifacts, i.e. a separate file for each partition. This is useful in
1985 conjunction with systemd-sysupdate or other tools, or to generate
1986 split dm-verity artifacts.
1988 * systemd-repart is now able to generate dm-verity partitions, including
1991 * systemd-repart can now set a partition UUID to zero, allowing it to
1992 be filled in later, such as when using verity partitions.
1994 * systemd-repart now supports drop-ins for its configuration files.
1996 * Package metadata logged by systemd-coredump in the system journal is
1999 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
2001 * systemd-oomd now automatically links against libatomic, if available.
2003 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
2006 * scope units now also provide oom-kill status.
2008 * systemd-pstore will now try to load only the efi_pstore kernel module
2009 before running, ensuring that pstore can be used.
2011 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
2012 session after a preconfigure timeout.
2014 * systemd-homed will now wait up to 30 seconds for workers to terminate,
2015 rather than indefinitely.
2017 * homectl gained a new '--luks-sector-size=' flag that allows users to
2018 select the preferred LUKS sector size. Must be a power of 2 between 512
2019 and 4096. systemd-userdbd records gained a corresponding field.
2021 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
2022 variable when generating the 'sp_lstchg' field, to ensure an image
2023 build can be reproducible.
2025 * 'udevadm wait' will now listen to kernel uevents too when called with
2028 * When naming network devices udev will now consult the Devicetree
2029 "alias" fields for the device.
2031 * systemd-udev will now create infiniband/by-path and
2032 infiniband/by-ibdev links for Infiniband verbs devices.
2034 * systemd-udev-trigger.service will now also prioritize input devices.
2036 * ConditionACPower= and systemd-ac-power will now assume the system is
2037 running on AC power if no battery can be found.
2039 * All features and tools using the TPM2 will now communicate with it
2040 using a bind key. Beforehand, the tpm2 support used encrypted sessions
2041 by creating a primary key that was used to encrypt traffic. This
2042 creates a problem as the key created for encrypting the traffic could
2043 be faked by an active interposer on the bus. In cases when a pin is
2044 used, a bind key will be used. The pin is used as the auth value for
2045 the seal key, aka the disk encryption key, and that auth value will be
2046 used in the session establishment. An attacker would need the pin
2047 value to create the secure session and thus an active interposer
2048 without the pin cannot interpose on TPM2 traffic.
2050 * systemd-growfs no longer requires udev to run.
2052 * systemd-backlight now will better support systems with multiple
2055 * systemd-cryptsetup's keyfile-timeout= option now also works when a
2056 device is used as a keyfile.
2058 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
2059 unlocking key from a key file (instead of prompting the user). Note
2060 that this is the key for unlocking the volume in order to be able to
2061 enroll a new key, but it is not the key that is enrolled.
2063 * systemd-dissect gained a new --umount switch that will safely and
2064 synchronously unmount all partitions of an image previously mounted
2065 with 'systemd-dissect --mount'.
2067 * When using gcrypt, all systemd tools and services will now configure
2068 it to prefer the OS random number generator if present.
2070 * All example code shipped with documentation has been relicensed from CC0
2073 * Unit tests will no longer fail when running on a system without
2076 Experimental features:
2078 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
2079 and bpftool >= 7.0).
2081 * sd-boot can automatically enroll SecureBoot keys from files found on
2082 the ESP. This enrollment can be either automatic ('force' mode) or
2083 controlled by the user ('manual' mode). It is sufficient to place the
2084 SecureBoot keys in the right place in the ESP and they will be picked
2085 up by sd-boot and shown in the boot menu.
2087 * The mkosi config in systemd gained support for automatically
2088 compiling a kernel with the configuration appropriate for testing
2089 systemd. This may be useful when developing or testing systemd in
2090 tandem with the kernel.
2092 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
2093 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
2094 Alexander Graf, Alexander Shopov, Alexander Wilson,
2095 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
2096 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
2097 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
2098 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
2099 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
2100 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
2101 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
2102 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
2103 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
2104 David Jaša, David Rheinsberg, David Seifert, David Tardon,
2105 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
2106 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
2107 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
2108 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
2109 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
2110 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
2111 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
2112 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
2113 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
2114 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
2115 JeroenHD, jiangchuangang, João Loureiro,
2116 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
2117 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
2118 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
2119 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
2120 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
2121 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
2122 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
2123 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
2124 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
2125 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
2126 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
2127 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
2128 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
2129 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
2130 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
2131 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
2132 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
2133 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
2134 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
2135 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
2136 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
2137 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
2138 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
2139 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
2141 – The Great Beyond, 2022-10-31 👻
2145 Backwards-incompatible changes:
2147 * The minimum kernel version required has been bumped from 3.13 to 4.15,
2148 and CLOCK_BOOTTIME is now assumed to always exist.
2150 * C11 with GNU extensions (aka "gnu11") is now used to build our
2151 components. Public API headers are still restricted to ISO C89.
2153 * In v250, a systemd-networkd feature that automatically configures
2154 routes to addresses specified in AllowedIPs= was added and enabled by
2155 default. However, this causes network connectivity issues in many
2156 existing setups. Hence, it has been disabled by default since
2157 systemd-stable 250.3. The feature can still be used by explicitly
2158 configuring RouteTable= setting in .netdev files.
2160 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
2161 when a Condition*= check does not succeed, restoring the JobRemoved
2162 signal to the behaviour it had before v250.
2164 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
2165 GetImageMetadataWithExtensions() have been fixed to provide an extra
2166 return parameter, containing the actual extension release metadata.
2167 The current implementation was judged to be broken and unusable, and
2168 thus the usual procedure of adding a new set of methods was skipped,
2169 and backward compatibility broken instead on the assumption that
2170 nobody can be affected given the current state of this interface.
2172 * All kernels supported by systemd mix bytes returned by RDRAND (or
2173 similar) into the entropy pool at early boot. This means that on
2174 those systems, even if /dev/urandom is not yet initialized, it still
2175 returns bytes that are of at least RDRAND quality. For that reason,
2176 we no longer have reason to invoke RDRAND from systemd itself, which
2177 has historically been a source of bugs. Furthermore, kernels ≥5.6
2178 provide the getrandom(GRND_INSECURE) interface for returning random
2179 bytes before the entropy pool is initialized without warning into
2180 kmsg, which is what we attempt to use if available. systemd's direct
2181 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
2182 running an older kernel may experience kmsg warnings that were not
2183 seen with 250. For newer kernels, non-x86 systems, or older x86
2184 systems, there should be no visible changes.
2186 * sd-boot will now measure the kernel command line into TPM PCR 12
2187 rather than PCR 8. This improves usefulness of the measurements on
2188 systems where sd-boot is chainloaded from Grub. Grub measures all
2189 commands its executes into PCR 8, which makes it very hard to use
2190 reasonably, hence separate ourselves from that and use PCR 12
2191 instead, which is what certain Ubuntu editions already do. To retain
2192 compatibility with systems running older systemd systems a new meson
2193 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
2194 If enabled, the measurement is done twice: into the new-style PCR 12
2195 *and* the old-style PCR 8. It's strongly advised to migrate all users
2196 to PCR 12 for this purpose in the long run, as we intend to remove
2197 this compatibility feature in two years' time.
2199 * busctl capture now writes output in the newer pcapng format instead
2202 * A udev rule that imported hwdb matches for USB devices with lowercase
2203 hexadecimal vendor/product ID digits was added in systemd 250. This
2204 has been reverted, since uppercase hexadecimal digits are supposed to
2205 be used, and we already had a rule with the appropriate match.
2207 Users might need to adjust their local hwdb entries.
2209 * arch_prctl(2) has been moved to the @default set in the syscall filters
2210 (as exposed via the SystemCallFilter= setting in service unit files).
2211 It is apparently used by the linker now.
2213 * The tmpfiles entries that create the /run/systemd/netif directory and
2214 its subdirectories were moved from tmpfiles.d/systemd.conf to
2215 tmpfiles.d/systemd-network.conf.
2217 Users might need to adjust their files that override tmpfiles.d/systemd.conf
2218 to account for this change.
2220 * The requirement for Portable Services images to contain a well-formed
2221 os-release file (i.e.: contain at least an ID field) is now enforced.
2222 This applies to base images and extensions, and also to systemd-sysext.
2224 Changes in the Boot Loader Specification, kernel-install and sd-boot:
2226 * kernel-install's and bootctl's Boot Loader Specification Type #1
2227 entry generation logic has been reworked. The user may now pick
2228 explicitly by which "token" string to name the installation's boot
2229 entries, via the new /etc/kernel/entry-token file or the new
2230 --entry-token= switch to bootctl. By default — as before — the
2231 entries are named after the local machine ID. However, in "golden
2232 image" environments, where the machine ID shall be initialized on
2233 first boot (as opposed to at installation time before first boot) the
2234 machine ID will not be available at build time. In this case the
2235 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
2236 file) may be used to override the "token" for the entries, for
2237 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
2238 make the OS images independent of any machine ID, and ensure that the
2239 images will not carry any identifiable information before first boot,
2240 but on the other hand means that multiple parallel installations of
2241 the very same image on the same disk cannot be supported.
2243 Summary: if you are building golden images that shall acquire
2244 identity information exclusively on first boot, make sure to both
2245 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
2246 value of the IMAGE_ID= or ID= field of /etc/os-release or another
2247 suitable identifier before deploying the image.
2249 * The Boot Loader Specification has been extended with
2250 /loader/entries.srel file located in the EFI System Partition (ESP)
2251 that disambiguates the format of the entries in the /loader/entries/
2252 directory (in order to discern them from incompatible uses of this
2253 directory by other projects). For entries that follow the
2254 Specification, the string "type1" is stored in this file.
2256 bootctl will now write this file automatically when installing the
2257 systemd-boot boot loader.
2259 * kernel-install supports a new initrd_generator= setting in
2260 /etc/kernel/install.conf, that is exported as
2261 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
2262 allows choosing different initrd generators.
2264 * kernel-install will now create a "staging area" (an initially-empty
2265 directory to gather files for a Boot Loader Specification Type #1
2266 entry). The path to this directory is exported as
2267 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
2268 drop files there instead of writing them directly to the final
2269 location. kernel-install will move them when all files have been
2270 prepared successfully.
2272 * New option sort-key= has been added to the Boot Loader Specification
2273 to override the sorting order of the entries in the boot menu. It is
2274 read by sd-boot and bootctl, and will be written by kernel-install,
2275 with the default value of IMAGE_ID= or ID= fields from
2276 os-release. Together, this means that on multiboot installations,
2277 entries should be grouped and sorted in a predictable way.
2279 * The sort order of boot entries has been updated: entries which have
2280 the new field sort-key= are sorted by it first, and all entries
2281 without it are ordered later. After that, entries are sorted by
2282 version so that newest entries are towards the beginning of the list.
2284 * The kernel-install tool gained a new 'inspect' verb which shows the
2285 paths and other settings used.
2287 * sd-boot can now optionally beep when the menu is shown and menu
2288 entries are selected, which can be useful on machines without a
2289 working display. (Controllable via a loader.conf setting.)
2291 * The --make-machine-id-directory= switch to bootctl has been replaced
2292 by --make-entry-directory=, given that the entry directory is not
2293 necessarily named after the machine ID, but after some other suitable
2294 ID as selected via --entry-token= described above. The old name of
2295 the option is still understood to maximize compatibility.
2297 * 'bootctl list' gained support for a new --json= switch to output boot
2298 menu entries in JSON format.
2300 * 'bootctl is-installed' now supports the --graceful, and various verbs
2301 omit output with the new option --quiet.
2303 Changes in systemd-homed:
2305 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
2306 of activated home directories it manages (if the kernel and selected
2307 file systems support it). So far it mapped three UID ranges: the
2308 range from 0…60000, the user's own UID, and the range 60514…65534,
2309 leaving everything else unmapped (in other words, the 16-bit UID range
2310 is mapped almost fully, with the exception of the UID subrange used
2311 for systemd-homed users, with one exception: the user's own UID).
2312 Unmapped UIDs may not be used for file ownership in the home
2313 directory — any chown() attempts with them will fail. With this
2314 release a fourth range is added to these mappings:
2315 524288…1879048191. This range is the UID range intended for container
2318 https://systemd.io/UIDS-GIDS
2320 This range may be used for container managers that place container OS
2321 trees in the home directory (which is a questionable approach, for
2322 quota, permission, SUID handling and network file system
2323 compatibility reasons, but nonetheless apparently commonplace). Note
2324 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
2325 UID assignments from the range are not managed or mapped by
2326 `systemd-homed`, and must be managed with other mechanisms, in the
2327 context of the local system.
2329 Typically, a better approach to user namespacing in relevant
2330 container managers would be to leave container OS trees on disk at
2331 UID offset 0, but then map them to a dynamically allocated runtime
2332 UID range via another UID mount map at container invocation
2333 time. That way user namespace UID ranges become strictly a runtime
2334 concept, and do not leak into persistent file systems, persistent
2335 user databases or persistent configuration, thus greatly simplifying
2336 handling, and improving compatibility with home directories intended
2337 to be portable like the ones managed by systemd-homed.
2339 Changes in shared libraries:
2341 * A new libsystemd-core-<version>.so private shared library is
2342 installed under /usr/lib/systemd/system, mirroring the existing
2343 libsystemd-shared-<version>.so library. This allows the total
2344 installation size to be reduced by binary code reuse.
2346 * The <version> tag used in the name of libsystemd-shared.so and
2347 libsystemd-core.so can be configured via the meson option
2348 'shared-lib-tag'. Distributions may build subsequent versions of the
2349 systemd package with unique tags (e.g. the full package version),
2350 thus allowing multiple installations of those shared libraries to be
2351 available at the same time. This is intended to fix an issue where
2352 programs that link to those libraries would fail to execute because
2353 they were installed earlier or later than the appropriate version of
2356 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
2357 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
2358 format instead of as a simple series of hex characters.
2360 * The sd-device API gained two new calls sd_device_new_from_devname()
2361 and sd_device_new_from_path() which permit allocating an sd_device
2362 object from a device node name or file system path.
2364 * sd-device also gained a new call sd_device_open() which will open the
2365 device node associated with a device for which an sd_device object
2366 has been allocated. The call is supposed to address races around
2367 device nodes being removed/recycled due to hotplug events, or media
2368 change events: the call checks internally whether the major/minor of
2369 the device node and the "diskseq" (in case of block devices) match
2370 with the metadata loaded in the sd_device object, thus ensuring that
2371 the device once opened really matches the provided sd_device object.
2373 Changes in PID1, systemctl, and systemd-oomd:
2375 * A new set of service monitor environment variables will be passed to
2376 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
2377 handler unit as OnFailure=/OnSuccess=. The variables are:
2378 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
2379 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
2380 handler needs to watch multiple units, use a templated handler.
2382 * A new ExtensionDirectories= setting in service unit files allows
2383 system extensions to be loaded from a directory. (It is similar to
2384 ExtensionImages=, but takes paths to directories, instead of
2387 'portablectl attach --extension=' now also accepts directory paths.
2389 * The user.delegate and user.invocation_id extended attributes on
2390 cgroups are used in addition to trusted.delegate and
2391 trusted.invocation_id. The latter pair requires privileges to set,
2392 but the former doesn't and can be also set by the unprivileged user
2395 (Only supported on kernels ≥5.6.)
2397 * Units that were killed by systemd-oomd will now have a service result
2398 of 'oom-kill'. The number of times a service was killed is tallied
2399 in the 'user.oomd_ooms' extended attribute.
2401 The OOMPolicy= unit file setting is now also honoured by
2404 * In unit files the new %y/%Y specifiers can be used to refer to
2405 normalized unit file path, which is particularly useful for symlinked
2408 The new %q specifier resolves to the pretty hostname
2409 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
2411 The new %d specifier resolves to the credentials directory of a
2412 service (same as $CREDENTIALS_DIRECTORY).
2414 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
2415 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
2416 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
2417 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
2418 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
2419 MountFlags= service settings now also work in unprivileged user
2420 services, i.e. those run by the user's --user service manager, as long
2421 as user namespaces are enabled on the system.
2423 * Services with Restart=always and a failing ExecCondition= will no
2424 longer be restarted, to bring ExecCondition= behaviour in line with
2425 Condition*= settings.
2427 * LoadCredential= now accepts a directory as the argument; all files
2428 from the directory will be loaded as credentials.
2430 * A new D-Bus property ControlGroupId is now exposed on service units,
2431 that encapsulates the service's numeric cgroup ID that newer kernels
2432 assign to each cgroup.
2434 * PID 1 gained support for configuring the "pre-timeout" of watchdog
2435 devices and the associated governor, via the new
2436 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
2437 options in /etc/systemd/system.conf.
2439 * systemctl's --timestamp= option gained a new choice "unix", to show
2440 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
2442 * A new "taint" flag named "old-kernel" is introduced which is set when
2443 the kernel systemd runs on is older then the current baseline version
2444 (see above). The flag is shown in "systemctl status" output.
2446 * Two additional taint flags "short-uid-range" and "short-gid-range"
2447 have been added as well, which are set when systemd notices it is run
2448 within a userns namespace that does not define the full 0…65535 UID
2451 * A new "unmerged-usr" taint flag has been added that is set whenever
2452 running on systems where /bin/ + /sbin/ are *not* symlinks to their
2453 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
2456 * Generators invoked by PID 1 will now have a couple of useful
2457 environment variables set describing the execution context a
2458 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
2459 system service manager, or from the per-user service
2460 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
2461 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
2462 systemd considers the current boot to be a "first"
2463 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
2464 detected and which type of hypervisor/container
2465 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
2466 kernel is built for.
2468 * PID 1 will now automatically pick up system credentials from qemu's
2469 fw_cfg interface, thus allowing passing arbitrary data into VM
2470 systems similar to how this is already supported for passing them
2471 into `systemd-nspawn` containers. Credentials may now also be passed
2472 in via the new kernel command line option `systemd.set_credential=`
2473 (note that kernel command line options are world-readable during
2474 runtime, and only useful for credentials that require no
2475 confidentiality). The credentials that can be passed to unified
2476 kernels that use the `systemd-stub` UEFI stub are now similarly
2477 picked up automatically. Automatic importing of system credentials
2478 this way can be turned off via the new
2479 `systemd.import_credentials=no` kernel command line option.
2481 * LoadCredential= will now automatically look for credentials in the
2482 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
2483 the argument is not an absolute path. Similarly,
2484 LoadCredentialEncrypted= will check the same directories plus
2485 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
2486 /usr/lib/credstore.encrypted/. The idea is to use those directories
2487 as the system-wide location for credentials that services should pick
2490 * System and service credentials are described in great detail in a new
2493 https://systemd.io/CREDENTIALS
2495 Changes in systemd-journald:
2497 * The journal JSON export format has been added to listed of stable
2498 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
2500 * journalctl --list-boots now supports JSON output and the --reverse option.
2502 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
2503 updated, BUILDING_IMAGES is new:
2505 https://systemd.io/JOURNAL_EXPORT_FORMATS
2506 https://systemd.io/BUILDING_IMAGES
2510 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
2511 calculators, etc.), the other AV production devices (DJ tables,
2512 keypads, etc.) that should accessible to the seat owner user by
2515 * udevadm trigger gained a new --prioritized-subsystem= option to
2516 process certain subsystems (and all their parent devices) earlier.
2518 systemd-udev-trigger.service now uses this new option to trigger
2519 block and TPM devices first, hopefully making the boot a bit faster.
2521 * udevadm trigger now implements --type=all, --initialized-match,
2522 --initialized-nomatch to trigger both subsystems and devices, only
2523 already-initialized devices, and only devices which haven't been
2524 initialized yet, respectively.
2526 * udevadm gained a new "wait" command for safely waiting for a specific
2527 device to show up in the udev device database. This is useful in
2528 scripts that asynchronously allocate a block device (e.g. through
2529 repartitioning, or allocating a loopback device or similar) and need
2530 to synchronize on the creation to complete.
2532 * udevadm gained a new "lock" command for locking one or more block
2533 devices while formatting it or writing a partition table to it. It is
2534 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
2535 usable in scripts dealing with block devices.
2537 * udevadm info will show a couple of additional device fields in its
2538 output, and will not apply a limited set of coloring to line types.
2540 * udevadm info --tree will now show a tree of objects (i.e. devices and
2541 suchlike) in the /sys/ hierarchy.
2543 * Block devices will now get a new set of device symlinks in
2544 /dev/disk/by-diskseq/<nr>, which may be used to reference block
2545 device nodes via the kernel's "diskseq" value. Note that this does
2546 not guarantee that opening a device by a symlink like this will
2547 guarantee that the opened device actually matches the specified
2548 diskseq value. To be safe against races, the actual diskseq value of
2549 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
2550 the one in the symlink path.
2552 * .link files gained support for setting MDI/MID-X on a link.
2554 * .link files gained support for [Match] Firmware= setting to match on
2555 the device firmware description string. By mistake, it was previously
2556 only supported in .network files.
2558 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
2559 and [SR-IOV] section to configure SR-IOV virtual functions.
2561 Changes in systemd-networkd:
2563 * The default scope for unicast routes configured through [Route]
2564 section is changed to "link", to make the behavior consistent with
2565 "ip route" command. The manual configuration of [Route] Scope= is
2568 * A new unit systemd-networkd-wait-online@<interface>.service has been
2569 added that can be used to wait for a specific network interface to be
2572 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
2573 that configures the eponymous kernel attribute on the bridge.
2575 * .netdev files now can be used to create virtual WLAN devices, and
2576 configure various settings on them, via the [WLAN] section.
2578 * .link/.network files gained support for [Match] Kind= setting to match
2579 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
2581 This value is also shown by 'networkctl status'.
2583 * The Local= setting in .netdev files for various virtual network
2584 devices gained support for specifying, in addition to the network
2585 address, the name of a local interface which must have the specified
2588 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
2589 files, to configure tunnels in external mode (a.k.a. collect metadata
2592 * [Network] L2TP= setting was removed. Please use interface specifier in
2593 Local= setting in .netdev files of corresponding L2TP interface.
2595 * New [DHCPServer] BootServerName=, BootServerAddress=, and
2596 BootFilename= settings can be used to configure the server address,
2597 server name, and file name sent in the DHCP packet (e.g. to configure
2600 Changes in systemd-resolved:
2602 * systemd-resolved is started earlier (in sysinit.target), so it
2603 available earlier and will also be started in the initrd if installed
2606 Changes in disk encryption:
2608 * systemd-cryptenroll can now control whether to require the user to
2609 enter a PIN when using TPM-based unlocking of a volume via the new
2610 --tpm2-with-pin= option.
2612 Option tpm2-pin= can be used in /etc/crypttab.
2614 * When unlocking devices via TPM, TPM2 parameter encryption is now
2615 used, to ensure that communication between CPU and discrete TPM chips
2616 cannot be eavesdropped to acquire disk encryption keys.
2618 * A new switch --fido2-credential-algorithm= has been added to
2619 systemd-cryptenroll allowing selection of the credential algorithm to
2620 use when binding encryption to FIDO2 tokens.
2622 Changes in systemd-hostnamed:
2624 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
2625 to override the values gleaned from the hwdb.
2627 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
2628 /sys/class/dmi/id) to override the chassis that is reported by
2631 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
2632 for reading the hardware serial number, as reportd by DMI. It also
2633 exposes a new method D-Bus property FirmwareVersion that encode the
2634 firmware version of the system.
2636 Changes in other components:
2638 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
2639 handling with the values that were configured during systemd build
2640 (if /etc/locale.conf has not been created through some other
2641 mechanism). This means that /etc/locale.conf should always have
2642 reasonable contents and we avoid a potential mismatch in defaults.
2644 * The userdbctl tool will now show UID range information as part of the
2645 list of known users.
2647 * A new build-time configuration setting default-user-shell= can be
2648 used to set the default shell for user records and nspawn shell
2649 invocations (instead of the default /bin/bash).
2651 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
2652 information dynamically at runtime via IPC.
2654 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
2655 whether a functioning TPM2 infrastructure is available, i.e. if
2656 firmware, kernel driver and systemd all have TPM2 support enabled and
2659 * The systemd-creds tool gained support for generating encrypted
2660 credentials that are using an empty encryption key. While this
2661 provides no integrity nor confidentiality it's useful to implement
2662 codeflows that work the same on TPM-ful and TPM2-less systems. The
2663 service manager will only accept credentials "encrypted" that way if
2664 a TPM2 device cannot be detected, to ensure that credentials
2665 "encrypted" like that cannot be used to trick TPM2 systems.
2667 * When deciding whether to colorize output, all systemd programs now
2668 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
2671 * Meson's new install_tag feature is now in use for several components,
2672 allowing to build and install select binaries only: pam, nss, devel
2673 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
2674 $ meson build systemd-boot
2675 $ meson install --tags systemd-boot --no-rebuild
2676 https://mesonbuild.com/Installing.html#installation-tags
2678 * A new build configuration option has been added, to allow selecting the
2679 default compression algorithm used by systemd-journald and systemd-coredump.
2680 This allows to build-in support for decompressing all supported formats,
2681 but choose a specific one for compression. E.g.:
2682 $ meson -Ddefault-compression=xz
2684 Experimental features:
2686 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
2687 loader.conf that implements booting Microsoft Windows from the
2688 sd-boot in a way that first reboots the system, to reset the TPM
2689 PCRs. This improves compatibility with BitLocker's TPM use, as the
2690 PCRs will only record the Windows boot process, and not sd-boot
2691 itself, thus retaining the PCR measurements not involving sd-boot.
2692 Note that this feature is experimental for now, and is likely going
2693 to be generalized and renamed in a future release, without retaining
2694 compatibility with the current implementation.
2696 * A new systemd-sysupdate component has been added that automatically
2697 discovers, downloads, and installs A/B-style updates for the host
2698 installation itself, or container images, portable service images,
2699 and other assets. See the new systemd-sysupdate man page for updates.
2701 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
2702 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
2703 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
2704 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
2705 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
2706 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
2707 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
2708 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
2709 David, David Bond, Davide Cavalca, David Tardon, davijosw,
2710 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
2711 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
2712 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
2713 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
2714 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
2715 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
2716 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
2717 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
2718 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
2719 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
2720 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
2721 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
2722 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
2723 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
2724 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
2725 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
2726 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
2727 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
2728 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
2729 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
2730 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
2731 Simon Ellmann, Sonali Srivastava, Stefan Seering,
2732 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
2733 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
2734 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
2735 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
2736 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
2737 Zbigniew Jędrzejewski-Szmek, наб
2739 — Edinburgh, 2022-05-21
2743 * Support for encrypted and authenticated credentials has been added.
2744 This extends the credential logic introduced with v247 to support
2745 non-interactive symmetric encryption and authentication, based on a
2746 key that is stored on the /var/ file system or in the TPM2 chip (if
2747 available), or the combination of both (by default if a TPM2 chip
2748 exists the combination is used, otherwise the /var/ key only). The
2749 credentials are automatically decrypted at the moment a service is
2750 started, and are made accessible to the service itself in unencrypted
2751 form. A new tool 'systemd-creds' encrypts credentials for this
2752 purpose, and two new service file settings LoadCredentialEncrypted=
2753 and SetCredentialEncrypted= configure such credentials.
2755 This feature is useful to store sensitive material such as SSL
2756 certificates, passwords and similar securely at rest and only decrypt
2757 them when needed, and in a way that is tied to the local OS
2758 installation or hardware.
2760 * systemd-gpt-auto-generator can now automatically set up discoverable
2761 LUKS2 encrypted swap partitions.
2763 * The GPT Discoverable Partitions Specification has been substantially
2764 extended with support for root and /usr/ partitions for the majority
2765 of architectures systemd supports. This includes platforms that do
2766 not natively support UEFI, because even though GPT is specified under
2767 UEFI umbrella, it is useful on other systems too. Specifically,
2768 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
2769 Portable Services use the concept without requiring UEFI.
2771 * The GPT Discoverable Partitions Specifications has been extended with
2772 a new set of partitions that may carry PKCS#7 signatures for Verity
2773 partitions, encoded in a simple JSON format. This implements a simple
2774 mechanism for building disk images that are fully authenticated and
2775 can be tested against a set of cryptographic certificates. This is
2776 now implemented for the various systemd tools that can operate with
2777 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
2778 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
2779 The PKCS#7 signatures are passed to the kernel (where they are
2780 checked against certificates from the kernel keyring), or can be
2781 verified against certificates provided in userspace (via a simple
2782 drop-in file mechanism).
2784 * systemd-dissect's inspection logic will now report for which uses a
2785 disk image is intended. Specifically, it will display whether an
2786 image is suitable for booting on UEFI or in a container (using
2787 systemd-nspawn's --image= switch), whether it can be used as portable
2788 service, or attached as system extension.
2790 * The system-extension.d/ drop-in files now support a new field
2791 SYSEXT_SCOPE= that may encode which purpose a system extension image
2792 is for: one of "initrd", "system" or "portable". This is useful to
2793 make images more self-descriptive, and to ensure system extensions
2794 cannot be attached in the wrong contexts.
2796 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
2797 be used in portable service images to indicate which unit prefixes
2800 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
2801 now is able to decode images for non-native architectures as well.
2802 This allows systemd-nspawn to boot images of non-native architectures
2803 if the corresponding user mode emulator is installed and
2804 systemd-binfmtd is running.
2806 * systemd-logind gained new settings HandlePowerKeyLongPress=,
2807 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
2808 HandleHibernateKeyLongPress= which may be used to configure actions
2809 when the relevant keys are pressed for more than 5s. This is useful
2810 on devices that only have hardware for a subset of these keys. By
2811 default, if the reboot key is pressed long the poweroff operation is
2812 now triggered, and when the suspend key is pressed long the hibernate
2813 operation is triggered. Long pressing the other two keys currently
2814 does not trigger any operation by default.
2816 * When showing unit status updates on the console during boot and
2817 shutdown, and a service is slow to start so that the cylon animation
2818 is shown, the most recent sd_notify() STATUS= text is now shown as
2819 well. Services may use this to make the boot/shutdown output easier
2820 to understand, and to indicate what precisely a service that is slow
2821 to start or stop is waiting for. In particular, the per-user service
2822 manager instance now reports what it is doing and which service it is
2823 waiting for this way to the system service manager.
2825 * The service manager will now re-execute on reception of the
2826 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
2827 only when running as PID 1. There was no signal to request this when
2828 running as per-user service manager, i.e. as any other PID than 1.
2829 SIGRTMIN+25 works for both system and user managers.
2831 * The hardware watchdog logic in PID 1 gained support for operating
2832 with the default timeout configured in the hardware, instead of
2833 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
2834 request this behavior.
2836 * A new kernel command line option systemd.watchdog_sec= is now
2837 understood which may be used to override the hardware watchdog
2838 time-out for the boot.
2840 * A new setting DefaultOOMScoreAdjust= is now supported in
2841 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
2842 to set the default process OOM score adjustment value for processes
2843 started by the service manager. For per-user service managers this
2844 now defaults to 100, but for per-system service managers is left as
2845 is. This means that by default now services forked off the user
2846 service manager are more likely to be killed by the OOM killer than
2847 system services or the managers themselves.
2849 * A new per-service setting RestrictFileSystems= as been added that
2850 restricts the file systems a service has access to by their type.
2851 This is based on the new BPF LSM of the Linux kernel. It provides an
2852 effective way to make certain API file systems unavailable to
2853 services (and thus minimizing attack surface). A new command
2854 "systemd-analyze filesystems" has been added that lists all known
2855 file system types (and how they are grouped together under useful
2858 * Services now support a new setting RestrictNetworkInterfaces= for
2859 restricting access to specific network interfaces.
2861 * Service unit files gained new settings StartupAllowedCPUs= and
2862 StartupAllowedMemoryNodes=. These are similar to their counterparts
2863 without the "Startup" prefix and apply during the boot process
2864 only. This is useful to improve boot-time behavior of the system and
2865 assign resources differently during boot than during regular
2866 runtime. This is similar to the preexisting StartupCPUWeight=
2869 * Related to this: the various StartupXYZ= settings
2870 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
2871 during shutdown. The settings not prefixed with "Startup" hence apply
2872 during regular runtime, and those that are prefixed like that apply
2873 during boot and shutdown.
2875 * A new per-unit set of conditions/asserts
2876 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
2877 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
2878 pressure is above the configured threshold, using the kernel PSI
2879 feature. For more details see systemd.unit(5) and
2880 https://docs.kernel.org/accounting/psi.html
2882 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
2883 ProtectKernelLogs=yes can now be used.
2885 * The default maximum numbers of inodes have been raised from 64k to 1M
2886 for /dev/, and from 400k to 1M for /tmp/.
2888 * The per-user service manager learnt support for communicating with
2889 systemd-oomd to acquire OOM kill information.
2891 * A new service setting ExecSearchPath= has been added that allows
2892 changing the search path for executables for services. It affects
2893 where we look for the binaries specified in ExecStart= and similar,
2894 and the specified directories are also added the $PATH environment
2895 variable passed to invoked processes.
2897 * A new setting RuntimeRandomizedExtraSec= has been added for service
2898 and scope units that allows extending the runtime time-out as
2899 configured by RuntimeMaxSec= with a randomized amount.
2901 * The syntax of the service unit settings RuntimeDirectory=,
2902 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
2903 if the specified value is now suffixed with a colon, followed by
2904 another filename, the latter will be created as symbolic link to the
2905 specified directory. This allows creating these service directories
2906 together with alias symlinks to make them available under multiple
2909 * Service unit files gained two new settings TTYRows=/TTYColumns= for
2910 configuring rows/columns of the TTY device passed to
2911 stdin/stdout/stderr of the service. This is useful to propagate TTY
2912 dimensions to a virtual machine.
2914 * A new service unit file setting ExitType= has been added that
2915 specifies when to assume a service has exited. By default systemd
2916 only watches the main process of a service. By setting
2917 ExitType=cgroup it can be told to wait for the last process in a
2920 * Automount unit files gained a new setting ExtraOptions= that can be
2921 used to configure additional mount options to pass to the kernel when
2922 mounting the autofs instance.
2924 * "Urlification" (generation of ESC sequences that generate clickable
2925 hyperlinks in modern terminals) may now be turned off altogether
2928 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
2929 settings that default to 200 and 2 s respectively. The ratelimit
2930 ensures that a path unit cannot cause PID1 to busy-loop when it is
2931 trying to trigger a service that is skipped because of a Condition*=
2932 not being satisfied. This matches the configuration and behaviour of
2935 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
2936 as a plug-in for cryptsetup. This means the plain cryptsetup command
2937 may now be used to unlock volumes set up this way.
2939 * The TPM2 logic in cryptsetup will now automatically detect systems
2940 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
2941 updates the SHA1 banks. In such a case PCR policies will be
2942 automatically bound to the latter, not the former. This makes the PCR
2943 policies reliable, but of course do not provide the same level of
2944 trust as SHA256 banks.
2946 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
2947 RSA primary keys in addition to ECC, improving compatibility with
2948 TPM2 chips that do not support ECC. RSA keys are much slower to use
2949 than ECC, and hence are only used if ECC is not available.
2951 * /etc/crypttab gained support for a new token-timeout= setting for
2952 encrypted volumes that allows configuration of the maximum time to
2953 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
2954 the logic will query the user for a regular passphrase/recovery key
2957 * Support for activating dm-integrity volumes at boot via a new file
2958 /etc/integritytab and the tool systemd-integritysetup have been
2959 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
2960 with dm-integrity instead of dm-crypt/dm-verity.
2962 * The systemd-veritysetup-generator now understands a new usrhash=
2963 kernel command line option for specifying the Verity root hash for
2964 the partition backing the /usr/ file system. A matching set of
2965 systemd.verity_usr_* kernel command line options has been added as
2966 well. These all work similar to the corresponding options for the
2969 * The sd-device API gained a new API call sd_device_get_diskseq() to
2970 return the DISKSEQ property of a device structure. The "disk
2971 sequence" concept is a new feature recently introduced to the Linux
2972 kernel that allows detecting reuse cycles of block devices, i.e. can
2973 be used to recognize when loopback block devices are reused for a
2974 different purpose or CD-ROM drives get their media changed.
2976 * A new unit systemd-boot-update.service has been added. If enabled
2977 (the default) and the sd-boot loader is detected to be installed, it
2978 is automatically updated to the newest version when out of date. This
2979 is useful to ensure the boot loader remains up-to-date, and updates
2980 automatically propagate from the OS tree in /usr/.
2982 * sd-boot will now build with SBAT by default in order to facilitate
2983 working with recent versions of Shim that require it to be present.
2985 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
2986 This is used to robustly generate boot entry titles for Windows.
2988 * A new generic target unit factory-reset.target has been added. It is
2989 hooked into systemd-logind similar in fashion to
2990 reboot/poweroff/suspend/hibernate, and is supposed to be used to
2991 initiate a factory reset operation. What precisely this operation
2992 entails is up for the implementer to decide, the primary goal of the
2993 new unit is provide a framework where to plug in the implementation
2994 and how to trigger it.
2996 * A new meson build-time option 'clock-valid-range-usec-max' has been
2997 added which takes a time in µs and defaults to 15 years. If the RTC
2998 time is noticed to be more than the specified time ahead of the
2999 built-in epoch of systemd (which by default is the release timestamp
3000 of systemd) it is assumed that the RTC is not working correctly, and
3001 the RTC is reset to the epoch. (It already is reset to the epoch when
3002 noticed to be before it.) This should increase the chance that time
3003 doesn't accidentally jump too far ahead due to faulty hardware or
3006 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
3007 which may be used to automatically save the current system time to
3008 disk in regular intervals. This is useful to maintain a roughly
3009 monotonic clock even without RTC hardware and with some robustness
3010 against abnormal system shutdown.
3012 * systemd-analyze verify gained support for a pair of new --image= +
3013 --root= switches for verifying units below a specific root
3014 directory/image instead of on the host.
3016 * systemd-analyze verify gained support for verifying unit files under
3017 an explicitly specified unit name, independently of what the filename
3020 * systemd-analyze verify gained a new switch --recursive-errors= which
3021 controls whether to only fail on errors found in the specified units
3022 or recursively any dependent units.
3024 * systemd-analyze security now supports a new --offline mode for
3025 analyzing unit files stored on disk instead of loaded units. It may
3026 be combined with --root=/--image to analyze unit files under a root
3027 directory or disk image. It also learnt a new --threshold= parameter
3028 for specifying an exposure level threshold: if the exposure level
3029 exceeds the specified value the call will fail. It also gained a new
3030 --security-policy= switch for configuring security policies to
3031 enforce on the units. A policy is a JSON file that lists which tests
3032 shall be weighted how much to determine the overall exposure
3033 level. Altogether these new features are useful for fully automatic
3034 analysis and enforcement of security policies on unit files.
3036 * systemd-analyze security gain a new --json= switch for JSON output.
3038 * systemd-analyze learnt a new --quiet switch for reducing
3039 non-essential output. It's honored by the "dot", "syscall-filter",
3040 "filesystems" commands.
3042 * systemd-analyze security gained a --profile= option that can be used
3043 to take into account a portable profile when analyzing portable
3044 services, since a lot of the security-related settings are enabled
3047 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
3048 files, binaries and executables and prints metadata information,
3049 including the build-id and other info described on:
3050 https://systemd.io/COREDUMP_PACKAGE_METADATA/
3052 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
3053 section, for automatically propagating DNS settings from other
3056 * The static lease DHCP server logic in systemd-networkd may now serve
3057 IP addresses outside of the configured IP pool range for the server.
3059 * CAN support in systemd-networkd gained four new settings Loopback=,
3060 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
3061 control modes. It gained a number of further settings for tweaking
3064 * The [CAN] section in .network file gained new TimeQuantaNSec=,
3065 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
3066 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
3067 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
3068 DataSyncJumpWidth= settings to control bit-timing processed by the
3071 * DHCPv4 client support in systemd-networkd learnt a new Label= option
3072 for configuring the address label to apply to configure IPv4
3075 * The [IPv6AcceptRA] section of .network files gained support for a new
3076 UseMTU= setting that may be used to control whether to apply the
3077 announced MTU settings to the local interface.
3079 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
3080 setting to control whether the DHCPv4 client request and process the
3083 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
3084 [DHCPPrefixDelegation], as now the prefix delegation is also supported
3085 with DHCPv4 protocol by enabling the Use6RD= setting.
3087 * The [DHCPPrefixDelegation] section in .network file gained a new
3088 setting UplinkInterface= to specify the upstream interface.
3090 * The [DHCPv6] section in .network file gained a new setting
3091 UseDelegatedPrefix= to control whether the delegated prefixes will be
3092 propagated to the downstream interfaces.
3094 * The [IPv6AcceptRA] section of .network files now understands two new
3095 settings UseGateway=/UseRoutePrefix= for explicitly configuring
3096 whether to use the relevant fields from the IPv6 Router Advertisement
3099 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
3100 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
3101 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
3102 [IPv6AcceptRA] section to control when the DHCPv6 client is started
3103 and how the delegated prefixes are handled by the DHCPv6 client.
3105 * The IPv6Token= section in the [Network] section is deprecated, and
3106 the [IPv6AcceptRA] section gained the Token= setting for its
3107 replacement. The [IPv6Prefix] section also gained the Token= setting.
3108 The Token= setting gained 'eui64' mode to explicitly configure an
3109 address with the EUI64 algorithm based on the interface MAC address.
3110 The 'prefixstable' mode can now optionally take a secret key. The
3111 Token= setting in the [DHCPPrefixDelegation] section now supports all
3112 algorithms supported by the same settings in the other sections.
3114 * The [RoutingPolicyRule] section of .network file gained a new
3115 SuppressInterfaceGroup= setting.
3117 * The IgnoreCarrierLoss= setting in the [Network] section of .network
3118 files now allows a duration to be specified, controlling how long to
3119 wait before reacting to carrier loss.
3121 * The [DHCPServer] section of .network file gained a new Router=
3122 setting to specify the router address.
3124 * The [CAKE] section of .network files gained various new settings
3125 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
3126 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
3127 and UseRawPacketSize= for configuring CAKE.
3129 * systemd-networkd now ships with new default .network files:
3130 80-container-vb.network which matches host-side network bridge device
3131 created by systemd-nspawn's --network-bridge or --network-zone
3132 switch, and 80-6rd-tunnel.network which matches automatically created
3133 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
3135 * systemd-networkd's handling of Endpoint= resolution for WireGuard
3136 interfaces has been improved.
3138 * systemd-networkd will now automatically configure routes to addresses
3139 specified in AllowedIPs=. This feature can be controlled via
3140 RouteTable= and RouteMetric= settings in [WireGuard] or
3141 [WireGuardPeer] sections.
3143 * systemd-networkd will now once again automatically generate persistent
3144 MAC addresses for batadv and bridge interfaces. Users can disable this
3145 by using MACAddress=none in .netdev files.
3147 * systemd-networkd and systemd-udevd now support IP over InfiniBand
3148 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
3149 systemd.netdev files gained the [IPoIB] section.
3151 * systemd-networkd and systemd-udevd now support net.ifname-policy=
3152 option on the kernel command-line. This is implemented through the
3153 systemd-network-generator service that automatically generates
3154 appropriate .link, .network, and .netdev files.
3156 * The various systemd-udevd "ethtool" buffer settings now understand
3157 the special value "max" to configure the buffers to the maximum the
3160 * systemd-udevd's .link files may now configure a large variety of
3161 NIC coalescing settings, plus more hardware offload settings.
3163 * .link files gained a new WakeOnLanPassword= setting in the [Link]
3164 section that allows to specify a WoL "SecureOn" password on hardware
3167 * systemd-nspawn's --setenv= switch now supports an additional syntax:
3168 if only a variable name is specified (i.e. without being suffixed by
3169 a '=' character and a value) the current value of the environment
3170 variable is propagated to the container. e.g. --setenv=FOO will
3171 lookup the current value of $FOO in the environment, and pass it down
3172 to the container. Similar behavior has been added to homectl's,
3173 machinectl's and systemd-run's --setenv= switch.
3175 * systemd-nspawn gained a new switch --suppress-sync= which may be used
3176 to optionally suppress the effect of the sync()/fsync()/fdatasync()
3177 system calls for the container payload. This is useful for build
3178 system environments where safety against abnormal system shutdown is
3179 not essential as all build artifacts can be regenerated any time, but
3180 the performance win is beneficial.
3182 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
3183 same value that PID 1 uses for most forked off processes.
3185 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
3186 uidmap/nouidmap options as last parameter. If "uidmap" is used the
3187 bind mounts are created with UID mapping taking place that ensures
3188 the host's file ownerships are mapped 1:1 to container file
3189 ownerships, even if user namespacing is used. This way
3190 files/directories bound into containers will no longer show up as
3191 owned by the nobody user as they typically did if no special care was
3192 taken to shift them manually.
3194 * When discovering Windows installations sd-boot will now attempt to
3195 show the Windows version.
3197 * The color scheme to use in sd-boot may now be configured at
3200 * sd-boot gained the ability to change screen resolution during
3201 boot-time, by hitting the "r" key. This will cycle through available
3202 resolutions and save the last selection.
3204 * sd-boot learnt a new hotkey "f". When pressed the system will enter
3205 firmware setup. This is useful in environments where it is difficult
3206 to hit the right keys early enough to enter the firmware, and works
3207 on any firmware regardless which key it natively uses.
3209 * sd-boot gained support for automatically booting into the menu item
3210 selected on the last boot (using the "@saved" identifier for menu
3213 * sd-boot gained support for automatically loading all EFI drivers
3214 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
3215 Partition (ESP). These drivers are loaded before the menu entries are
3216 loaded. This is useful e.g. to load additional file system drivers
3217 for the XBOOTLDR partition.
3219 * systemd-boot will now paint the input cursor on its own instead of
3220 relying on the firmware to do so, increasing compatibility with broken
3221 firmware that doesn't make the cursor reasonably visible.
3223 * sd-boot now embeds a .osrel PE section like we expect from Boot
3224 Loader Specification Type #2 Unified Kernels. This means sd-boot
3225 itself may be used in place of a Type #2 Unified Kernel. This is
3226 useful for debugging purposes as it allows chain-loading one a
3227 (development) sd-boot instance from another.
3229 * sd-boot now supports a new "devicetree" field in Boot Loader
3230 Specification Type #1 entries: if configured the specified device
3231 tree file is installed before the kernel is invoked. This is useful
3232 for installing/applying new devicetree files without updating the
3235 * Similarly, sd-stub now can read devicetree data from a PE section
3236 ".dtb" and apply it before invoking the kernel.
3238 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
3239 gained the ability to pick up credentials and sysext files, wrap them
3240 in a cpio archive, and pass as an additional initrd to the invoked
3241 Linux kernel, in effect placing those files in the /.extra/ directory
3242 of the initrd environment. This is useful to implement trusted initrd
3243 environments which are fully authenticated but still can be extended
3244 (via sysexts) and parameterized (via encrypted/authenticated
3245 credentials, see above).
3247 Credentials can be located next to the kernel image file (credentials
3248 specific to a single boot entry), or in one of the shared directories
3249 (credentials applicable to multiple boot entries).
3251 * sd-stub now comes with a full man page, that explains its feature set
3252 and how to combine a kernel image, an initrd and the stub to build a
3253 complete EFI unified kernel image, implementing Boot Loader
3254 Specification Type #2.
3256 * sd-stub may now provide the initrd to the executed kernel via the
3257 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
3258 non-x86 architectures.
3260 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
3261 may be used to set the boot menu time-out of the boot loader (for all
3262 or just the subsequent boot).
3264 * bootctl and kernel-install will now read variables
3265 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
3266 /etc/kernel/install.conf. When set, it specifies the layout to use
3267 for installation directories on the boot partition, so that tools
3268 don't need to guess it based on the already-existing directories. The
3269 only value that is defined natively is "bls", corresponding to the
3271 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
3272 kernel-install that implement a different layout can declare other
3273 values for this variable.
3275 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
3276 assumption that if the user installed sd-boot to the ESP, they intend
3277 to use the entry layout understood by sd-boot. It'll also write
3278 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
3279 (and it wasn't specified in the config file yet). Similarly,
3280 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
3281 wasn't specified in the config file yet). Effectively, those changes
3282 mean that the machine-id used for boot loader entry installation is
3283 "frozen" upon first use and becomes independent of the actual
3286 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
3287 images created for distribution ("golden images") are built with no
3288 machine-id, so that a unique machine-id can be created on the first
3289 boot. But those images may contain boot loader entries with the
3290 machine-id used during build included in paths. Using a "frozen"
3291 value allows unambiguously identifying entries that match the
3292 specific installation, while still permitting parallel installations
3295 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
3296 kernel-install to guess the installation layout. This fixes the
3297 problem where a (possibly empty) directory in the boot partition is
3298 created from a different layout causing kernel-install plugins to
3299 assume the wrong layout. A particular example of how this may happen
3300 is the grub2 package in Fedora which includes directories under /boot
3301 directly in its file list. Various other packages pull in grub2 as a
3302 dependency, so it may be installed even if unused, breaking
3303 installations that use the bls layout.
3305 * bootctl and systemd-bless-boot can now be linked statically.
3307 * systemd-sysext now optionally doesn't insist on extension-release.d/
3308 files being placed in the image under the image's file name. If the
3309 file system xattr user.extension-release.strict is set on the
3310 extension release file, it is accepted regardless of its name. This
3311 relaxes security restrictions a bit, as system extension may be
3312 attached under a wrong name this way.
3314 * udevadm's test-builtin command learnt a new --action= switch for
3315 testing the built-in with the specified action (in place of the
3318 * udevadm info gained new switches --property=/--value for showing only
3319 specific udev properties/values instead of all.
3321 * A new hwdb database has been added that contains matches for various
3322 types of signal analyzers (protocol analyzers, logic analyzers,
3323 oscilloscopes, multimeters, bench power supplies, etc.) that should
3324 be accessible to regular users.
3326 * A new hwdb database entry has been added that carries information
3327 about types of cameras (regular or infrared), and in which direction
3328 they point (front or back).
3330 * A new rule to allow console users access to rfkill by default has been
3333 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
3334 now also owned by the system group "sgx".
3336 * A new build-time meson option "extra-net-naming-schemes=" has been
3337 added to define additional naming schemes for udev's network
3338 interface naming logic. This is useful for enterprise distributions
3339 and similar which want to pin the schemes of certain distribution
3340 releases under a specific name and previously had to patch the
3341 sources to introduce new named schemes.
3343 * The predictable naming logic for network interfaces has been extended
3344 to generate stable names from Xen netfront device information.
3346 * hostnamed's chassis property can now be sourced from chassis-type
3347 field encoded in devicetree (in addition to the existing DMI
3350 * systemd-cgls now optionally displays cgroup IDs and extended
3351 attributes for each cgroup. (Controllable via the new --xattr= +
3352 --cgroup-id= switches.)
3354 * coredumpctl gained a new --all switch for operating on all
3355 Journal files instead of just the local ones.
3357 * systemd-coredump will now use libdw/libelf via dlopen() rather than
3358 directly linking, allowing users to easily opt-out of backtrace/metadata
3359 analysis of core files, and reduce image sizes when this is not needed.
3361 * systemd-coredump will now analyze core files with libdw/libelf in a
3362 forked, sandboxed process.
3364 * systemd-homed will now try to unmount an activate home area in
3365 regular intervals once the user logged out fully. Previously this was
3366 attempted exactly once but if the home directory was busy for some
3367 reason it was not tried again.
3369 * systemd-homed's LUKS2 home area backend will now create a BSD file
3370 system lock on the image file while the home area is active
3371 (i.e. mounted). If a home area is found to be locked, logins are
3372 politely refused. This should improve behavior when using home areas
3373 images that are accessible via the network from multiple clients, and
3374 reduce the chance of accidental file system corruption in that case.
3376 * Optionally, systemd-homed will now drop the kernel buffer cache once
3377 a user has fully logged out, configurable via the new --drop-caches=
3380 * systemd-homed now makes use of UID mapped mounts for the home areas.
3381 If the kernel and used file system support it, files are now
3382 internally owned by the "nobody" user (i.e. the user typically used
3383 for indicating "this ownership is not mapped"), and dynamically
3384 mapped to the UID used locally on the system via the UID mapping
3385 mount logic of recent kernels. This makes migrating home areas
3386 between different systems cheaper because recursively chown()ing file
3387 system trees is no longer necessary.
3389 * systemd-homed's CIFS backend now optionally supports CIFS service
3390 names with a directory suffix, in order to place home directories in
3391 a subdirectory of a CIFS share, instead of the top-level directory.
3393 * systemd-homed's CIFS backend gained support for specifying additional
3394 mount options in the JSON user record (cifsExtraMountOptions field,
3395 and --cifs-extra-mount-options= homectl switch). This is for example
3396 useful for configuring mount options such as "noserverino" that some
3397 SMB3 services require (use that to run a homed home directory from a
3398 FritzBox SMB3 share this way).
3400 * systemd-homed will now default to btrfs' zstd compression for home
3401 areas. This is inspired by Fedora's recent decision to switch to zstd
3404 * Additional mount options to use when mounting the file system of
3405 LUKS2 volumes in systemd-homed has been added. Via the
3406 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
3407 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
3408 systemd-homed or via the luksExtraMountOptions user record JSON
3409 property. (Exposed via homectl --luks-extra-mount-options)
3411 * homectl's resize command now takes the special size specifications
3412 "min" and "max" to shrink/grow the home area to the minimum/maximum
3413 size possible, taking disk usage/space constraints and file system
3414 limitations into account. Resizing is now generally graceful: the
3415 logic will try to get as close to the specified size as possible, but
3416 not consider it a failure if the request couldn't be fulfilled
3419 * systemd-homed gained the ability to automatically shrink home areas
3420 on logout to their minimal size and grow them again on next
3421 login. This ensures that while inactive, a home area only takes up
3422 the minimal space necessary, but once activated, it provides
3423 sufficient space for the user's needs. This behavior is only
3424 supported if btrfs is used as file system inside the home area
3425 (because only for btrfs online growing/shrinking is implemented in
3426 the kernel). This behavior is now enabled by default, but may be
3427 controlled via the new --auto-resize-mode= setting of homectl.
3429 * systemd-homed gained support for automatically re-balancing free disk
3430 space among active home areas, in case the LUKS2 backends are used,
3431 and no explicit disk size was requested. This way disk space is
3432 automatically managed and home areas resized in regular intervals and
3433 manual resizing when disk space becomes scarce should not be
3434 necessary anymore. This behavior is only supported if btrfs is used
3435 within the home areas (as only then online shrinking and growing is
3436 supported), and may be configured via the new rebalanceWeight JSON
3437 user record field (as exposed via the new --rebalance-weight= homectl
3438 setting). Re-balancing is mostly automatic, but can also be requested
3439 explicitly via "homectl rebalance", which is synchronous, and thus
3440 may be used to wait until the rebalance run is complete.
3442 * userdbctl gained a --json= switch for configured the JSON formatting
3443 to use when outputting user or group records.
3445 * userdbctl gained a new --multiplexer= switch for explicitly
3446 configuring whether to use the systemd-userdbd server side user
3447 record resolution logic.
3449 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
3450 for chaining up another command to execute after completing the
3451 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
3452 configuration of a single command to invoke, this maybe used to
3453 invoke multiple: first userdbctl's own implementation, and then any
3454 other also configured in the command line.
3456 * The sd-event API gained a new function sd_event_add_inotify_fd() that
3457 is similar to sd_event_add_inotify() but accepts a file descriptor
3458 instead of a path in the file system for referencing the inode to
3461 * The sd-event API gained a new function
3462 sd_event_source_set_ratelimit_expire_callback() that may be used to
3463 define a callback function that is called whenever an event source
3464 leaves the rate limiting phase.
3466 * New documentation has been added explaining which steps are necessary
3467 to port systemd to a new architecture:
3469 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
3471 * The x-systemd.makefs option in /etc/fstab now explicitly supports
3472 ext2, ext3, and f2fs file systems.
3474 * Mount units and units generated from /etc/fstab entries with 'noauto'
3475 are now ordered the same as other units. Effectively, they will be
3476 started earlier (if something actually pulled them in) and stopped
3477 later, similarly to normal mount units that are part of
3478 fs-local.target. This change should be invisible to users, but
3479 should prevent those units from being stopped too early during
3482 * The systemd-getty-generator now honors a new kernel command line
3483 argument systemd.getty_auto= and a new environment variable
3484 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
3485 example useful to turn off gettys inside of containers or similar
3488 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
3489 (in addition to 127.0.0.53, as before). If DNS requests are sent to
3490 this address they are propagated in "bypass" mode only, i.e. are
3491 almost not processed locally, but mostly forwarded as-is to the
3492 current upstream DNS servers. This provides a stable DNS server
3493 address that proxies all requests dynamically to the right upstream
3494 DNS servers even if these dynamically change. This stub does not do
3495 mDNS/LLMNR resolution. However, it will translate look-ups to
3496 DNS-over-TLS if necessary. This new stub is particularly useful in
3497 container/VM environments, or for tethering setups: use DNAT to
3498 redirect traffic to any IP address to this stub.
3500 * systemd-importd now honors new environment variables
3501 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
3502 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
3503 generation, btrfs quota setup and disk synchronization.
3505 * systemd-importd and systemd-resolved can now be optionally built with
3506 OpenSSL instead of libgcrypt.
3508 * systemd-repart no longer requires OpenSSL.
3510 * systemd-sysusers will no longer create the redundant 'nobody' group
3511 by default, as the 'nobody' user is already created with an
3512 appropriate primary group.
3514 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
3516 * systemctl show-environment gained support for --output=json.
3518 * pam_systemd will now first try to use the X11 abstract socket, and
3519 fallback to the socket file in /tmp/.X11-unix/ only if that does not
3522 * systemd-journald will no longer go back to volatile storage
3523 regardless of configuration when its unit is restarted.
3525 * Initial support for the LoongArch architecture has been added (system
3526 call lists, GPT partition table UUIDs, etc).
3528 * systemd-journald's own logging messages are now also logged to the
3529 journal itself when systemd-journald logs to /dev/kmsg.
3531 * systemd-journald now re-enables COW for archived journal files on
3532 filesystems that support COW. One benefit of this change is that
3533 archived journal files will now get compressed on btrfs filesystems
3534 that have compression enabled.
3536 * systemd-journald now deduplicates fields in a single log message
3537 before adding it to the journal. In archived journal files, it will
3538 also punch holes for unused parts and truncate the file as
3539 appropriate, leading to reductions in disk usage.
3541 * journalctl --verify was extended with more informative error
3544 * More of sd-journal's functions are now resistant against journal file
3547 * The shutdown command learnt a new option --show, to display the
3550 * A LICENSES/ directory is now included in the git tree. It contains a
3551 README.md file that explains the licenses used by source files in
3552 this repository. It also contains the text of all applicable
3553 licenses as they appear on spdx.org.
3555 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
3556 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
3557 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
3558 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
3559 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
3560 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
3561 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
3562 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
3563 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
3564 Christian Brauner, Christian Göttsche, Christian Wehrli,
3565 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
3566 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
3567 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
3568 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
3569 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
3570 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
3571 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
3572 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
3573 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
3574 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
3575 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
3576 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
3577 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
3578 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
3579 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
3580 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
3581 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
3582 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
3583 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
3584 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
3585 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
3586 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
3587 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
3588 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
3589 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
3590 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
3591 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
3592 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
3593 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
3594 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
3595 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
3596 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
3597 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
3598 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
3599 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
3600 Дамјан Георгиевски, наб
3602 — Warsaw, 2021-12-23
3606 * When operating on disk images via the --image= switch of various
3607 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
3608 no 'root=' parameter on the kernel command line, and multiple
3609 suitable root or /usr/ partitions exist in the image, then a simple
3610 comparison inspired by strverscmp() is done on the GPT partition
3611 label, and the newest partition is picked. This permits a simple and
3612 generic whole-file-system A/B update logic where new operating system
3613 versions are dropped into partitions whose label is then updated with
3614 a matching version identifier.
3616 * systemd-sysusers now supports querying the passwords to set for the
3617 users it creates via the "credentials" logic introduced in v247: the
3618 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
3619 credentials are consulted for the password to use (either in UNIX
3620 hashed form, or literally). By default these credentials are inherited
3621 down from PID1 (which in turn imports it from a container manager if
3622 there is one). This permits easy configuration of user passwords
3623 during first boot. Example:
3625 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
3627 Note that systemd-sysusers operates in purely additive mode: it
3628 executes no operation if the declared users already exist, and hence
3629 doesn't set any passwords as effect of the command line above if the
3630 specified root user exists already in the image. (Note that
3631 --volatile=yes ensures it doesn't, though.)
3633 * systemd-firstboot now also supports querying various system
3634 parameters via the credential subsystems. Thus, as above this may be
3635 used to initialize important system parameters on first boot of
3636 previously unprovisioned images (i.e. images with a mostly empty
3639 * PID 1 may now show both the unit name and the unit description
3640 strings in its status output during boot. This may be configured with
3641 StatusUnitFormat=combined in system.conf or
3642 systemd.status-unit-format=combined on the kernel command line.
3644 * The systemd-machine-id-setup tool now supports a --image= switch for
3645 provisioning a machine ID file into an OS disk image, similar to how
3646 --root= operates on an OS file tree. This matches the existing switch
3647 of the same name for systemd-tmpfiles, systemd-firstboot, and
3648 systemd-sysusers tools.
3650 * Similarly, systemd-repart gained support for the --image= switch too.
3651 In combination with the existing --size= option, this makes the tool
3652 particularly useful for easily growing disk images in a single
3653 invocation, following the declarative rules included in the image
3656 * systemd-repart's partition configuration files gained support for a
3657 new switch MakeDirectories= which may be used to create arbitrary
3658 directories inside file systems that are created, before registering
3659 them in the partition table. This is useful in particular for root
3660 partitions to create mount point directories for other partitions
3661 included in the image. For example, a disk image that contains a
3662 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
3663 create /home/ and /var/ as empty directories in the root file system
3664 on its creation, so that the resulting image can be mounted
3665 immediately, even in read-only mode.
3667 * systemd-repart's CopyBlocks= setting gained support for the special
3668 value "auto". If used, a suitable matching partition on the booted OS
3669 is found as source to copy blocks from. This is useful when
3670 implementing replicating installers, that are booted from one medium
3671 and then stream their own root partition onto the target medium.
3673 * systemd-repart's partition configuration files gained support for a
3674 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
3675 GPT partition flags for the created partitions: this is useful for
3676 marking newly created partitions as read-only, or as not being
3677 subject for automatic mounting from creation on.
3679 * The /etc/os-release file has been extended with two new (optional)
3680 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
3681 information for OS images that are updated comprehensively and
3682 atomically as one image. Two new specifiers %M, %A now resolve to
3683 these two fields in the various configuration options that resolve
3686 * portablectl gained a new switch --extension= for enabling portable
3687 service images with extensions that follow the extension image
3688 concept introduced with v248, and thus allows layering multiple
3689 images when setting up the root filesystem of the service.
3691 * systemd-coredump will now extract ELF build-id information from
3692 processes dumping core and include it in the coredump report.
3693 Moreover, it will look for ELF .note.package sections with
3694 distribution packaging meta-information about the crashing process.
3695 This is useful to directly embed the rpm or deb (or any other)
3696 package name and version in ELF files, making it easy to match
3697 coredump reports with the specific package for which the software was
3698 compiled. This is particularly useful on environments with ELF files
3699 from multiple vendors, different distributions and versions, as is
3700 common today in our containerized and sand-boxed world. For further
3703 https://systemd.io/COREDUMP_PACKAGE_METADATA
3705 * A new udev hardware database has been added for FireWire devices
3708 * The "net_id" built-in of udev has been updated with three
3709 backwards-incompatible changes:
3711 - PCI hotplug slot names on s390 systems are now parsed as
3712 hexadecimal numbers. They were incorrectly parsed as decimal
3713 previously, or ignored if the name was not a valid decimal
3716 - PCI onboard indices up to 65535 are allowed. Previously, numbers
3717 above 16383 were rejected. This primarily impacts s390 systems,
3718 where values up to 65535 are used.
3720 - Invalid characters in interface names are replaced with "_".
3722 The new version of the net naming scheme is "v249". The previous
3723 scheme can be selected via the "net.naming-scheme=v247" kernel
3724 command line parameter.
3726 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
3727 NULL bus object, for which they will return false. Or in other words,
3728 an unallocated bus connection is neither ready nor open.
3730 * The sd-device API acquired a new API function
3731 sd_device_get_usec_initialized() that returns the monotonic time when
3732 the udev device first appeared in the database.
3734 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
3735 sd_device_get_trigger_uuid(). The former is similar to
3736 sd_device_trigger() but returns a randomly generated UUID that is
3737 associated with the synthetic uevent generated by the call. This UUID
3738 may be read from the sd_device object a monitor eventually receives,
3739 via the sd_device_get_trigger_uuid(). This interface requires kernel
3740 4.13 or above to work, and allows tracking a synthetic uevent through
3741 the entire device management stack. The "udevadm trigger --settle"
3742 logic has been updated to make use of this concept if available to
3743 wait precisely for the uevents it generates. "udevadm trigger" also
3744 gained a new parameter --uuid that prints the UUID for each generated
3747 * sd-device also gained new APIs sd_device_new_from_ifname() and
3748 sd_device_new_from_ifindex() for allocating an sd-device object for
3749 the specified network interface. The former accepts an interface name
3750 (either a primary or an alternative name), the latter an interface
3753 * The native Journal protocol has been documented. Clients may talk
3754 this as alternative to the classic BSD syslog protocol for locally
3755 delivering log records to the Journal. The protocol has been stable
3756 for a long time and in fact been implemented already in a variety
3757 of alternative client libraries. This documentation makes the support
3760 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
3762 * A new BPFProgram= setting has been added to service files. It may be
3763 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
3764 file, or a bind mount or symlink to one. This may be used to upload
3765 and manage BPF programs externally and then hook arbitrary systemd
3768 * The "home.arpa" domain that has been officially declared as the
3769 choice for domain for local home networks per RFC 8375 has been added
3770 to the default NTA list of resolved, since DNSSEC is generally not
3771 available on private domains.
3773 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
3775 * A new ManageForeignRoutingPolicyRules= setting has been added to
3776 .network files which may be used to exclude foreign-created routing
3777 policy rules from systemd-networkd management.
3779 * systemd-network-wait-online gained two new switches -4 and -6 that
3780 may be used to tweak whether to wait for only IPv4 or only IPv6
3783 * .network files gained a new RequiredFamilyForOnline= setting to
3784 fine-tune whether to require an IPv4 or IPv6 address in order to
3785 consider an interface "online".
3787 * networkctl will now show an over-all "online" state in the per-link
3790 * In .network files a new OutgoingInterface= setting has been added to
3791 specify the output interface in bridge FDB setups.
3793 * In .network files the Multipath group ID may now be configured for
3794 [NextHop] entries, via the new Group= setting.
3796 * The DHCP server logic configured in .network files gained a new
3797 setting RelayTarget= that turns the server into a DHCP server relay.
3798 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
3799 to further tweak the DHCP relay behaviour.
3801 * The DHCP server logic also gained a new ServerAddress= setting in
3802 .network files that explicitly specifies the server IP address to
3803 use. If not specified, the address is determined automatically, as
3806 * The DHCP server logic in systemd-networkd gained support for static
3807 DHCP leases, configurable via the [DHCPServerStaticLease]
3808 section. This allows explicitly mapping specific MAC addresses to
3809 fixed IP addresses and vice versa.
3811 * The RestrictAddressFamilies= setting in service files now supports a
3812 new special value "none". If specified sockets of all address
3813 families will be made unavailable to services configured that way.
3815 * systemd-fstab-generator and systemd-repart have been updated to
3816 support booting from disks that carry only a /usr/ partition but no
3817 root partition yet, and where systemd-repart can add it in on the
3818 first boot. This is useful for implementing systems that ship with a
3819 single /usr/ file system, and whose root file system shall be set up
3820 and formatted on a LUKS-encrypted volume whose key is generated
3821 locally (and possibly enrolled in the TPM) during the first boot.
3823 * The [Address] section of .network files now accepts a new
3824 RouteMetric= setting that configures the routing metric to use for
3825 the prefix route created as effect of the address configuration.
3826 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
3827 gained matching settings for their prefix routes. (The option of the
3828 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
3829 it conceptually belongs there; the old option is still understood for
3832 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
3835 * A new udev property ID_NET_DHCP_BROADCAST on network interface
3836 devices is now honoured by systemd-networkd, controlling whether to
3837 issue DHCP offers via broadcasting. This is used to ensure that s390
3838 layer 3 network interfaces work out-of-the-box with systemd-networkd.
3840 * nss-myhostname and systemd-resolved will now synthesize address
3841 records for a new special hostname "_outbound". The name will always
3842 resolve to the local IP addresses most likely used for outbound
3843 connections towards the default routes. On multi-homed hosts this is
3844 useful to have a stable handle referring to "the" local IP address
3845 that matters most, to the point where this is defined.
3847 * The Discoverable Partition Specification has been updated with a new
3848 GPT partition flag "grow-file-system" defined for its partition
3849 types. Whenever partitions with this flag set are automatically
3850 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
3851 of systemd-nspawn or other tools; and as opposed to explicit mounting
3852 via /etc/fstab), the file system within the partition is
3853 automatically grown to the full size of the partition. If the file
3854 system size already matches the partition size this flag has no
3855 effect. Previously, this functionality has been available via the
3856 explicit x-systemd.growfs mount option, and this new flag extends
3857 this to automatically discovered mounts. A new GrowFileSystem=
3858 setting has been added to systemd-repart drop-in files that allows
3859 configuring this partition flag. This new flag defaults to on for
3860 partitions automatically created by systemd-repart, except if they
3861 are marked read-only. See the specification for further details:
3863 https://systemd.io/DISCOVERABLE_PARTITIONS
3865 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
3866 section. If enabled (which is the default), and an NTP server address
3867 is acquired through a DHCP lease on this interface an explicit route
3868 to this address is created on this interface to ensure that NTP
3869 traffic to the NTP server acquired on an interface is also routed
3870 through that interface. The pre-existing RoutesToDNS= setting that
3871 implements the same for DNS servers is now enabled by default.
3873 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
3874 been added that may be used to restrict the network interfaces
3875 sockets created by the service may be bound to. This is implemented
3878 * A new ConditionFirmware= setting has been added to unit files to
3879 conditionalize on certain firmware features. At the moment it may
3880 check whether running on a UEFI system, a device.tree system, or if
3881 the system is compatible with some specified device-tree feature.
3883 * A new ConditionOSRelease= setting has been added to unit files to
3884 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
3885 operators may be used to check if some field has some specific value
3886 or do an alphanumerical comparison. Equality comparisons are useful
3887 for fields like ID, but relative comparisons for fields like
3888 VERSION_ID or IMAGE_VERSION.
3890 * hostnamed gained a new Describe() D-Bus method that returns a JSON
3891 serialization of the host data it exposes. This is exposed via
3892 "hostnamectl --json=" to acquire a host identity description in JSON.
3893 It's our intention to add a similar features to most services and
3894 objects systemd manages, in order to simplify integration with
3895 program code that can consume JSON.
3897 * Similarly, networkd gained a Describe() method on its Manager and
3898 Link bus objects. This is exposed via "networkctl --json=".
3900 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
3901 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
3902 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
3903 that is used both to get the value (when no argument is given), and
3904 to set the value (when an argument is specified). The old names
3905 continue to be supported for compatibility.
3907 * systemd-detect-virt and ConditionVirtualization= are now able to
3908 correctly identify Amazon EC2 environments.
3910 * The LogLevelMax= setting of unit files now applies not only to log
3911 messages generated *by* the service, but also to log messages
3912 generated *about* the service by PID 1. To suppress logs concerning a
3913 specific service comprehensively, set this option to a high log
3916 * bootctl gained support for a new --make-machine-id-directory= switch
3917 that allows precise control on whether to create the top-level
3918 per-machine directory in the boot partition that typically contains
3919 Type 1 boot loader entries.
3921 * During build SBAT data to include in the systemd-boot EFI PE binaries
3922 may be specified now.
3924 * /etc/crypttab learnt a new option "headless". If specified any
3925 requests to query the user interactively for passwords or PINs will
3926 be skipped. This is useful on systems that are headless, i.e. where
3927 an interactive user is generally not present.
3929 * /etc/crypttab also learnt a new option "password-echo=" that allows
3930 configuring whether the encryption password prompt shall echo the
3931 typed password and if so, do so literally or via asterisks. (The
3932 default is the same behaviour as before: provide echo feedback via
3935 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
3936 systemd-homed has been updated to allow explicit configuration of the
3937 "user presence" and "user verification" checks, as well as whether a
3938 PIN is required for authentication, via the new switches
3939 --fido2-with-user-presence=, --fido2-with-user-verification=,
3940 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
3941 features are available, and may be enabled or disabled depends on the
3944 * systemd-nspawn's --private-user= switch now accepts the special value
3945 "identity" which configures a user namespacing environment with an
3946 identity mapping of 65535 UIDs. This means the container UID 0 is
3947 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
3948 this doesn't appear to be useful, however it does reduce the attack
3949 surface a bit, since the resulting container will possess process
3950 capabilities only within its namespace and not on the host.
3952 * systemd-nspawn's --private-user-chown switch has been replaced by a
3953 more generic --private-user-ownership= switch that accepts one of
3954 three values: "chown" is equivalent to the old --private-user-chown,
3955 and "off" is equivalent to the absence of the old switch. The value
3956 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
3957 of files and directories of the underlying image to the chosen UID
3958 range for the container. "auto" is equivalent to "map" if UID mapping
3959 mount are supported, otherwise it is equivalent to "chown". The short
3960 -U switch systemd-nspawn now implies --private-user-ownership=auto
3961 instead of the old --private-user-chown. Effectively this means: if
3962 the backing file system supports UID mapping mounts the feature is
3963 now used by default if -U is used. Generally, it's a good idea to use
3964 UID mapping mounts instead of recursive chown()ing, since it allows
3965 running containers off immutable images (since no modifications of
3966 the images need to take place), and share images between multiple
3967 instances. Moreover, the recursive chown()ing operation is slow and
3968 can be avoided. Conceptually it's also a good thing if transient UID
3969 range uses do not leak into persistent file ownership anymore. TLDR:
3970 finally, the last major drawback of user namespacing has been
3971 removed, and -U should always be used (unless you use btrfs, where
3972 UID mapped mounts do not exist; or your container actually needs
3973 privileges on the host).
3975 * nss-systemd now synthesizes user and group shadow records in addition
3976 to the main user and group records. Thus, hashed passwords managed by
3977 systemd-homed are now accessible via the shadow database.
3979 * The userdb logic (and thus nss-systemd, and so on) now read
3980 additional user/group definitions in JSON format from the drop-in
3981 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
3982 /usr/lib/userdb/. This is a simple and powerful mechanism for making
3983 additional users available to the system, with full integration into
3984 NSS including the shadow databases. Since the full JSON user/group
3985 record format is supported this may also be used to define users with
3986 resource management settings and other runtime settings that
3987 pam_systemd and systemd-logind enforce at login.
3989 * The userdbctl tool gained two new switches --with-dropin= and
3990 --with-varlink= which can be used to fine-tune the sources used for
3991 user database lookups.
3993 * systemd-nspawn gained a new switch --bind-user= for binding a host
3994 user account into the container. This does three things: the user's
3995 home directory is bind mounted from the host into the container,
3996 below the /run/userdb/home/ hierarchy. A free UID is picked in the
3997 container, and a user namespacing UID mapping to the host user's UID
3998 installed. And finally, a minimal JSON user and group record (along
3999 with its hashed password) is dropped into /run/host/userdb/. These
4000 records are picked up automatically by the userdb drop-in logic
4001 describe above, and allow the user to login with the same password as
4002 on the host. Effectively this means: if host and container run new
4003 enough systemd versions making a host user available to the container
4004 is trivially simple.
4006 * systemd-journal-gatewayd now supports the switches --user, --system,
4007 --merge, --file= that are equivalent to the same switches of
4008 journalctl, and permit exposing only the specified subset of the
4011 * The OnFailure= dependency between units is now augmented with a
4012 implicit reverse dependency OnFailureOf= (this new dependency cannot
4013 be configured directly it's only created as effect of an OnFailure=
4014 dependency in the reverse order — it's visible in "systemctl show"
4015 however). Similar, Slice= now has an reverse dependency SliceOf=,
4016 that is also not configurable directly, but useful to determine all
4017 units that are members of a slice.
4019 * A pair of new dependency types between units PropagatesStopTo= +
4020 StopPropagatedFrom= has been added, that allows propagation of unit
4021 stop events between two units. It operates similar to the existing
4022 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
4024 * A new dependency type OnSuccess= has been added (plus the reverse
4025 dependency OnSuccessOf=, which cannot be configured directly, but
4026 exists only as effect of the reverse OnSuccess=). It is similar to
4027 OnFailure=, but triggers in the opposite case: when a service exits
4028 cleanly. This allows "chaining up" of services where one or more
4029 services are started once another service has successfully completed.
4031 * A new dependency type Upholds= has been added (plus the reverse
4032 dependency UpheldBy=, which cannot be configured directly, but exists
4033 only as effect of Upholds=). This dependency type is a stronger form
4034 of Wants=: if a unit has an UpHolds= dependency on some other unit
4035 and the former is active then the latter is started whenever it is
4036 found inactive (and no job is queued for it). This is an alternative
4037 to Restart= inside service units, but less configurable, and the
4038 request to uphold a unit is not encoded in the unit itself but in
4039 another unit that intends to uphold it.
4041 * The systemd-ask-password tool now also supports reading passwords
4042 from the credentials subsystem, via the new --credential= switch.
4044 * The systemd-ask-password tool learnt a new switch --emoji= which may
4045 be used to explicit control whether the lock and key emoji (🔐) is
4046 shown in the password prompt on suitable TTYs.
4048 * The --echo switch of systemd-ask-password now optionally takes a
4049 parameter that controls character echo. It may either show asterisks
4050 (default, as before), turn echo off entirely, or echo the typed
4051 characters literally.
4053 * The systemd-ask-password tool also gained a new -n switch for
4054 suppressing output of a trailing newline character when writing the
4055 acquired password to standard output, similar to /bin/echo's -n
4058 * New documentation has been added that describes the organization of
4059 the systemd source code tree:
4061 https://systemd.io/ARCHITECTURE
4063 * Units using ConditionNeedsUpdate= will no longer be activated in
4066 * It is now possible to list a template unit in the WantedBy= or
4067 RequiredBy= settings of the [Install] section of another template
4068 unit, which will be instantiated using the same instance name.
4070 * A new MemoryAvailable property is available for units. If the unit,
4071 or the slices it is part of, have a memory limit set via MemoryMax=/
4072 MemoryHigh=, MemoryAvailable will indicate how much more memory the
4073 unit can claim before hitting the limits.
4075 * systemd-coredump will now try to stay below the cgroup memory limit
4076 placed on itself or one of the slices it runs under, if the storage
4077 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
4078 since files written on such filesystems count toward the cgroup memory
4079 limit. If there is not enough available memory in such cases to store
4080 the core file uncompressed, systemd-coredump will skip to compressed
4081 storage directly (if enabled) and it will avoid analyzing the core file
4082 to print backtrace and metadata in the journal.
4084 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
4085 of a path matches the configured expectations, and remove it if not.
4087 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
4088 specify which of the several available filesystem timestamps (access
4089 time, birth time, change time, modification time) to look at when
4090 deciding whether a path has aged enough to be cleaned.
4092 * A new IPv6StableSecretAddress= setting has been added to .network
4093 files, which takes an IPv6 address to use as secret for IPv6 address
4096 * The [DHCPServer] logic in .network files gained support for a new
4097 UplinkInterface= setting that permits configuration of the uplink
4098 interface name to propagate DHCP lease information from.
4100 * The WakeOnLan= setting in .link files now accepts a list of flags
4101 instead of a single one, to configure multiple wake-on-LAN policies.
4103 * User-space defined tracepoints (USDT) have been added to udev at
4104 strategic locations. This is useful for tracing udev behaviour and
4105 performance with bpftrace and similar tools.
4107 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
4108 setting a network timeout time.
4110 * If a system service is running in a new mount namespace (RootDirectory=
4111 and friends), all file systems will be mounted with MS_NOSUID by
4112 default, unless the system is running with SELinux enabled.
4114 * When enumerating time zones the timedatectl tool will now consult the
4115 'tzdata.zi' file shipped by the IANA time zone database package, in
4116 addition to 'zone1970.tab', as before. This makes sure time zone
4117 aliases are now correctly supported. Some distributions so far did
4118 not install this additional file, most do however. If you
4119 distribution does not install it yet, it might make sense to change
4122 * Intel HID rfkill event is no longer masked, since it's the only
4123 source of rfkill event on newer HP laptops. To have both backward and
4124 forward compatibility, userspace daemon needs to debounce duplicated
4125 events in a short time window.
4127 Contributions from: Aakash Singh, adrian5, Albert Brox,
4128 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
4129 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
4130 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
4131 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
4132 Christian Hesse, Daniel Schaefer, Dan Streetman,
4133 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
4134 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
4135 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
4136 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
4137 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
4138 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
4139 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
4140 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
4141 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
4142 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
4143 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
4144 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
4145 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
4146 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
4147 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
4148 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
4149 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
4150 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
4151 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
4152 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
4153 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
4154 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
4155 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
4156 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
4157 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
4159 — Edinburgh, 2021-07-07
4163 * A concept of system extension images is introduced. Such images may
4164 be used to extend the /usr/ and /opt/ directory hierarchies at
4165 runtime with additional files (even if the file system is read-only).
4166 When a system extension image is activated, its /usr/ and /opt/
4167 hierarchies and os-release information are combined via overlayfs
4168 with the file system hierarchy of the host OS.
4170 A new systemd-sysext tool can be used to merge, unmerge, list, and
4171 refresh system extension hierarchies. See
4172 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
4174 The systemd-sysext.service automatically merges installed system
4175 extensions during boot (before basic.target, but not in very early
4176 boot, since various file systems have to be mounted first).
4178 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
4179 supported system extension level.
4181 * A new ExtensionImages= unit setting can be used to apply the same
4182 system extension image concept from systemd-sysext to the namespaced
4183 file hierarchy of specific services, following the same rules and
4186 * Support for a new special "root=tmpfs" kernel command-line option has
4187 been added. When specified, a tmpfs is mounted on /, and mount.usr=
4188 should be used to point to the operating system implementation.
4190 * A new configuration file /etc/veritytab may be used to configure
4191 dm-verity integrity protection for block devices. Each line is in the
4192 format "volume-name data-device hash-device roothash options",
4193 similar to /etc/crypttab.
4195 * A new kernel command-line option systemd.verity.root_options= may be
4196 used to configure dm-verity behaviour for the root device.
4198 * The key file specified in /etc/crypttab (the third field) may now
4199 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
4200 acquired by connecting to that socket and reading from it. This
4201 allows the implementation of a service to provide key information
4202 dynamically, at the moment when it is needed.
4204 * When the hostname is set explicitly to "localhost", systemd-hostnamed
4205 will respect this. Previously such a setting would be mostly silently
4206 ignored. The goal is to honour configuration as specified by the
4209 * The fallback hostname that will be used by the system manager and
4210 systemd-hostnamed can now be configured in two new ways: by setting
4211 DEFAULT_HOSTNAME= in os-release(5), or by setting
4212 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
4213 also be configured during compilation. The environment variable is
4214 intended for testing and local overrides, the os-release(5) field is
4215 intended to allow customization by different variants of a
4216 distribution that share the same compiled packages.
4218 * The environment block of the manager itself may be configured through
4219 a new ManagerEnvironment= setting in system.conf or user.conf. This
4220 complements existing ways to set the environment block (the kernel
4221 command line for the system manager, the inherited environment and
4222 user@.service unit file settings for the user manager).
4224 * systemd-hostnamed now exports the default hostname and the source of
4225 the configured hostname ("static", "transient", or "default") as
4228 * systemd-hostnamed now exports the "HardwareVendor" and
4229 "HardwareModel" D-Bus properties, which are supposed to contain a
4230 pair of cleaned up, human readable strings describing the system's
4231 vendor and model. It's typically sourced from the firmware's DMI
4232 tables, but may be augmented from a new hwdb database. hostnamectl
4233 shows this in the status output.
4235 * Support has been added to systemd-cryptsetup for extracting the
4236 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
4237 metadata header. This allows the information how to open the
4238 encrypted device to be embedded directly in the device and obviates
4239 the need for configuration in an external file.
4241 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
4242 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
4243 pre-existing support for PKCS#11 security tokens).
4245 * systemd-repart may enroll encrypted partitions using TPM2
4246 hardware. This may be useful for example to create an encrypted /var
4247 partition bound to the machine on first boot.
4249 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
4250 and PKCS#11 security tokens to LUKS volumes, list and destroy
4253 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
4255 It also supports enrolling "recovery keys" and regular passphrases.
4257 * The libfido2 dependency is now based on dlopen(), so that the library
4258 is used at runtime when installed, but is not a hard runtime
4261 * systemd-cryptsetup gained support for two new options in
4262 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
4263 request synchronous processing of encryption/decryption IO.
4265 * The manager may be configured at compile time to use the fexecve()
4266 instead of the execve() system call when spawning processes. Using
4267 fexecve() closes a window between checking the security context of an
4268 executable and spawning it, but unfortunately the kernel displays
4269 stale information in the process' "comm" field, which impacts ps
4272 * The configuration option -Dcompat-gateway-hostname has been dropped.
4273 "_gateway" is now the only supported name.
4275 * The ConditionSecurity=tpm2 unit file setting may be used to check if
4276 the system has at least one TPM2 (tpmrm class) device.
4278 * A new ConditionCPUFeature= has been added that may be used to
4279 conditionalize units based on CPU features. For example,
4280 ConditionCPUFeature=rdrand will condition a unit so that it is only
4281 run when the system CPU supports the RDRAND opcode.
4283 * The existing ConditionControlGroupController= setting has been
4284 extended with two new values "v1" and "v2". "v2" means that the
4285 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
4286 hierarchy or the hybrid hierarchy are used.
4288 * A new PrivateIPC= setting on a unit file allows executed processes to
4289 be moved into a private IPC namespace, with separate System V IPC
4290 identifiers and POSIX message queues.
4292 A new IPCNamespacePath= allows the unit to be joined to an existing
4295 * The tables of system calls in seccomp filters are now automatically
4296 generated from kernel lists exported on
4297 https://fedora.juszkiewicz.com.pl/syscalls.html.
4299 The following architectures should now have complete lists:
4300 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
4301 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
4303 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
4304 on /run/ if it is not already a mount point. A writable /run/ has
4305 always been a requirement for a functioning system, but this was not
4306 guaranteed when using a read-only image.
4308 Users can always specify BindPaths= or InaccessiblePaths= as
4309 overrides, and they will take precedence. If the host's root mount
4310 point is used, there is no change in behaviour.
4312 * New bind mounts and file system image mounts may be injected into the
4313 mount namespace of a service (without restarting it). This is exposed
4314 respectively as 'systemctl bind <unit> <path>…' and
4315 'systemctl mount-image <unit> <image>…'.
4317 * The StandardOutput= and StandardError= settings can now specify files
4318 to be truncated for output (as "truncate:<path>").
4320 * The ExecPaths= and NoExecPaths= settings may be used to specify
4321 noexec for parts of the file system.
4323 * sd-bus has a new function sd_bus_open_user_machine() to open a
4324 connection to the session bus of a specific user in a local container
4325 or on the local host. This is exposed in the existing -M switch to
4326 systemctl and similar tools:
4328 systemctl --user -M lennart@foobar start foo
4330 This will connect to the user bus of a user "lennart" in container
4331 "foobar". If no container name is specified, the specified user on
4332 the host itself is connected to
4334 systemctl --user -M lennart@ start quux
4336 * sd-bus also gained a convenience function sd_bus_message_send() to
4337 simplify invocations of sd_bus_send(), taking only a single
4338 parameter: the message to send.
4340 * sd-event allows rate limits to be set on event sources, for dealing
4341 with high-priority event sources that might starve out others. See
4342 the new man page sd_event_source_set_ratelimit(3) for details.
4344 * systemd.link files gained a [Link] Promiscuous= switch, which allows
4345 the device to be raised in promiscuous mode.
4347 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
4348 number of TX and RX queues to be configured.
4350 New [Link] TransmitQueueLength= setting allows the size of the TX
4351 queue to be configured.
4353 New [Link] GenericSegmentOffloadMaxBytes= and
4354 GenericSegmentOffloadMaxSegments= allow capping the packet size and
4355 the number of segments accepted in Generic Segment Offload.
4357 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
4358 wireless routing protocol that operates on ISO/OSI Layer 2 only and
4359 uses ethernet frames to route/bridge packets. This encompasses a new
4360 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
4361 new settings in .netdev files, and a new BatmanAdvanced= setting in
4364 * systemd.network files gained a [Network] RouteTable= configuration
4365 switch to select the routing policy table.
4367 systemd.network files gained a [RoutingPolicyRule] Type=
4368 configuration switch (one of "blackhole, "unreachable", "prohibit").
4370 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
4371 RouteAllowList= settings to ignore/accept route advertisements from
4372 routers matching specified prefixes. The DenyList= setting has been
4373 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
4376 systemd.network files gained a [DHCPv6] UseAddress= setting to
4377 optionally ignore the address provided in the lease.
4379 systemd.network files gained a [DHCPv6PrefixDelegation]
4380 ManageTemporaryAddress= switch.
4382 systemd.network files gained a new ActivationPolicy= setting which
4383 allows configuring how the UP state of an interface shall be managed,
4384 i.e. whether the interface is always upped, always downed, or may be
4385 upped/downed by the user using "ip link set dev".
4387 * The default for the Broadcast= setting in .network files has slightly
4388 changed: the broadcast address will not be configured for wireguard
4391 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
4392 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
4393 configuration options for VLAN packet handling.
4395 * udev rules may now set log_level= option. This allows debug logs to
4396 be enabled for select events, e.g. just for a specific subsystem or
4397 even a single device.
4399 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
4400 DATA_PREPARED_ID properties for block devices with ISO9660 file
4403 * udev now exports decoded DMI information about installed memory slots
4404 as device properties under the /sys/class/dmi/id/ pseudo device.
4406 * /dev/ is not mounted noexec anymore. This didn't provide any
4407 significant security benefits and would conflict with the executable
4408 mappings used with /dev/sgx device nodes. The previous behaviour can
4409 be restored for individual services with NoExecPaths=/dev (or by allow-
4410 listing and excluding /dev from ExecPaths=).
4412 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
4413 and /dev/vhost-net are owned by the kvm group.
4415 * The hardware database has been extended with a list of fingerprint
4416 readers that correctly support USB auto-suspend using data from
4419 * systemd-resolved can now answer DNSSEC questions through the stub
4420 resolver interface in a way that allows local clients to do DNSSEC
4421 validation themselves. For a question with DO+CD set, it'll proxy the
4422 DNS query and respond with a mostly unmodified packet received from
4423 the upstream server.
4425 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
4426 resolved.conf. If true the service will provide caching even for DNS
4427 lookups made to an upstream DNS server on the 127.0.0.1/::1
4428 addresses. By default (and when the option is false) systemd-resolved
4429 will not cache such lookups, in order to avoid duplicate local
4430 caching, under the assumption the local upstream server caches
4433 * systemd-resolved now implements RFC5001 NSID in its local DNS
4434 stub. This may be used by local clients to determine whether they are
4435 talking to the DNS resolver stub or a different DNS server.
4437 * When resolving host names and other records resolvectl will now
4438 report where the data was acquired from (i.e. the local cache, the
4439 network, locally synthesized, …) and whether the network traffic it
4440 effected was encrypted or not. Moreover the tool acquired a number of
4441 new options --cache=, --synthesize=, --network=, --zone=,
4442 --trust-anchor=, --validate= that take booleans and may be used to
4443 tweak a lookup, i.e. whether it may be answered from cached
4444 information, locally synthesized information, information acquired
4445 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
4446 anchor, and whether DNSSEC validation shall be executed for the
4449 * systemd-nspawn gained a new --ambient-capability= setting
4450 (AmbientCapability= in .nspawn files) to configure ambient
4451 capabilities passed to the container payload.
4453 * systemd-nspawn gained the ability to configure the firewall using the
4454 nftables subsystem (in addition to the existing iptables
4455 support). Similarly, systemd-networkd's IPMasquerade= option now
4456 supports nftables as back-end, too. In both cases NAT on IPv6 is now
4457 supported too, in addition to IPv4 (the iptables back-end still is
4460 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
4461 retains its meaning, but has been deprecated. Please switch to either
4462 "ivp4" or "both" (if covering IPv6 is desired).
4464 * systemd-importd will now download .verity and .roothash.p7s files
4465 along with the machine image (as exposed via machinectl pull-raw).
4467 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
4468 setting to configure the time a unit's cgroup needs to exceed memory
4469 pressure limits before action will be taken, and a new
4470 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
4473 systemd-oomd is now considered fully supported (the usual
4474 backwards-compatibility promises apply). Swap is not required for
4475 operation, but it is still recommended.
4477 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
4478 configures the retry delay when trying to contact servers.
4480 * systemd-stdio-bridge gained --system/--user options to connect to the
4481 system bus (previous default) or the user session bus.
4483 * systemd-localed may now call locale-gen to generate missing locales
4484 on-demand (UTF-8-only). This improves integration with Debian-based
4485 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
4487 * systemctl --check-inhibitors=true may now be used to obey inhibitors
4488 even when invoked non-interactively. The old --ignore-inhibitors
4489 switch is now deprecated and replaced by --check-inhibitors=false.
4491 * systemctl import-environment will now emit a warning when called
4492 without any arguments (i.e. to import the full environment block of
4493 the called program). This command will usually be invoked from a
4494 shell, which means that it'll inherit a bunch of variables which are
4495 specific to that shell, and usually to the TTY the shell is connected
4496 to, and don't have any meaning in the global context of the system or
4497 user service manager. Instead, only specific variables should be
4498 imported into the manager environment block.
4500 Similarly, programs which update the manager environment block by
4501 directly calling the D-Bus API of the manager, should also push
4502 specific variables, and not the full inherited environment.
4504 * systemctl's status output now shows unit state with a more careful
4505 choice of Unicode characters: units in maintenance show a "○" symbol
4506 instead of the usual "●", failed units show "×", and services being
4509 * coredumpctl gained a --debugger-arguments= switch to pass arguments
4510 to the debugger. It also gained support for showing coredump info in
4511 a simple JSON format.
4513 * systemctl/loginctl/machinectl's --signal= option now accept a special
4514 value "list", which may be used to show a brief table with known
4515 process signals and their numbers.
4517 * networkctl now shows the link activation policy in status.
4519 * Various tools gained --pager/--no-pager/--json= switches to
4520 enable/disable the pager and provide JSON output.
4522 * Various tools now accept two new values for the SYSTEMD_COLORS
4523 environment variable: "16" and "256", to configure how many terminal
4524 colors are used in output.
4526 * less 568 or newer is now required for the auto-paging logic of the
4527 various tools. Hyperlink ANSI sequences in terminal output are now
4528 used even if a pager is used, and older versions of less are not able
4529 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
4530 disable this output again.
4532 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
4533 builds, non-merged-usr builds) are now officially deprecated. A
4534 warning is emitted during build. Support is slated to be removed in
4535 about a year (when the Debian Bookworm release development starts).
4537 * Systems with the legacy cgroup v1 hierarchy are now marked as
4538 "tainted", to make it clearer that using the legacy hierarchy is not
4541 * systemd-localed will now refuse to configure a keymap which is not
4542 installed in the file system. This is intended as a bug fix, but
4543 could break cases where systemd-localed was used to configure the
4544 keymap in advanced of it being installed. It is necessary to install
4545 the keymap file first.
4547 * The main git development branch has been renamed to 'main'.
4549 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
4550 for partitions, as in the vast majority of cases they contain none
4551 and are used internally by the bootloader (eg: uboot).
4553 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
4554 spawned processes to the PID of the process itself. This may be used
4555 by programs for detecting whether they were forked off by the service
4556 manager itself or are a process forked off further down the tree.
4558 * The sd-device API gained four new calls: sd_device_get_action() to
4559 determine the uevent add/remove/change/… action the device object has
4560 been seen for, sd_device_get_seqno() to determine the uevent sequence
4561 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
4562 object from stat(2) data of a device node, and sd_device_trigger() to
4563 write to the 'uevent' attribute of a device.
4565 * For most tools the --no-legend= switch has been replaced by
4566 --legend=no and --legend=yes, to force whether tables are shown with
4569 * Units acquired a new property "Markers" that takes a list of zero,
4570 one or two of the following strings: "needs-reload" and
4571 "needs-restart". These markers may be set via "systemctl
4572 set-property". Once a marker is set, "systemctl reload-or-restart
4573 --marked" may be invoked to execute the operation the units are
4574 marked for. This is useful for package managers that want to mark
4575 units for restart/reload while updating, but effect the actual
4576 operations at a later step at once.
4578 * The sd_bus_message_read_strv() API call of sd-bus may now also be
4579 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
4582 * bootctl will now report whether the UEFI firmware used a TPM2 device
4583 and measured the boot process into it.
4585 * systemd-tmpfiles learnt support for a new environment variable
4586 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
4587 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
4588 even if the root fs of the system is not itself a btrfs volume.
4590 * systemd-detect-virt/ConditionVirtualization= will now explicitly
4591 detect Docker/Podman environments where possible. Moreover, they
4592 should be able to generically detect any container manager as long as
4593 it assigns the container a cgroup.
4595 * portablectl gained a new "reattach" verb for detaching/reattaching a
4596 portable service image, useful for updating images on-the-fly.
4598 * Intel SGX enclave device nodes (which expose a security feature of
4599 newer Intel CPUs) will now be owned by a new system group "sgx".
4601 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
4602 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
4603 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
4604 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
4605 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
4606 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
4607 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
4608 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
4609 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
4610 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
4611 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
4612 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
4613 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
4614 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
4615 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
4616 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
4617 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
4618 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
4619 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
4620 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
4621 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
4622 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
4623 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
4624 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
4625 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
4626 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
4627 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
4628 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
4629 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
4630 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
4631 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
4632 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
4633 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
4634 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
4635 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
4636 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
4637 Zmicer Turok, Дамјан Георгиевски
4639 — Berlin, 2021-03-30
4643 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
4644 "bind" and "unbind" to the Linux device model. When this kernel
4645 change was made, systemd-udevd was only minimally updated to handle
4646 and propagate these new event types. The introduction of these new
4647 uevents (which are typically generated for USB devices and devices
4648 needing a firmware upload before being functional) resulted in a
4649 number of issues which we so far didn't address. We hoped the kernel
4650 maintainers would themselves address these issues in some form, but
4651 that did not happen. To handle them properly, many (if not most) udev
4652 rules files shipped in various packages need updating, and so do many
4653 programs that monitor or enumerate devices with libudev or sd-device,
4654 or otherwise process uevents. Please note that this incompatibility
4655 is not fault of systemd or udev, but caused by an incompatible kernel
4656 change that happened back in Linux 4.14, but is becoming more and
4657 more visible as the new uevents are generated by more kernel drivers.
4659 To minimize issues resulting from this kernel change (but not avoid
4660 them entirely) starting with systemd-udevd 247 the udev "tags"
4661 concept (which is a concept for marking and filtering devices during
4662 enumeration and monitoring) has been reworked: udev tags are now
4663 "sticky", meaning that once a tag is assigned to a device it will not
4664 be removed from the device again until the device itself is removed
4665 (i.e. unplugged). This makes sure that any application monitoring
4666 devices that match a specific tag is guaranteed to both see uevents
4667 where the device starts being relevant, and those where it stops
4668 being relevant (the latter now regularly happening due to the new
4669 "unbind" uevent type). The udev tags concept is hence now a concept
4670 tied to a *device* instead of a device *event* — unlike for example
4671 udev properties whose lifecycle (as before) is generally tied to a
4672 device event, meaning that the previously determined properties are
4673 forgotten whenever a new uevent is processed.
4675 With the newly redefined udev tags concept, sometimes it's necessary
4676 to determine which tags are the ones applied by the most recent
4677 uevent/database update, in order to discern them from those
4678 originating from earlier uevents/database updates of the same
4679 device. To accommodate for this a new automatic property CURRENT_TAGS
4680 has been added that works similar to the existing TAGS property but
4681 only lists tags set by the most recent uevent/database
4682 update. Similarly, the libudev/sd-device API has been updated with
4683 new functions to enumerate these 'current' tags, in addition to the
4684 existing APIs that now enumerate the 'sticky' ones.
4686 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
4687 essential that all udev rules files and applications are updated to
4688 handle the new events. Specifically:
4690 • All rule files that currently use a header guard similar to
4691 ACTION!="add|change",GOTO="xyz_end" should be updated to use
4692 ACTION=="remove",GOTO="xyz_end" instead, so that the
4693 properties/tags they add are also applied whenever "bind" (or
4694 "unbind") is seen. (This is most important for all physical device
4695 types — those for which "bind" and "unbind" are currently
4696 generated, for all other device types this change is still
4697 recommended but not as important — but certainly prepares for
4698 future kernel uevent type additions).
4700 • Similarly, all code monitoring devices that contains an 'if' branch
4701 discerning the "add" + "change" uevent actions from all other
4702 uevents actions (i.e. considering devices only relevant after "add"
4703 or "change", and irrelevant on all other events) should be reworked
4704 to instead negatively check for "remove" only (i.e. considering
4705 devices relevant after all event types, except for "remove", which
4706 invalidates the device). Note that this also means that devices
4707 should be considered relevant on "unbind", even though conceptually
4708 this — in some form — invalidates the device. Since the precise
4709 effect of "unbind" is not generically defined, devices should be
4710 considered relevant even after "unbind", however I/O errors
4711 accessing the device should then be handled gracefully.
4713 • Any code that uses device tags for deciding whether a device is
4714 relevant or not most likely needs to be updated to use the new
4715 udev_device_has_current_tag() API (or sd_device_has_current_tag()
4716 in case sd-device is used), to check whether the tag is set at the
4717 moment an uevent is seen (as opposed to the existing
4718 udev_device_has_tag() API which checks if the tag ever existed on
4719 the device, following the API concept redefinition explained
4722 We are very sorry for this breakage and the requirement to update
4723 packages using these interfaces. We'd again like to underline that
4724 this is not caused by systemd/udev changes, but result of a kernel
4727 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
4728 packages have not retriggered devices once the udev package (or any
4729 auxiliary package installing additional udev rules) is updated. We
4730 intend to work with major distributions to change this, so that
4731 "udevadm trigger -c change" is issued on such upgrades, ensuring that
4732 the updated ruleset is applied to the devices already discovered, so
4733 that (asynchronously) after the upgrade completed the udev database
4734 is consistent with the updated rule set. This means udev rules must
4735 be ready to be retriggered with a "change" action any time, and
4736 result in correct and complete udev database entries. While the
4737 majority of udev rule files known to us currently get this right,
4738 some don't. Specifically, there are udev rules files included in
4739 various packages that only set udev properties on the "add" action,
4740 but do not handle the "change" action. If a device matching those
4741 rules is retriggered with the "change" action (as is intended here)
4742 it would suddenly lose the relevant properties. This always has been
4743 problematic, but as soon as all udev devices are triggered on relevant
4744 package upgrades this will become particularly so. It is strongly
4745 recommended to fix offending rules so that they can handle a "change"
4746 action at any time, and acquire all necessary udev properties even
4747 then. Or in other words: the header guard mentioned above
4748 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
4749 this, as it makes sure rules are rerun on "change" correctly, and
4750 accumulate the correct and complete set of udev properties. udev rule
4751 definitions that cannot handle "change" events being triggered at
4752 arbitrary times should be considered buggy.
4754 * The MountAPIVFS= service file setting now defaults to on if
4755 RootImage= and RootDirectory= are used, which means that with those
4756 two settings /proc/, /sys/ and /dev/ are automatically properly set
4757 up for services. Previous behaviour may be restored by explicitly
4758 setting MountAPIVFS=off.
4760 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
4761 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
4762 latter it takes precedence over the former, similar to how most of
4763 systemd's own configuration is handled. Given that PAM stack
4764 definitions are primarily put together by OS vendors/distributions
4765 (though possibly overridden by users), this systemd release moves its
4766 own PAM stack configuration for the "systemd-user" PAM service (i.e.
4767 for the PAM session invoked by the per-user user@.service instance)
4768 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
4769 packages' vendor versions of their PAM stack definitions from
4770 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
4771 desired the location to which systemd installs its PAM stack
4772 configuration may be changed via the -Dpamconfdir Meson option.
4774 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
4775 libpwquality and libcryptsetup have been changed to be based on
4776 dlopen(): instead of regular dynamic library dependencies declared in
4777 the binary ELF headers, these libraries are now loaded on demand
4778 only, if they are available. If the libraries cannot be found the
4779 relevant operations will fail gracefully, or a suitable fallback
4780 logic is chosen. This is supposed to be useful for general purpose
4781 distributions, as it allows minimizing the list of dependencies the
4782 systemd packages pull in, permitting building of more minimal OS
4783 images, while still making use of these "weak" dependencies should
4784 they be installed. Since many package managers automatically
4785 synthesize package dependencies from ELF shared library dependencies,
4786 some additional manual packaging work has to be done now to replace
4787 those (slightly downgraded from "required" to "recommended" or
4788 whatever is conceptually suitable for the package manager). Note that
4789 this change does not alter build-time behaviour: as before the
4790 build-time dependencies have to be installed during build, even if
4791 they now are optional during runtime.
4793 * sd-event.h gained a new call sd_event_add_time_relative() for
4794 installing timers relative to the current time. This is mostly a
4795 convenience wrapper around the pre-existing sd_event_add_time() call
4796 which installs absolute timers.
4798 * sd-event event sources may now be placed in a new "exit-on-failure"
4799 mode, which may be controlled via the new
4800 sd_event_source_get_exit_on_failure() and
4801 sd_event_source_set_exit_on_failure() functions. If enabled, any
4802 failure returned by the event source handler functions will result in
4803 exiting the event loop (unlike the default behaviour of just
4804 disabling the event source but continuing with the event loop). This
4805 feature is useful to set for all event sources that define "primary"
4806 program behaviour (where failure should be fatal) in contrast to
4807 "auxiliary" behaviour (where failure should remain local).
4809 * Most event source types sd-event supports now accept a NULL handler
4810 function, in which case the event loop is exited once the event
4811 source is to be dispatched, using the userdata pointer — converted to
4812 a signed integer — as exit code of the event loop. Previously this
4813 was supported for IO and signal event sources already. Exit event
4814 sources still do not support this (simply because it makes little
4815 sense there, as the event loop is already exiting when they are
4818 * A new per-unit setting RootImageOptions= has been added which allows
4819 tweaking the mount options for any file system mounted as effect of
4820 the RootImage= setting.
4822 * Another new per-unit setting MountImages= has been added, that allows
4823 mounting additional disk images into the file system tree accessible
4826 * Timer units gained a new FixedRandomDelay= boolean setting. If
4827 enabled, the random delay configured with RandomizedDelaySec= is
4828 selected in a way that is stable on a given system (though still
4829 different for different units).
4831 * Socket units gained a new setting Timestamping= that takes "us", "ns"
4832 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
4835 * systemd-repart now generates JSON output when requested with the new
4838 * systemd-machined's OpenMachineShell() bus call will now pass
4839 additional policy metadata data fields to the PolicyKit
4840 authentication request.
4842 * systemd-tmpfiles gained a new -E switch, which is equivalent to
4843 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
4844 --exclude=/sys. It's particularly useful in combination with --root=,
4845 when operating on OS trees that do not have any of these four runtime
4846 directories mounted, as this means no files below these subtrees are
4847 created or modified, since those mount points should probably remain
4850 * systemd-tmpfiles gained a new --image= switch which is like --root=,
4851 but takes a disk image instead of a directory as argument. The
4852 specified disk image is mounted inside a temporary mount namespace
4853 and the tmpfiles.d/ drop-ins stored in the image are executed and
4854 applied to the image. systemd-sysusers similarly gained a new
4855 --image= switch, that allows the sysusers.d/ drop-ins stored in the
4856 image to be applied onto the image.
4858 * Similarly, the journalctl command also gained an --image= switch,
4859 which is a quick one-step solution to look at the log data included
4862 * journalctl's --output=cat option (which outputs the log content
4863 without any metadata, just the pure text messages) will now make use
4864 of terminal colors when run on a suitable terminal, similarly to the
4867 * JSON group records now support a "description" string that may be
4868 used to add a human-readable textual description to such groups. This
4869 is supposed to match the user's GECOS field which traditionally
4870 didn't have a counterpart for group records.
4872 * The "systemd-dissect" tool that may be used to inspect OS disk images
4873 and that was previously installed to /usr/lib/systemd/ has now been
4874 moved to /usr/bin/, reflecting its updated status of an officially
4875 supported tool with a stable interface. It gained support for a new
4876 --mkdir switch which when combined with --mount has the effect of
4877 creating the directory to mount the image to if it is missing
4878 first. It also gained two new commands --copy-from and --copy-to for
4879 copying files and directories in and out of an OS image without the
4880 need to manually mount it. It also acquired support for a new option
4881 --json= to generate JSON output when inspecting an OS image.
4883 * The cgroup2 file system is now mounted with the
4884 "memory_recursiveprot" mount option, supported since kernel 5.7. This
4885 means that the MemoryLow= and MemoryMin= unit file settings now apply
4886 recursively to whole subtrees.
4888 * systemd-homed now defaults to using the btrfs file system — if
4889 available — when creating home directories in LUKS volumes. This may
4890 be changed with the DefaultFileSystemType= setting in homed.conf.
4891 It's now the default file system in various major distributions and
4892 has the major benefit for homed that it can be grown and shrunk while
4893 mounted, unlike the other contenders ext4 and xfs, which can both be
4894 grown online, but not shrunk (in fact xfs is the technically most
4895 limited option here, as it cannot be shrunk at all).
4897 * JSON user records managed by systemd-homed gained support for
4898 "recovery keys". These are basically secondary passphrases that can
4899 unlock user accounts/home directories. They are computer-generated
4900 rather than user-chosen, and typically have greater entropy.
4901 homectl's --recovery-key= option may be used to add a recovery key to
4902 a user account. The generated recovery key is displayed as a QR code,
4903 so that it can be scanned to be kept in a safe place. This feature is
4904 particularly useful in combination with systemd-homed's support for
4905 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
4906 security tokens are lost. Recovery keys may be entered wherever the
4907 system asks for a password.
4909 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
4910 home directory which indicates that a home directory has not been
4911 deactivated cleanly when offline. This flag is useful to identify
4912 home directories for which the offline discard logic did not run when
4913 offlining, and where it would be a good idea to log in again to catch
4916 * systemctl gained a new parameter --timestamp= which may be used to
4917 change the style in which timestamps are output, i.e. whether to show
4918 them in local timezone or UTC, or whether to show µs granularity.
4920 * Alibaba's "pouch" container manager is now detected by
4921 systemd-detect-virt, ConditionVirtualization= and similar
4922 constructs. Similar, they now also recognize IBM PowerVM machine
4925 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
4926 place to use for propagating external mounts into the
4927 container. Similarly /run/host/notify is now used as the socket path
4928 for container payloads to communicate with the container manager
4929 using sd_notify(). The container manager now uses the
4930 /run/host/inaccessible/ directory to place "inaccessible" file nodes
4931 of all relevant types which may be used by the container payload as
4932 bind mount source to over-mount inodes to make them inaccessible.
4933 /run/host/container-manager will now be initialized with the same
4934 string as the $container environment variable passed to the
4935 container's PID 1. /run/host/container-uuid will be initialized with
4936 the same string as $container_uuid. This means the /run/host/
4937 hierarchy is now the primary way to make host resources available to
4938 the container. The Container Interface documents these new files and
4941 https://systemd.io/CONTAINER_INTERFACE
4943 * Support for the "ConditionNull=" unit file condition has been
4944 deprecated and undocumented for 6 years. systemd started to warn
4945 about its use 1.5 years ago. It has now been removed entirely.
4947 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
4948 a sd_bus_error struct and a list of error names, and checks if the
4949 error matches one of these names. It's a convenience wrapper that is
4950 useful in cases where multiple errors shall be handled the same way.
4952 * A new system call filter list "@known" has been added, that contains
4953 all system calls known at the time systemd was built.
4955 * Behaviour of system call filter allow lists has changed slightly:
4956 system calls that are contained in @known will result in EPERM by
4957 default, while those not contained in it result in ENOSYS. This
4958 should improve compatibility because known system calls will thus be
4959 communicated as prohibited, while unknown (and thus newer ones) will
4960 be communicated as not implemented, which hopefully has the greatest
4961 chance of triggering the right fallback code paths in client
4964 * "systemd-analyze syscall-filter" will now show two separate sections
4965 at the bottom of the output: system calls known during systemd build
4966 time but not included in any of the filter groups shown above, and
4967 system calls defined on the local kernel but known during systemd
4970 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
4971 systemd-nspawn all system call filter violations will be logged by
4972 the kernel (audit). This is useful for tracking down system calls
4973 invoked by container payloads that are prohibited by the container's
4974 system call filter policy.
4976 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
4977 systemd-nspawn (and other programs that use seccomp) all seccomp
4978 filtering is turned off.
4980 * Two new unit file settings ProtectProc= and ProcSubset= have been
4981 added that expose the hidepid= and subset= mount options of procfs.
4982 All processes of the unit will only see processes in /proc that are
4983 are owned by the unit's user. This is an important new sandboxing
4984 option that is recommended to be set on all system services. All
4985 long-running system services that are included in systemd itself set
4986 this option now. This option is only supported on kernel 5.8 and
4987 above, since the hidepid= option supported on older kernels was not a
4988 per-mount option but actually applied to the whole PID namespace.
4990 * Socket units gained a new boolean setting FlushPending=. If enabled
4991 all pending socket data/connections are flushed whenever the socket
4992 unit enters the "listening" state, i.e. after the associated service
4995 * The unit file setting NUMAMask= gained a new "all" value: when used,
4996 all existing NUMA nodes are added to the NUMA mask.
4998 * A new "credentials" logic has been added to system services. This is
4999 a simple mechanism to pass privileged data to services in a safe and
5000 secure way. It's supposed to be used to pass per-service secret data
5001 such as passwords or cryptographic keys but also associated less
5002 private information such as user names, certificates, and similar to
5003 system services. Each credential is identified by a short user-chosen
5004 name and may contain arbitrary binary data. Two new unit file
5005 settings have been added: SetCredential= and LoadCredential=. The
5006 former allows setting a credential to a literal string, the latter
5007 sets a credential to the contents of a file (or data read from a
5008 user-chosen AF_UNIX stream socket). Credentials are passed to the
5009 service via a special credentials directory, one file for each
5010 credential. The path to the credentials directory is passed in a new
5011 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
5012 are passed in the file system they may be easily referenced in
5013 ExecStart= command lines too, thus no explicit support for the
5014 credentials logic in daemons is required (though ideally daemons
5015 would look for the bits they need in $CREDENTIALS_DIRECTORY
5016 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
5017 backed by unswappable memory if privileges allow it, immutable if
5018 privileges allow it, is accessible only to the service's UID, and is
5019 automatically destroyed when the service stops.
5021 * systemd-nspawn supports the same credentials logic. It can both
5022 consume credentials passed to it via the aforementioned
5023 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
5024 to its payload. The service manager/PID 1 has been updated to match
5025 this: it can also accept credentials from the container manager that
5026 invokes it (in fact: any process that invokes it), and passes them on
5027 to its services. Thus, credentials can be propagated recursively down
5028 the tree: from a system's service manager to a systemd-nspawn
5029 service, to the service manager that runs as container payload and to
5030 the service it runs below. Credentials may also be added on the
5031 systemd-nspawn command line, using new --set-credential= and
5032 --load-credential= command line switches that match the
5033 aforementioned service settings.
5035 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
5036 the partition drop-ins which may be used to format/LUKS
5037 encrypt/populate any created partitions. The partitions are
5038 encrypted/formatted/populated before they are registered in the
5039 partition table, so that they appear atomically: either the
5040 partitions do not exist yet or they exist fully encrypted, formatted,
5041 and populated — there is no time window where they are
5042 "half-initialized". Thus the system is robust to abrupt shutdown: if
5043 the tool is terminated half-way during its operations on next boot it
5044 will start from the beginning.
5046 * systemd-repart's --size= operation gained a new "auto" value. If
5047 specified, and operating on a loopback file it is automatically sized
5048 to the minimal size the size constraints permit. This is useful to
5049 use "systemd-repart" as an image builder for minimally sized images.
5051 * systemd-resolved now gained a third IPC interface for requesting name
5052 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
5053 interface is now supported. The nss-resolve NSS module has been
5054 modified to use this new interface instead of D-Bus. Using Varlink
5055 has a major benefit over D-Bus: it works without a broker service,
5056 and thus already during earliest boot, before the dbus daemon has
5057 been started. This means name resolution via systemd-resolved now
5058 works at the same time systemd-networkd operates: from earliest boot
5059 on, including in the initrd.
5061 * systemd-resolved gained support for a new DNSStubListenerExtra=
5062 configuration file setting which may be used to specify additional IP
5063 addresses the built-in DNS stub shall listen on, in addition to the
5064 main one on 127.0.0.53:53.
5066 * Name lookups issued via systemd-resolved's D-Bus and Varlink
5067 interfaces (and thus also via glibc NSS if nss-resolve is used) will
5068 now honour a trailing dot in the hostname: if specified the search
5069 path logic is turned off. Thus "resolvectl query foo." is now
5070 equivalent to "resolvectl query --search=off foo.".
5072 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
5073 exposes how /etc/resolv.conf is currently managed: by resolved (and
5074 in which mode if so) or another subsystem. "resolvctl" will display
5075 this property in its status output.
5077 * The resolv.conf snippets systemd-resolved provides will now set "."
5078 as the search domain if no other search domain is known. This turns
5079 off the derivation of an implicit search domain by nss-dns for the
5080 hostname, when the hostname is set to an FQDN. This change is done to
5081 make nss-dns using resolv.conf provided by systemd-resolved behave
5082 more similarly to nss-resolve.
5084 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
5085 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
5086 "birth" time (btime) of a file in addition to the atime, mtime, and
5089 * systemd-analyze gained a new verb "capability" that lists all known
5090 capabilities by the systemd build and by the kernel.
5092 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
5093 advance the system clock to it at boot if it is noticed to be before
5094 that time. Previously, PID 1 would only advance the time to an epoch
5095 time that is set during build-time. With this new file OS builders
5096 can change this epoch timestamp on individual OS images without
5097 having to rebuild systemd.
5099 * systemd-logind will now listen to the KEY_RESTART key from the Linux
5100 input layer and reboot the system if it is pressed, similarly to how
5101 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
5102 was originally defined in the Multimedia context (to restart playback
5103 of a song or film), but is now primarily used in various embedded
5104 devices for "Reboot" buttons. Accordingly, systemd-logind will now
5105 honour it as such. This may configured in more detail via the new
5106 HandleRebootKey= and RebootKeyIgnoreInhibited=.
5108 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
5109 copying OS trees, for example in "systemd-nspawn --ephemeral",
5110 "systemd-nspawn --template=", "machinectl clone" and similar. This is
5111 useful when operating with OSTree images, which use hardlinks heavily
5112 throughout, and where such copies previously resulting in "exploding"
5115 * systemd-nspawn's --console= setting gained support for a new
5116 "autopipe" value, which is identical to "interactive" when invoked on
5117 a TTY, and "pipe" otherwise.
5119 * systemd-networkd's .network files gained support for explicitly
5120 configuring the multicast membership entries of bridge devices in the
5121 [BridgeMDB] section. It also gained support for the PIE queuing
5122 discipline in the [FlowQueuePIE] sections.
5124 * systemd-networkd's .netdev files may now be used to create "BareUDP"
5125 tunnels, configured in the new [BareUDP] setting.
5127 * systemd-networkd's Gateway= setting in .network files now accepts the
5128 special values "_dhcp4" and "_ipv6ra" to configure additional,
5129 locally defined, explicit routes to the gateway acquired via DHCP or
5130 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
5131 but still accepted for backwards compatibility.
5133 * systemd-networkd's [IPv6PrefixDelegation] section and
5134 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
5135 IPv6SendRA= (the old names are still accepted for backwards
5138 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
5139 boolean setting in [Network] section. If enabled, the delegated prefix
5140 gained by another link will be configured, and an address within the
5141 prefix will be assigned.
5143 * systemd-networkd's .network files gained the Announce= boolean setting
5144 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
5145 prefix will be announced through IPv6 router advertisement (IPv6 RA).
5146 The setting is enabled by default.
5148 * VXLAN tunnels may now be marked as independent of any underlying
5149 network interface via the new Independent= boolean setting.
5151 * systemctl gained support for two new verbs: "service-log-level" and
5152 "service-log-target" may be used on services that implement the
5153 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
5154 adjust the log level and target. All of systemd's long-running
5155 services support this now, but ideally all system services would
5156 implement this interface to make the system more uniformly
5159 * The SystemCallErrorNumber= unit file setting now accepts the new
5160 "kill" and "log" actions, in addition to arbitrary error number
5161 specifications as before. If "kill" the processes are killed on the
5162 event, if "log" the offending system call is audit logged.
5164 * A new SystemCallLog= unit file setting has been added that accepts a
5165 list of system calls that shall be logged about (audit).
5167 * The OS image dissection logic (as used by RootImage= in unit files or
5168 systemd-nspawn's --image= switch) has gained support for identifying
5169 and mounting explicit /usr/ partitions, which are now defined in the
5170 discoverable partition specification. This should be useful for
5171 environments where the root file system is
5172 generated/formatted/populated dynamically on first boot and combined
5173 with an immutable /usr/ tree that is supplied by the vendor.
5175 * In the final phase of shutdown, within the systemd-shutdown binary
5176 we'll now try to detach MD devices (i.e software RAID) in addition to
5177 loopback block devices and DM devices as before. This is supposed to
5178 be a safety net only, in order to increase robustness if things go
5179 wrong. Storage subsystems are expected to properly detach their
5180 storage volumes during regular shutdown already (or in case of
5181 storage backing the root file system: in the initrd hook we return to
5184 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
5185 will now log the thread ID in their log output. This is useful when
5186 working with heavily threaded programs.
5188 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
5189 not use the RDRAND CPU instruction. This is useful in environments
5190 such as replay debuggers where non-deterministic behaviour is not
5193 * The autopaging logic in systemd's various tools (such as systemctl)
5194 has been updated to turn on "secure" mode in "less"
5195 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
5196 detected. This disables invoking external programs from the pager,
5197 via the pipe logic. This behaviour may be overridden via the new
5198 $SYSTEMD_PAGERSECURE environment variable.
5200 * Units which have resource limits (.service, .mount, .swap, .slice,
5201 .socket, and .slice) gained new configuration settings
5202 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
5203 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
5204 limits and optional action taken by systemd-oomd.
5206 * A new service systemd-oomd has been added. It monitors resource
5207 contention for selected parts of the unit hierarchy using the PSI
5208 information reported by the kernel, and kills processes when memory
5209 or swap pressure is above configured limits. This service is only
5210 enabled by default in developer mode (see below) and should be
5211 considered a preview in this release. Behaviour details and option
5212 names are subject to change without the usual backwards-compatibility
5215 * A new helper oomctl has been added to introspect systemd-oomd state.
5216 It is only enabled by default in developer mode and should be
5217 considered a preview without the usual backwards-compatibility
5220 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
5221 enabled, systemd reads the system UID boundaries from /etc/login.defs
5222 at runtime, instead of using the built-in values selected during
5223 build. This is an option to improve compatibility for upgrades from
5224 old systems. It's strongly recommended not to make use of this
5225 functionality on new systems (or even enable it during build), as it
5226 makes something runtime-configurable that is mostly an implementation
5227 detail of the OS, and permits avoidable differences in deployments
5228 that create all kinds of problems in the long run.
5230 * New meson option '-Dmode=developer|release' has been added. When
5231 'developer', additional checks and features are enabled that are
5232 relevant during upstream development, e.g. verification that
5233 semi-automatically-generated documentation has been properly updated
5234 following API changes. Those checks are considered hints for
5235 developers and are not actionable in downstream builds. In addition,
5236 extra features that are not ready for general consumption may be
5237 enabled in developer mode. It is thus recommended to set
5238 '-Dmode=release' in end-user and distro builds.
5240 * systemd-cryptsetup gained support for processing detached LUKS
5241 headers specified on the kernel command line via the header=
5242 parameter of the luks.options= kernel command line option. The same
5243 device/path syntax as for key files is supported for header files
5246 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
5247 slot index data for devices that are connected through a PCI bridge
5248 where the _SUN index is associated with the bridge instead of the
5249 network device itself. Previously this would create ambiguous device
5250 naming if multiple network interfaces were connected to the same PCI
5251 bridge. Since this is a naming scheme incompatibility on systems that
5252 possess hardware like this it has been introduced as new naming
5253 scheme "v247". The previous scheme can be selected via the
5254 "net.naming-scheme=v245" kernel command line parameter.
5256 * ConditionFirstBoot= semantics have been modified to be safe towards
5257 abnormal system power-off during first boot. Specifically, the
5258 "systemd-machine-id-commit.service" service now acts as boot
5259 milestone indicating when the first boot process is sufficiently
5260 complete in order to not consider the next following boot also a
5261 first boot. If the system is reset before this unit is reached the
5262 first time, the next boot will still be considered a first boot; once
5263 it has been reached, no further boots will be considered a first
5264 boot. The "first-boot-complete.target" unit now acts as official hook
5265 point to order against this. If a service shall be run on every boot
5266 until the first boot fully succeeds it may thus be ordered before
5267 this target unit (and pull it in) and carry ConditionFirstBoot=
5270 * bootctl's set-default and set-oneshot commands now accept the three
5271 special strings "@default", "@oneshot", "@current" in place of a boot
5272 entry id. These strings are resolved to the current default and
5273 oneshot boot loader entry, as well as the currently booted one. Thus
5274 a command "bootctl set-default @current" may be used to make the
5275 currently boot menu item the new default for all subsequent boots.
5277 * "systemctl edit" has been updated to show the original effective unit
5278 contents in commented form in the text editor.
5280 * Units in user mode are now segregated into three new slices:
5281 session.slice (units that form the core of graphical session),
5282 app.slice ("normal" user applications), and background.slice
5283 (low-priority tasks). Unless otherwise configured, user units are
5284 placed in app.slice. The plan is to add resource limits and
5285 protections for the different slices in the future.
5287 * New GPT partition types for RISCV32/64 for the root and /usr
5288 partitions, and their associated Verity partitions have been defined,
5289 and are now understood by systemd-gpt-auto-generator, and the OS
5290 image dissection logic.
5292 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
5293 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
5294 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
5295 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
5296 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
5297 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
5298 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
5299 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
5300 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
5301 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
5302 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
5303 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
5304 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
5305 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
5306 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
5307 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
5308 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
5309 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
5310 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
5311 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
5312 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
5313 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
5314 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
5315 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
5316 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
5317 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
5318 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
5319 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
5320 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
5321 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
5322 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
5323 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
5324 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
5326 – Warsaw, 2020-11-26
5330 * The service manager gained basic support for cgroup v2 freezer. Units
5331 can now be suspended or resumed either using new systemctl verbs,
5332 freeze and thaw respectively, or via D-Bus.
5334 * PID 1 may now automatically load pre-compiled AppArmor policies from
5335 /etc/apparmor/earlypolicy during early boot.
5337 * The CPUAffinity= setting in service unit files now supports a new
5338 special value "numa" that causes the CPU affinity masked to be set
5339 based on the NUMA mask.
5341 * systemd will now log about all left-over processes remaining in a
5342 unit when the unit is stopped. It will now warn about services using
5343 KillMode=none, as this is generally an unsafe thing to make use of.
5345 * Two new unit file settings
5346 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
5347 added. They may be used to check whether a specific file system path
5348 resides on a block device that is encrypted on the block level
5349 (i.e. using dm-crypt/LUKS).
5351 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
5352 has been added that may be used for simple environment checks. This
5353 is particularly useful when passing in environment variables from a
5354 container manager (or from PAM in case of the systemd --user
5357 * .service unit files now accept a new setting CoredumpFilter= which
5358 allows configuration of the memory sections coredumps of the
5359 service's processes shall include.
5361 * .mount units gained a new ReadWriteOnly= boolean option. If set
5362 it will not be attempted to mount a file system read-only if mounting
5363 in read-write mode doesn't succeed. An option x-systemd.rw-only is
5364 available in /etc/fstab to control the same.
5366 * .socket units gained a new boolean setting PassPacketInfo=. If
5367 enabled, the kernel will attach additional per-packet metadata to all
5368 packets read from the socket, as an ancillary message. This controls
5369 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
5370 depending on socket type.
5372 * .service units gained a new setting RootHash= which may be used to
5373 specify the root hash for verity enabled disk images which are
5374 specified in RootImage=. RootVerity= may be used to specify a path to
5375 the Verity data matching a RootImage= file system. (The latter is
5376 only useful for images that do not contain the Verity data embedded
5377 into the same image that carries a GPT partition table following the
5378 Discoverable Partition Specification). Similarly, systemd-nspawn
5379 gained a new switch --verity-data= that takes a path to a file with
5380 the verity data of the disk image supplied in --image=, if the image
5381 doesn't contain the verity data itself.
5383 * .service units gained a new setting RootHashSignature= which takes
5384 either a base64 encoded PKCS#7 signature of the root hash specified
5385 with RootHash=, or a path to a file to read the signature from. This
5386 allows validation of the root hash against public keys available in
5387 the kernel keyring, and is only supported on recent kernels
5388 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
5389 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
5390 this mechanism has also been added to systemd-veritysetup.
5392 * .service unit files gained two new options
5393 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
5394 tune behaviour if a start or stop timeout is hit, i.e. whether to
5395 terminate the service with SIGTERM, SIGABRT or SIGKILL.
5397 * Most options in systemd that accept hexadecimal values prefixed with
5398 0x in additional to the usual decimal notation now also support octal
5399 notation when the 0o prefix is used and binary notation if the 0b
5402 * Various command line parameters and configuration file settings that
5403 configure key or certificate files now optionally take paths to
5404 AF_UNIX sockets in the file system. If configured that way a stream
5405 connection is made to the socket and the required data read from
5406 it. This is a simple and natural extension to the existing regular
5407 file logic, and permits other software to provide keys or
5408 certificates via simple IPC services, for example when unencrypted
5409 storage on disk is not desired. Specifically, systemd-networkd's
5410 Wireguard and MACSEC key file settings as well as
5411 systemd-journal-gatewayd's and systemd-journal-remote's PEM
5412 key/certificate parameters support this now.
5414 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
5415 configuration files that support specifier expansion learnt six new
5416 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
5417 resolve to the various ID fields from /etc/os-release, %l resolves to
5418 the "short" hostname of the system, i.e. the hostname configured in
5419 the kernel truncated at the first dot.
5421 * Support for the .include syntax in unit files has been removed. The
5422 concept has been obsolete for 6 years and we started warning about
5423 its pending removal 2 years ago (also see NEWS file below). It's
5426 * StandardError= and StandardOutput= in unit files no longer support
5427 the "syslog" and "syslog-console" switches. They were long removed
5428 from the documentation, but will now result in warnings when used,
5429 and be converted to "journal" and "journal+console" automatically.
5431 * If the service setting User= is set to the "nobody" user, a warning
5432 message is now written to the logs (but the value is nonetheless
5433 accepted). Setting User=nobody is unsafe, since the primary purpose
5434 of the "nobody" user is to own all files whose owner cannot be mapped
5435 locally. It's in particular used by the NFS subsystem and in user
5436 namespacing. By running a service under this user's UID it might get
5437 read and even write access to all these otherwise unmappable files,
5438 which is quite likely a major security problem.
5440 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
5441 and others) now have a size and inode limits applied (50% of RAM for
5442 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
5443 that the implicit kernel default is 50% too, so there is no change
5444 in the size limit for /tmp and /dev/shm.
5446 * nss-mymachines lost support for resolution of users and groups, and
5447 now only does resolution of hostnames. This functionality is now
5448 provided by nss-systemd. Thus, the 'mymachines' entry should be
5449 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
5450 (and 'systemd' added if it is not already there).
5452 * A new kernel command line option systemd.hostname= has been added
5453 that allows controlling the hostname that is initialized early during
5456 * A kernel command line option "udev.blockdev_read_only" has been
5457 added. If specified all hardware block devices that show up are
5458 immediately marked as read-only by udev. This option is useful for
5459 making sure that a specific boot under no circumstances modifies data
5460 on disk. Use "blockdev --setrw" to undo the effect of this, per
5463 * A new boolean kernel command line option systemd.swap= has been
5464 added, which may be used to turn off automatic activation of swap
5465 devices listed in /etc/fstab.
5467 * New kernel command line options systemd.condition-needs-update= and
5468 systemd.condition-first-boot= have been added, which override the
5469 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
5472 * A new kernel command line option systemd.clock-usec= has been added
5473 that allows setting the system clock to the specified time in µs
5474 since Jan 1st, 1970 early during boot. This is in particular useful
5475 in order to make test cases more reliable.
5477 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
5478 systemd-coredump to save core files for suid processes. When saving
5479 the core file, systemd-coredump will use the effective uid and gid of
5480 the process that faulted.
5482 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
5483 now automatically set to "Y" at boot, in order to enable pstore
5484 generation for collection with systemd-pstore.
5486 * We provide a set of udev rules to enable auto-suspend on PCI and USB
5487 devices that were tested to correctly support it. Previously, this
5488 was distributed as a set of udev rules, but has now been replaced by
5489 by a set of hwdb entries (and a much shorter udev rule to take action
5490 if the device modalias matches one of the new hwdb entries).
5492 As before, entries are periodically imported from the database
5493 maintained by the ChromiumOS project. If you have a device that
5494 supports auto-suspend correctly and where it should be enabled by
5495 default, please submit a patch that adds it to the database (see
5496 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
5498 * systemd-udevd gained the new configuration option timeout_signal= as well
5499 as a corresponding kernel command line option udev.timeout_signal=.
5500 The option can be used to configure the UNIX signal that the main
5501 daemon sends to the worker processes on timeout. Setting the signal
5502 to SIGABRT is useful for debugging.
5504 * .link files managed by systemd-udevd gained options RxFlowControl=,
5505 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
5506 order to configure various flow control parameters. They also gained
5507 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
5508 frame ring buffer sizes.
5510 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
5511 enabled systemd-networkd manages all routes configured by other tools.
5513 * .network files managed by systemd-networkd gained a new section
5514 [SR-IOV], in order to configure SR-IOV capable network devices.
5516 * systemd-networkd's [IPv6Prefix] section in .network files gained a
5517 new boolean setting Assign=. If enabled an address from the prefix is
5518 automatically assigned to the interface.
5520 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
5521 controls delegated prefixes assigned by DHCPv6 client. The section
5522 has three settings: SubnetID=, Assign=, and Token=. The setting
5523 SubnetID= allows explicit configuration of the preferred subnet that
5524 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
5525 Assign= is enabled (which is the default) an address from any acquired
5526 delegated prefix is automatically chosen and assigned to the
5527 interface. The setting Token= specifies an optional address generation
5530 * systemd-networkd's [Network] section gained a new setting
5531 IPv4AcceptLocal=. If enabled the interface accepts packets with local
5534 * systemd-networkd gained support for configuring the HTB queuing
5535 discipline in the [HierarchyTokenBucket] and
5536 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
5537 be configured in the [PFIFO] section, "GRED" in
5538 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
5539 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
5540 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
5541 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
5542 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
5543 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
5545 * systemd-networkd gained support for a new Termination= setting in the
5546 [CAN] section for configuring the termination resistor. It also
5547 gained a new ListenOnly= setting for controlling whether to only
5548 listen on CAN interfaces, without interfering with traffic otherwise
5549 (which is useful for debugging/monitoring CAN network
5550 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
5551 been added to configure various CAN-FD aspects.
5553 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
5554 When enabled, DHCPv6 will be attempted right-away without requiring an
5555 Router Advertisement packet suggesting it first (i.e. without the 'M'
5556 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
5557 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
5558 the RA packets suggest it.
5560 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
5561 which may be used to turn off use of the gateway information provided
5562 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
5563 used to configure how to process leases that lack a lifetime option.
5565 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
5566 setting SendVendorOption= allowing configuration of additional vendor
5567 options to send in the DHCP requests/responses. The [DHCPv6] section
5568 gained a new SendOption= setting for sending arbitrary DHCP
5569 options. RequestOptions= has been added to request arbitrary options
5570 from the server. UserClass= has been added to set the DHCP user class
5573 * systemd-networkd's [DHCPServer] section gained a new set of options
5574 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
5575 information about these three protocols in the DHCP lease. It also
5576 gained support for including "MUD" URLs ("Manufacturer Usage
5577 Description"). Support for "MUD" URLs was also added to the LLDP
5578 stack, configurable in the [LLDP] section in .network files.
5580 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
5581 mode. Also, the sections now support a new setting SourceMACAddress=.
5583 * systemd-networkd's .netdev files now support a new setting
5584 VLANProtocol= in the [Bridge] section that allows configuration of
5585 the VLAN protocol to use.
5587 * systemd-networkd supports a new Group= setting in the [Link] section
5588 of the .network files, to control the link group.
5590 * systemd-networkd's [Network] section gained a new
5591 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
5592 link local address is generated.
5594 * A new default .network file is now shipped that matches TUN/TAP
5595 devices that begin with "vt-" in their name. Such interfaces will
5596 have IP routing onto the host links set up automatically. This is
5597 supposed to be used by VM managers to trivially acquire a network
5598 interface which is fully set up for host communication, simply by
5599 carefully picking an interface name to use.
5601 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
5602 which sets the route priority for routes specified by the DHCP server.
5604 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
5605 which configures the vendor class information sent to DHCP server.
5607 * The BlackList= settings in .network files' [DHCPv4] and
5608 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
5609 are still understood to provide compatibility.
5611 * networkctl gained the new "forcerenew" command for forcing all DHCP
5612 server clients to renew their lease. The interface "status" output
5613 will now show numerous additional fields of information about an
5614 interface. There are new "up" and "down" commands to bring specific
5615 interfaces up or down.
5617 * systemd-resolved's DNS= configuration option now optionally accepts a
5618 port number (after ":") and a host name (after "#"). When the host
5619 name is specified, the DNS-over-TLS certificate is validated to match
5620 the specified hostname. Additionally, in case of IPv6 addresses, an
5621 interface may be specified (after "%").
5623 * systemd-resolved may be configured to forward single-label DNS names.
5624 This is not standard-conformant, but may make sense in setups where
5625 public DNS servers are not used.
5627 * systemd-resolved's DNS-over-TLS support gained SNI validation.
5629 * systemd-nspawn's --resolv-conf= switch gained a number of new
5630 supported values. Specifically, options starting with "replace-" are
5631 like those prefixed "copy-" but replace any existing resolv.conf
5632 file. And options ending in "-uplink" and "-stub" can now be used to
5633 propagate other flavours of resolv.conf into the container (as
5634 defined by systemd-resolved).
5636 * The various programs included in systemd can now optionally output
5637 their log messages on stderr prefixed with a timestamp, controlled by
5638 the $SYSTEMD_LOG_TIME environment variable.
5640 * systemctl gained a new "-P" switch that is a shortcut for "--value
5643 * "systemctl list-units" and "systemctl list-machines" no longer hide
5644 their first output column with --no-legend. To hide the first column,
5647 * "systemctl reboot" takes the option "--reboot-argument=".
5648 The optional positional argument to "systemctl reboot" is now
5649 being deprecated in favor of this option.
5651 * systemd-run gained a new switch --slice-inherit. If specified the
5652 unit it generates is placed in the same slice as the systemd-run
5655 * systemd-journald gained support for zstd compression of large fields
5656 in journal files. The hash tables in journal files have been hardened
5657 against hash collisions. This is an incompatible change and means
5658 that journal files created with new systemd versions are not readable
5659 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
5660 environment variable for systemd-journald.service is set to 0 this
5661 new hardening functionality may be turned off, so that generated
5662 journal files remain compatible with older journalctl
5665 * journalctl will now include a clickable link in the default output for
5666 each log message for which a URL with further documentation is
5667 known. This is only supported on terminal emulators that support
5668 clickable hyperlinks, and is turned off if a pager is used (since
5669 "less" still doesn't support hyperlinks,
5670 unfortunately). Documentation URLs may be included in log messages
5671 either by including a DOCUMENTATION= journal field in it, or by
5672 associating a journal message catalog entry with the log message's
5673 MESSAGE_ID, which then carries a "Documentation:" tag.
5675 * journald.conf gained a new boolean setting Audit= that may be used to
5676 control whether systemd-journald will enable audit during
5679 * when systemd-journald's log stream is broken up into multiple lines
5680 because the PID of the sender changed this is indicated in the
5681 generated log records via the _LINE_BREAK=pid-change field.
5683 * journalctl's "-o cat" output mode will now show one or more journal
5684 fields specified with --output-fields= instead of unconditionally
5685 MESSAGE=. This is useful to retrieve a very specific set of fields
5686 without any decoration.
5688 * The sd-journal.h API gained two new functions:
5689 sd_journal_enumerate_available_unique() and
5690 sd_journal_enumerate_available_data() that operate like their
5691 counterparts that lack the _available_ in the name, but skip items
5692 that cannot be read and processed by the local implementation
5693 (i.e. are compressed in an unsupported format or such),
5695 * coredumpctl gained a new --file= switch, matching the same one in
5696 journalctl: a specific journal file may be specified to read the
5699 * coredumps collected by systemd-coredump may now be compressed using
5702 * systemd-binfmt gained a new switch --unregister for unregistering all
5703 registered entries at once. This is now invoked automatically at
5704 shutdown, so that binary formats registered with the "F" flag will
5705 not block clean file system unmounting.
5707 * systemd-notify's --pid= switch gained new values: "parent", "self",
5708 "auto" for controlling which PID to send to the service manager: the
5709 systemd-notify process' PID, or the one of the process invoking it.
5711 * systemd-logind's Session bus object learnt a new method call
5712 SetType() for temporarily updating the session type of an already
5713 allocated session. This is useful for upgrading tty sessions to
5714 graphical ones once a compositor is invoked.
5716 * systemd-socket-proxy gained a new switch --exit-idle-time= for
5717 configuring an exit-on-idle time.
5719 * systemd-repart's --empty= setting gained a new value "create". If
5720 specified a new empty regular disk image file is created under the
5721 specified name. Its size may be specified with the new --size=
5722 option. The latter is also supported without the "create" mode, in
5723 order to grow existing disk image files to the specified size. These
5724 two new options are useful when creating or manipulating disk images
5725 instead of operating on actual block devices.
5727 * systemd-repart drop-ins now support a new UUID= setting to control
5728 the UUID to assign to a newly created partition.
5730 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
5733 * systemd-repart's Label= setting now support the usual, simple
5734 specifier expansion.
5736 * systemd-homed's LUKS backend gained the ability to discard empty file
5737 system blocks automatically when the user logs out. This is enabled
5738 by default to ensure that home directories take minimal space when
5739 logged out but get full size guarantees when logged in. This may be
5740 controlled with the new --luks-offline-discard= switch to homectl.
5742 * If systemd-homed detects that /home/ is encrypted as a whole it will
5743 now default to the directory or subvolume backends instead of the
5744 LUKS backend, in order to avoid double encryption. The default
5745 storage and file system may now be configured explicitly, too, via
5746 the new /etc/systemd/homed.conf configuration file.
5748 * systemd-homed now supports unlocking home directories with FIDO2
5749 security tokens that support the 'hmac-secret' extension, in addition
5750 to the existing support for PKCS#11 security token unlocking
5751 support. Note that many recent hardware security tokens support both
5752 interfaces. The FIDO2 support is accessible via homectl's
5753 --fido2-device= option.
5755 * homectl's --pkcs11-uri= setting now accepts two special parameters:
5756 if "auto" is specified and only one suitable PKCS#11 security token
5757 is plugged in, its URL is automatically determined and enrolled for
5758 unlocking the home directory. If "list" is specified a brief table of
5759 suitable PKCS#11 security tokens is shown. Similar, the new
5760 --fido2-device= option also supports these two special values, for
5761 automatically selecting and listing suitable FIDO2 devices.
5763 * The /etc/crypttab tmp option now optionally takes an argument
5764 selecting the file system to use. Moreover, the default is now
5765 changed from ext2 to ext4.
5767 * There's a new /etc/crypttab option "keyfile-erase". If specified the
5768 key file listed in the same line is removed after use, regardless if
5769 volume activation was successful or not. This is useful if the key
5770 file is only acquired transiently at runtime and shall be erased
5771 before the system continues to boot.
5773 * There's also a new /etc/crypttab option "try-empty-password". If
5774 specified, before asking the user for a password it is attempted to
5775 unlock the volume with an empty password. This is useful for
5776 installing encrypted images whose password shall be set on first boot
5777 instead of at installation time.
5779 * systemd-cryptsetup will now attempt to load the keys to unlock
5780 volumes with automatically from files in
5781 /etc/cryptsetup-keys.d/<volume>.key and
5782 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
5784 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
5785 /etc/crypttab, during boot.
5787 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
5788 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
5791 * A new generator systemd-xdg-autostart-generator has been added. It
5792 generates systemd unit files from XDG autostart .desktop files, and
5793 may be used to let the systemd user instance manage services that are
5794 started automatically as part of the desktop session.
5796 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
5797 to query and change the firmware's 'Reboot Into Firmware Interface'
5800 * systemd-firstboot gained a new switch --kernel-command-line= that may
5801 be used to initialize the /etc/kernel/cmdline file of the image. It
5802 also gained a new switch --root-password-hashed= which is like
5803 --root-password= but accepts a pre-hashed UNIX password as
5804 argument. The new option --delete-root-password may be used to unset
5805 any password for the root user (dangerous!). The --root-shell= switch
5806 may be used to control the shell to use for the root account. A new
5807 --force option may be used to override any already set settings with
5808 the parameters specified on the command line (by default, the tool
5809 will not override what has already been set before, i.e. is purely
5812 * systemd-firstboot gained support for a new --image= switch, which is
5813 similar to --root= but accepts the path to a disk image file, on
5814 which it then operates.
5816 * A new sd-path.h API has been added to libsystemd. It provides a
5817 simple API for retrieving various search paths and primary
5818 directories for various resources.
5820 * A new call sd_notify_barrier() has been added to the sd-daemon.h
5821 API. The call will block until all previously sent sd_notify()
5822 messages have been processed by the service manager. This is useful
5823 to remove races caused by a process already having disappeared at the
5824 time a notification message is processed by the service manager,
5825 making correct attribution impossible. The systemd-notify tool will
5826 now make use of this call implicitly, but this can be turned off again
5827 via the new --no-block switch.
5829 * When sending a file descriptor (fd) to the service manager to keep
5830 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
5831 may be specified. If passed the service manager will refrain from
5832 poll()ing on the file descriptor. Traditionally (and when the
5833 parameter is not specified), the service manager will poll it for
5834 POLLHUP or POLLERR events, and immediately close the fds in that
5837 * The service manager (PID1) gained a new D-Bus method call
5838 SetShowStatus() which may be used to control whether it shall show
5839 boot-time status output on the console. This method has a similar
5840 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
5842 * The sd-bus API gained a number of convenience functions that take
5843 va_list arguments rather than "...". For example, there's now
5844 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
5845 it easier to build wrappers that accept variadic arguments and want
5846 to pass a ready va_list structure to sd-bus.
5848 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
5849 flag which alters how the userdata pointer to pass to the callbacks
5850 is determined. When the flag is set, the offset field is converted
5851 as-is into a pointer, without adding it to the object pointer the
5852 vtable is associated with.
5854 * sd-bus now exposes four new functions:
5855 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
5856 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
5857 validate strings to check if they qualify as various D-Bus concepts.
5859 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
5860 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
5861 that simplify adding argument names to D-Bus methods and signals.
5863 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
5865 * Various D-Bus APIs of systemd daemons now have man pages that
5866 document the methods, signals and properties.
5868 * The expectations on user/group name syntax are now documented in
5869 detail; documentation on how classic home directories may be
5870 converted into home directories managed by homed has been added;
5871 documentation regarding integration of homed/userdb functionality in
5872 desktops has been added:
5874 https://systemd.io/USER_NAMES
5875 https://systemd.io/CONVERTING_TO_HOMED
5876 https://systemd.io/USERDB_AND_DESKTOPS
5878 * Documentation for the on-disk Journal file format has been updated
5879 and has now moved to:
5881 https://systemd.io/JOURNAL_FILE_FORMAT
5883 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
5884 has been extended by a set of environment variables that expose
5885 select fields from the host's os-release file to the container
5886 payload. Similarly, host's os-release files can be mounted into the
5887 container underneath /run/host. Together, those mechanisms provide a
5888 standardized way to expose information about the host to the
5889 container payload. Both interfaces are implemented in systemd-nspawn.
5891 * All D-Bus services shipped in systemd now implement the generic
5892 LogControl1 D-Bus API which allows clients to change log level +
5893 target of the service during runtime.
5895 * Only relevant for developers: the mkosi.default symlink has been
5896 dropped from version control. Please create a symlink to one of the
5897 distribution-specific defaults in .mkosi/ based on your preference.
5899 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
5900 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
5901 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
5902 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
5903 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
5904 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
5905 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
5906 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
5907 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
5908 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
5909 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
5910 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
5911 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
5912 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
5913 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
5914 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
5915 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
5916 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
5917 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
5918 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
5919 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
5920 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
5921 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
5922 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
5923 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
5924 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
5925 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
5926 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
5927 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
5928 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
5929 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
5930 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
5931 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
5932 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
5933 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
5934 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
5935 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
5936 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
5938 – Warsaw, 2020-07-30
5942 * A new tool "systemd-repart" has been added, that operates as an
5943 idempotent declarative repartitioner for GPT partition tables.
5944 Specifically, a set of partitions that must or may exist can be
5945 configured via drop-in files, and during every boot the partition
5946 table on disk is compared with these files, creating missing
5947 partitions or growing existing ones based on configurable relative
5948 and absolute size constraints. The tool is strictly incremental,
5949 i.e. does not delete, shrink or move partitions, but only adds and
5950 grows them. The primary use-case is OS images that ship in minimized
5951 form, that on first boot are grown to the size of the underlying
5952 block device or augmented with additional partitions. For example,
5953 the root partition could be extended to cover the whole disk, or a
5954 swap or /home partitions could be added on first boot. It can also be
5955 used for systems that use an A/B update scheme but ship images with
5956 just the A partition, with B added on first boot. The tool is
5957 primarily intended to be run in the initrd, shortly before
5958 transitioning into the host OS, but can also be run after the
5959 transition took place. It automatically discovers the disk backing
5960 the root file system, and should hence not require any additional
5961 configuration besides the partition definition drop-ins. If no
5962 configuration drop-ins are present, no action is taken.
5964 * A new component "userdb" has been added, along with a small daemon
5965 "systemd-userdbd.service" and a client tool "userdbctl". The framework
5966 allows defining rich user and group records in a JSON format,
5967 extending on the classic "struct passwd" and "struct group"
5968 structures. Various components in systemd have been updated to
5969 process records in this format, including systemd-logind and
5970 pam-systemd. The user records are intended to be extensible, and
5971 allow setting various resource management, security and runtime
5972 parameters that shall be applied to processes and sessions of the
5973 user as they log in. This facility is intended to allow associating
5974 such metadata directly with user/group records so that they can be
5975 produced, extended and consumed in unified form. We hope that
5976 eventually frameworks such as sssd will generate records this way, so
5977 that for the first time resource management and various other
5978 per-user settings can be configured in LDAP directories and then
5979 provided to systemd (specifically to systemd-logind and pam-system)
5980 to apply on login. For further details see:
5982 https://systemd.io/USER_RECORD
5983 https://systemd.io/GROUP_RECORD
5984 https://systemd.io/USER_GROUP_API
5986 * A small new service systemd-homed.service has been added, that may be
5987 used to securely manage home directories with built-in encryption.
5988 The complete user record data is unified with the home directory,
5989 thus making home directories naturally migratable. Its primary
5990 back-end is based on LUKS volumes, but fscrypt, plain directories,
5991 and other storage schemes are also supported. This solves a couple of
5992 problems we saw with traditional ways to manage home directories, in
5993 particular when it comes to encryption. For further discussion of
5994 this, see the video of Lennart's talk at AllSystemsGo! 2019:
5996 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
5998 For further details about the format and expectations on home
5999 directories this new daemon makes, see:
6001 https://systemd.io/HOME_DIRECTORY
6003 * systemd-journald is now multi-instantiable. In addition to the main
6004 instance systemd-journald.service there's now a template unit
6005 systemd-journald@.service, with each instance defining a new named
6006 log 'namespace' (whose name is specified via the instance part of the
6007 unit name). A new unit file setting LogNamespace= has been added,
6008 taking such a namespace name, that assigns services to the specified
6009 log namespaces. As each log namespace is serviced by its own
6010 independent journal daemon, this functionality may be used to improve
6011 performance and increase isolation of applications, at the price of
6012 losing global message ordering. Each instance of journald has a
6013 separate set of configuration files, with possibly different disk
6014 usage limitations and other settings.
6016 journalctl now takes a new option --namespace= to show logs from a
6017 specific log namespace. The sd-journal.h API gained
6018 sd_journal_open_namespace() for opening the log stream of a specific
6019 log namespace. systemd-journald also gained the ability to exit on
6020 idle, which is useful in the context of log namespaces, as this means
6021 log daemons for log namespaces can be activated automatically on
6022 demand and will stop automatically when no longer used, minimizing
6025 * When systemd-tmpfiles copies a file tree using the 'C' line type it
6026 will now label every copied file according to the SELinux database.
6028 * When systemd/PID 1 detects it is used in the initrd it will now boot
6029 into initrd.target rather than default.target by default. This should
6030 make it simpler to build initrds with systemd as for many cases the
6031 only difference between a host OS image and an initrd image now is
6032 the presence of the /etc/initrd-release file.
6034 * A new kernel command line option systemd.cpu_affinity= is now
6035 understood. It's equivalent to the CPUAffinity= option in
6036 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
6037 itself and the default for all other processes.
6039 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
6040 equivalent), the SELinux database is now reloaded, ensuring that
6041 sockets and other file system objects are generated taking the new
6042 database into account.
6044 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
6045 "quiet" has been changed to imply that instead of
6046 "systemd.show-status=auto". In this mode, only messages about errors
6047 and significant delays in boot are shown on the console.
6049 * The sd-event.h API gained native support for the new Linux "pidfd"
6050 concept. This permits watching processes using file descriptors
6051 instead of PID numbers, which fixes a number of races and makes
6052 process supervision more robust and efficient. All of systemd's
6053 components will now use pidfds if the kernel supports it for process
6054 watching, with the exception of PID 1 itself, unfortunately. We hope
6055 to move PID 1 to exclusively using pidfds too eventually, but this
6056 requires some more kernel work first. (Background: PID 1 watches
6057 processes using waitid() with the P_ALL flag, and that does not play
6058 together nicely with pidfds yet.)
6060 * Closely related to this, the sd-event.h API gained two new calls
6061 sd_event_source_send_child_signal() (for sending a signal to a
6062 watched process) and sd_event_source_get_child_process_own() (for
6063 marking a process so that it is killed automatically whenever the
6064 event source watching it is freed).
6066 * systemd-networkd gained support for configuring Token Bucket Filter
6067 (TBF) parameters in its qdisc configuration support. Similarly,
6068 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
6069 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
6071 * systemd-networkd gained support for Intermediate Functional Block
6072 (IFB) network devices.
6074 * systemd-networkd gained support for configuring multi-path IP routes,
6075 using the new MultiPathRoute= setting in the [Route] section.
6077 * systemd-networkd's DHCPv4 client has been updated to support a new
6078 SendDecline= option. If enabled, duplicate address detection is done
6079 after a DHCP offer is received from the server. If a conflict is
6080 detected, the address is declined. The DHCPv4 client also gained
6081 support for a new RouteMTUBytes= setting that allows to configure the
6082 MTU size to be used for routes generated from DHCPv4 leases.
6084 * The PrefixRoute= setting in systemd-networkd's [Address] section of
6085 .network files has been deprecated, and replaced by AddPrefixRoute=,
6086 with its sense inverted.
6088 * The Gateway= setting of [Route] sections of .network files gained
6089 support for a special new value "_dhcp". If set, the configured
6090 static route uses the gateway host configured via DHCP.
6092 * New User= and SuppressPrefixLength= settings have been implemented
6093 for the [RoutingPolicyRule] section of .network files to configure
6094 source routing based on UID ranges and prefix length, respectively.
6096 * The Type= match property of .link files has been generalized to
6097 always match the device type shown by 'networkctl status', even for
6098 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
6101 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
6102 D-Bus message object as "sensitive". Those objects are erased from
6103 memory when they are freed. This concept is intended to be used for
6104 messages that contain security sensitive data. A new flag
6105 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
6106 in sd-bus vtables, causing any incoming and outgoing messages of
6107 those methods to be implicitly marked as "sensitive".
6109 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
6110 contents of a message (or parts thereof) to standard output for
6113 * systemd-sysusers gained support for creating users with the primary
6114 group named differently than the user.
6116 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
6117 gained support for growing XFS partitions. Previously it supported
6118 only ext4 and btrfs partitions.
6120 * The support for /etc/crypttab gained a new x-initrd.attach option. If
6121 set, the specified encrypted volume is unlocked already in the
6122 initrd. This concept corresponds to the x-initrd.mount option in
6125 * systemd-cryptsetup gained native support for unlocking encrypted
6126 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
6127 encryption of volumes to YubiKeys. This is exposed in the new
6128 pkcs11-uri= option in /etc/crypttab.
6130 * The /etc/fstab support in systemd now supports two new mount options
6131 x-systemd.{required,wanted}-by=, for explicitly configuring the units
6132 that the specified mount shall be pulled in by, in place of
6133 the usual local-fs.target/remote-fs.target.
6135 * The https://systemd.io/ web site has been relaunched, directly
6136 populated with most of the documentation included in the systemd
6137 repository. systemd also acquired a new logo, thanks to Tobias
6140 * systemd-udevd gained support for managing "alternative" network
6141 interface names, as supported by new Linux kernels. For the first
6142 time this permits assigning multiple (and longer!) names to a network
6143 interface. systemd-udevd will now by default assign the names
6144 generated via all supported naming schemes to each interface. This
6145 may be further tweaked with .link files and the AlternativeName= and
6146 AlternativeNamesPolicy= settings. Other components of systemd have
6147 been updated to support the new alternative names wherever
6148 appropriate. For example, systemd-nspawn will now generate
6149 alternative interface names for the host-facing side of container
6150 veth links based on the full container name without truncation.
6152 * systemd-nspawn interface naming logic has been updated in another way
6153 too: if the main interface name (i.e. as opposed to new-style
6154 "alternative" names) based on the container name is truncated, a
6155 simple hashing scheme is used to give different interface names to
6156 multiple containers whose names all begin with the same prefix. Since
6157 this changes the primary interface names pointing to containers if
6158 truncation happens, the old scheme may still be requested by
6159 selecting an older naming scheme, via the net.naming-scheme= kernel
6160 command line option.
6162 * PrivateUsers= in service files now works in services run by the
6163 systemd --user per-user instance of the service manager.
6165 * A new per-service sandboxing option ProtectClock= has been added that
6166 locks down write access to the system clock. It takes away device
6167 node access to /dev/rtc as well as the system calls that set the
6168 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
6169 Note that this option does not affect access to auxiliary services
6170 that allow changing the clock, for example access to
6173 * The systemd-id128 tool gained a new "show" verb for listing or
6174 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
6175 GPT partition table types.
6177 * The Discoverable Partitions Specification has been updated to support
6178 /var and /var/tmp partition discovery. Support for this has been
6179 added to systemd-gpt-auto-generator. For details see:
6181 https://systemd.io/DISCOVERABLE_PARTITIONS
6183 * "systemctl list-unit-files" has been updated to show a new column
6184 with the suggested enablement state based on the vendor preset files
6185 for the respective units.
6187 * "systemctl" gained a new option "--with-dependencies". If specified
6188 commands such as "systemctl status" or "systemctl cat" will now show
6189 all specified units along with all units they depend on.
6191 * networkctl gained support for showing per-interface logs in its
6194 * systemd-networkd-wait-online gained support for specifying the maximum
6195 operational state to wait for, and to wait for interfaces to
6198 * The [Match] section of .link and .network files now supports a new
6199 option PermanentMACAddress= which may be used to check against the
6200 permanent MAC address of a network device even if a randomized MAC
6203 * The [TrafficControlQueueingDiscipline] section in .network files has
6204 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
6205 dropped from the individual setting names.
6207 * Any .link and .network files that have an empty [Match] section (this
6208 also includes empty and commented-out files) will now be
6209 rejected. systemd-udev and systemd-networkd started warning about
6210 such files in version 243.
6212 * systemd-logind will now validate access to the operation of changing
6213 the virtual terminal via a polkit action. By default, only users
6214 with at least one session on a local VT are granted permission.
6216 * When systemd sets up PAM sessions that invoked service processes
6217 shall run in, the pam_setcred() API is now invoked, thus permitting
6218 PAM modules to set additional credentials for the processes.
6220 * portablectl attach/detach verbs now accept --now and --enable options
6221 to combine attachment with enablement and invocation, or detachment
6222 with stopping and disablement.
6224 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
6225 fixed, which in turn exposed bugs in unit configuration of services
6226 which have Type=oneshot and should only run once, but do not have
6227 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
6228 service may be started again after exiting successfully, for example
6229 as a dependency in another transaction. Affected services included
6230 some internal systemd services (most notably
6231 systemd-vconsole-setup.service, which was updated to have
6232 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
6233 plymouth has been suitably updated or patched before upgrading to
6234 this systemd release. See
6235 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
6236 additional discussion.
6238 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
6239 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
6240 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
6241 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
6242 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
6243 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
6244 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
6245 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
6246 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
6247 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
6248 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
6249 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
6250 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
6251 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
6252 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
6253 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
6254 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
6255 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
6256 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
6257 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
6258 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
6259 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
6260 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
6261 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
6264 – Warsaw, 2020-03-06
6268 * Support for the cpuset cgroups v2 controller has been added.
6269 Processes may be restricted to specific CPUs using the new
6270 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
6271 AllowedMemoryNodes= setting.
6273 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
6274 now be configured using a new RestartKillSignal= setting. This
6275 allows units which signals to request termination to implement
6276 different behaviour when stopping in preparation for a restart.
6278 * "systemctl clean" may now be used also for socket, mount, and swap
6281 * systemd will also read configuration options from the EFI variable
6282 SystemdOptions. This may be used to configure systemd behaviour when
6283 modifying the kernel command line is inconvenient, but configuration
6284 on disk is read too late, for example for the options related to
6285 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
6286 set the EFI variable.
6288 * systemd will now disable printk ratelimits in early boot. This should
6289 allow us to capture more logs from the early boot phase where normal
6290 storage is not available and the kernel ring buffer is used for
6291 logging. Configuration on the kernel command line has higher priority
6292 and overrides the systemd setting.
6294 systemd programs which log to /dev/kmsg directly use internal
6295 ratelimits to prevent runaway logging. (Normally this is only used
6296 during early boot, so in practice this change has very little
6299 * Unit files now support top level dropin directories of the form
6300 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
6301 that affects all corresponding unit files.
6303 * systemctl gained support for 'stop --job-mode=triggering' which will
6304 stop the specified unit and any units which could trigger it.
6306 * Unit status display now includes units triggering and triggered by
6307 the unit being shown.
6309 * The RuntimeMaxSec= setting is now supported by scopes, not just
6310 .service units. This is particularly useful for PAM sessions which
6311 create a scope unit for the user login. systemd.runtime_max_sec=
6312 setting may used with the pam_systemd module to limit the duration
6313 of the PAM session, for example for time-limited logins.
6315 * A new @pkey system call group is now defined to make it easier to
6316 allow-list memory protection syscalls for containers and services
6317 which need to use them.
6319 * systemd-udevd: removed the 30s timeout for killing stale workers on
6320 exit. systemd-udevd now waits for workers to finish. The hard-coded
6321 exit timeout of 30s was too short for some large installations, where
6322 driver initialization could be prematurely interrupted during initrd
6323 processing if the root file system had been mounted and init was
6324 preparing to switch root. If udevd is run without systemd and workers
6325 are hanging while udevd receives an exit signal, udevd will now exit
6326 when udev.event_timeout is reached for the last hanging worker. With
6327 systemd, the exit timeout can additionally be configured using
6328 TimeoutStopSec= in systemd-udevd.service.
6330 * udev now provides a program (fido_id) that identifies FIDO CTAP1
6331 ("U2F")/CTAP2 security tokens based on the usage declared in their
6332 report and descriptor and outputs suitable environment variables.
6333 This replaces the externally maintained allow lists of all known
6334 security tokens that were used previously.
6336 * Automatically generated autosuspend udev rules for allow-listed
6337 devices have been imported from the Chromium OS project. This should
6338 improve power saving with many more devices.
6340 * udev gained a new "CONST{key}=value" setting that allows matching
6341 against system-wide constants without forking a helper binary.
6342 Currently "arch" and "virt" keys are supported.
6344 * udev now opens CDROMs in non-exclusive mode when querying their
6345 capabilities. This should fix issues where other programs trying to
6346 use the CDROM cannot gain access to it, but carries a risk of
6347 interfering with programs writing to the disk, if they did not open
6348 the device in exclusive mode as they should.
6350 * systemd-networkd does not create a default route for IPv4 link local
6351 addressing anymore. The creation of the route was unexpected and was
6352 breaking routing in various cases, but people who rely on it being
6353 created implicitly will need to adjust. Such a route may be requested
6354 with DefaultRouteOnDevice=yes.
6356 Similarly, systemd-networkd will not assign a link-local IPv6 address
6357 when IPv6 link-local routing is not enabled.
6359 * Receive and transmit buffers may now be configured on links with
6360 the new RxBufferSize= and TxBufferSize= settings.
6362 * systemd-networkd may now advertise additional IPv6 routes. A new
6363 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
6366 * systemd-networkd may now configure "next hop" routes using the
6367 [NextHop] section and Gateway= and Id= settings.
6369 * systemd-networkd will now retain DHCP config on restarts by default
6370 (but this may be overridden using the KeepConfiguration= setting).
6371 The default for SendRelease= has been changed to true.
6373 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
6374 received from the server.
6376 The client will use the received SIP server list if UseSIP=yes is
6379 The client may be configured to request specific options from the
6380 server using a new RequestOptions= setting.
6382 The client may be configured to send arbitrary options to the server
6383 using a new SendOption= setting.
6385 A new IPServiceType= setting has been added to configure the "IP
6386 service type" value used by the client.
6388 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
6389 request prefix hints in the DHCPv6 solicitation.
6391 * The DHCPv4 server may be configured to send arbitrary options using
6392 a new SendOption= setting.
6394 * The DHCPv4 server may now be configured to emit SIP server list using
6395 the new EmitSIP= and SIP= settings.
6397 * systemd-networkd and networkctl may now renew DHCP leases on demand.
6398 networkctl has a new 'networkctl renew' verb.
6400 * systemd-networkd may now reconfigure links on demand. networkctl
6401 gained two new verbs: "reload" will reload the configuration, and
6402 "reconfigure DEVICE…" will reconfigure one or more devices.
6404 * .network files may now match on SSID and BSSID of a wireless network,
6405 i.e. the access point name and hardware address using the new SSID=
6406 and BSSID= options. networkctl will display the current SSID and
6407 BSSID for wireless links.
6409 .network files may also match on the wireless network type using the
6410 new WLANInterfaceType= option.
6412 * systemd-networkd now includes default configuration that enables
6413 link-local addressing when connected to an ad-hoc wireless network.
6415 * systemd-networkd may configure the Traffic Control queueing
6416 disciplines in the kernel using the new
6417 [TrafficControlQueueingDiscipline] section and Parent=,
6418 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
6419 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
6420 NetworkEmulatorDuplicateRate= settings.
6422 * systemd-tmpfiles gained a new w+ setting to append to files.
6424 * systemd-analyze dump will now report when the memory configuration in
6425 the kernel does not match what systemd has configured (usually,
6426 because some external program has modified the kernel configuration
6429 * systemd-analyze gained a new --base-time= switch instructs the
6430 'calendar' verb to resolve times relative to that timestamp instead
6431 of the present time.
6433 * journalctl --update-catalog now produces deterministic output (making
6434 reproducible image builds easier).
6436 * A new devicetree-overlay setting is now documented in the Boot Loader
6439 * The default value of the WatchdogSec= setting used in systemd
6440 services (the ones bundled with the project itself) may be set at
6441 configuration time using the -Dservice-watchdog= setting. If set to
6442 empty, the watchdogs will be disabled.
6444 * systemd-resolved validates IP addresses in certificates now when GnuTLS
6447 * libcryptsetup >= 2.0.1 is now required.
6449 * A configuration option -Duser-path= may be used to override the $PATH
6450 used by the user service manager. The default is again to use the same
6451 path as the system manager.
6453 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
6454 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
6457 * Service units gained a new sandboxing option ProtectKernelLogs= which
6458 makes sure the program cannot get direct access to the kernel log
6459 buffer anymore, i.e. the syslog() system call (not to be confused
6460 with the API of the same name in libc, which is not affected), the
6461 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
6462 inaccessible to the service. It's recommended to enable this setting
6463 for all services that should not be able to read from or write to the
6464 kernel log buffer, which are probably almost all.
6466 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
6467 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
6468 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
6469 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
6470 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
6471 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
6472 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
6473 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
6474 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
6475 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
6476 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
6477 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
6478 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
6479 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
6480 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
6481 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
6482 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
6483 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
6484 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
6485 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
6486 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
6487 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
6488 Zach Smith, Zbigniew Jędrzejewski-Szmek
6490 – Warsaw, 2019-11-29
6494 * This release enables unprivileged programs (i.e. requiring neither
6495 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
6496 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
6497 kernel for the whole UNIX group range, i.e. all processes. This
6498 change should be reasonably safe, as the kernel support for it was
6499 specifically implemented to allow safe access to ICMP Echo for
6500 processes lacking any privileges. If this is not desirable, it can be
6501 disabled again by setting the parameter to "1 0".
6503 * Previously, filters defined with SystemCallFilter= would have the
6504 effect that any calling of an offending system call would terminate
6505 the calling thread. This behaviour never made much sense, since
6506 killing individual threads of unsuspecting processes is likely to
6507 create more problems than it solves. With this release the default
6508 action changed from killing the thread to killing the whole
6509 process. For this to work correctly both a kernel version (>= 4.14)
6510 and a libseccomp version (>= 2.4.0) supporting this new seccomp
6511 action is required. If an older kernel or libseccomp is used the old
6512 behaviour continues to be used. This change does not affect any
6513 services that have no system call filters defined, or that use
6514 SystemCallErrorNumber= (and thus see EPERM or another error instead
6515 of being killed when calling an offending system call). Note that
6516 systemd documentation always claimed that the whole process is
6517 killed. With this change behaviour is thus adjusted to match the
6520 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
6521 4194304 by default, i.e. the full 22bit range the kernel allows, up
6522 from the old 16-bit range. This should improve security and
6523 robustness, as PID collisions are made less likely (though certainly
6524 still possible). There are rumours this might create compatibility
6525 problems, though at this moment no practical ones are known to
6526 us. Downstream distributions are hence advised to undo this change in
6527 their builds if they are concerned about maximum compatibility, but
6528 for everybody else we recommend leaving the value bumped. Besides
6529 improving security and robustness this should also simplify things as
6530 the maximum number of allowed concurrent tasks was previously bounded
6531 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
6532 only a single knob is left ("kernel.threads-max"). There have been
6533 concerns that usability is affected by this change because larger PID
6534 numbers are harder to type, but we believe the change from 5 digits
6535 to 7 digits doesn't hamper usability.
6537 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
6538 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
6539 hierarchically set default memory protection values for a particular
6540 subtree of the unit hierarchy.
6542 * Memory protection directives can now take a value of zero, allowing
6543 explicit opting out of a default value propagated by an ancestor.
6545 * systemd now defaults to the "unified" cgroup hierarchy setup during
6546 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
6547 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
6548 change reflects the fact that cgroupsv2 support has matured
6549 substantially in both systemd and in the kernel, and is clearly the
6550 way forward. Downstream production distributions might want to
6551 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
6552 their builds as unfortunately the popular container managers have not
6553 caught up with the kernel API changes.
6555 * Man pages are not built by default anymore (html pages were already
6556 disabled by default), to make development builds quicker. When
6557 building systemd for a full installation with documentation, meson
6558 should be called with -Dman=true and/or -Dhtml=true as appropriate.
6559 The default was changed based on the assumption that quick one-off or
6560 repeated development builds are much more common than full optimized
6561 builds for installation, and people need to pass various other
6562 options to when doing "proper" builds anyway, so the gain from making
6563 development builds quicker is bigger than the one time disruption for
6566 Two scripts are created in the *build* directory to generate and
6567 preview man and html pages on demand, e.g.:
6569 build/man/man systemctl
6570 build/man/html systemd.index
6572 * libidn2 is used by default if both libidn2 and libidn are installed.
6573 Please use -Dlibidn=true if libidn is preferred.
6575 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
6576 big-endian machines. Before, bytes were written and read in native
6577 machine order as exposed by the native libc __cpu_mask interface.
6578 Now, little-endian order is always used (CPUs 0–7 are described by
6579 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
6580 This change fixes D-Bus calls that cross endianness boundary.
6582 The presentation format used for CPUAffinity= by "systemctl show" and
6583 "systemd-analyze dump" is changed to present CPU indices instead of
6584 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
6585 shown as CPUAffinity=03000000000000000000000000000… (on
6586 little-endian) or CPUAffinity=00000000000000300000000000000… (on
6587 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
6588 input format. The maximum integer that will be printed in the new
6589 format is 8191 (four digits), while the old format always used a very
6590 long number (with the length varying by architecture), so they can be
6591 unambiguously distinguished.
6593 * /usr/sbin/halt.local is no longer supported. Implementation in
6594 distributions was inconsistent and it seems this functionality was
6597 To replace this functionality, users should:
6598 - either define a new unit and make it a dependency of final.target
6599 (systemctl add-wants final.target my-halt-local.service)
6600 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
6601 and ensure that it accepts "halt", "poweroff", "reboot", and
6602 "kexec" as an argument, see the description in systemd-shutdown(8).
6604 * When a [Match] section in .link or .network file is empty (contains
6605 no match patterns), a warning will be emitted. Please add any "match
6606 all" pattern instead, e.g. OriginalName=* or Name=* in case all
6607 interfaces should really be matched.
6609 * A new setting NUMAPolicy= may be used to set process memory
6610 allocation policy. This setting can be specified in
6611 /etc/systemd/system.conf and hence will set the default policy for
6612 PID1. The default policy can be overridden on a per-service
6613 basis. The related setting NUMAMask= is used to specify NUMA node
6614 mask that should be associated with the selected policy.
6616 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
6617 generates when processes it manages are reaching their memory limits,
6618 and will place their units in a special state, and optionally kill or
6619 stop the whole unit.
6621 * The service manager will now expose bus properties for the IO
6622 resources used by units. This information is also shown in "systemctl
6623 status" now (for services that have IOAccounting=yes set). Moreover,
6624 the IO accounting data is included in the resource log message
6625 generated whenever a unit stops.
6627 * Units may now configure an explicit timeout to wait for when killed
6628 with SIGABRT, for example when a service watchdog is hit. Previously,
6629 the regular TimeoutStopSec= timeout was applied in this case too —
6630 now a separate timeout may be set using TimeoutAbortSec=.
6632 * Services may now send a special WATCHDOG=trigger message with
6633 sd_notify() to trigger an immediate "watchdog missed" event, and thus
6634 trigger service termination. This is useful both for testing watchdog
6635 handling, but also for defining error paths in services, that shall
6636 be handled the same way as watchdog events.
6638 * There are two new per-unit settings IPIngressFilterPath= and
6639 IPEgressFilterPath= which allow configuration of a BPF program
6640 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
6641 to apply to the IP packet ingress/egress path of all processes of a
6642 unit. This is useful to allow running systemd services with BPF
6643 programs set up externally.
6645 * systemctl gained a new "clean" verb for removing the state, cache,
6646 runtime or logs directories of a service while it is terminated. The
6647 new verb may also be used to remove the state maintained on disk for
6648 timer units that have Persistent= configured.
6650 * During the last phase of shutdown systemd will now automatically
6651 increase the log level configured in the "kernel.printk" sysctl so
6652 that any relevant loggable events happening during late shutdown are
6653 made visible. Previously, loggable events happening so late during
6654 shutdown were generally lost if the "kernel.printk" sysctl was set to
6655 high thresholds, as regular logging daemons are terminated at that
6656 time and thus nothing is written to disk.
6658 * If processes terminated during the last phase of shutdown do not exit
6659 quickly systemd will now show their names after a short time, to make
6660 debugging easier. After a longer timeout they are forcibly killed,
6663 * journalctl (and the other tools that display logs) will now highlight
6664 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
6665 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
6666 are now shown in blue color, to separate them visually from regular
6667 logs. References to configuration files are now turned into clickable
6668 links on terminals that support that.
6670 * systemd-journald will now stop logging to /var/log/journal during
6671 shutdown when /var/ is on a separate mount, so that it can be
6672 unmounted safely during shutdown.
6674 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
6676 * systemd-resolved "Cache=" configuration option in resolved.conf has
6677 been extended to also accept the 'no-negative' value. Previously,
6678 only a boolean option was allowed (yes/no), having yes as the
6679 default. If this option is set to 'no-negative', negative answers are
6680 not cached while the old cache heuristics are used positive answers.
6681 The default remains unchanged.
6683 * The predictable naming scheme for network devices now supports
6684 generating predictable names for "netdevsim" devices.
6686 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
6689 Those two changes form a new net.naming-policy-scheme= entry.
6690 Distributions which want to preserve naming stability may want to set
6691 the -Ddefault-net-naming-scheme= configuration option.
6693 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
6694 interfaces natively.
6696 * systemd-networkd's bridge FDB support now allows configuration of a
6697 destination address for each entry (Destination=), as well as the
6698 VXLAN VNI (VNI=), as well as an option to declare what an entry is
6699 associated with (AssociatedWith=).
6701 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
6702 option for configuring the maximum number of DHCP lease requests. It
6703 also learnt a new BlackList= option for deny-listing DHCP servers (a
6704 similar setting has also been added to the IPv6 RA client), as well
6705 as a SendRelease= option for configuring whether to send a DHCP
6706 RELEASE message when terminating.
6708 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
6709 separately in the [DHCPv4] and [DHCPv6] sections.
6711 * systemd-networkd's DHCP support will now optionally create an
6712 implicit host route to the DNS server specified in the DHCP lease, in
6713 addition to the routes listed explicitly in the lease. This should
6714 ensure that in multi-homed systems DNS traffic leaves the systems on
6715 the interface that acquired the DNS server information even if other
6716 routes such as default routes exist. This behaviour may be turned on
6717 with the new RoutesToDNS= option.
6719 * systemd-networkd's VXLAN support gained a new option
6720 GenericProtocolExtension= for enabling VXLAN Generic Protocol
6721 Extension support, as well as IPDoNotFragment= for setting the IP
6722 "Don't fragment" bit on outgoing packets. A similar option has been
6723 added to the GENEVE support.
6725 * In systemd-networkd's [Route] section you may now configure
6726 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
6727 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
6728 propagation. The Type= setting now supports local, broadcast,
6729 anycast, multicast, any, xresolve routes, too.
6731 * systemd-networkd's [Network] section learnt a new option
6732 DefaultRouteOnDevice= for automatically configuring a default route
6733 onto the network device.
6735 * systemd-networkd's bridging support gained two new options ProxyARP=
6736 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
6737 MulticastRouter= for configuring multicast routing behaviour. A new
6738 option MulticastIGMPVersion= may be used to change bridge's multicast
6739 Internet Group Management Protocol (IGMP) version.
6741 * systemd-networkd's FooOverUDP support gained the ability to configure
6742 local and peer IP addresses via Local= and Peer=. A new option
6743 PeerPort= may be used to configure the peer's IP port.
6745 * systemd-networkd's TUN support gained a new setting VnetHeader= for
6746 tweaking Generic Segment Offload support.
6748 * The address family for policy rules may be specified using the new
6749 Family= option in the [RoutingPolicyRule] section.
6751 * networkctl gained a new "delete" command for removing virtual network
6752 devices, as well as a new "--stats" switch for showing device
6755 * networkd.conf gained a new setting SpeedMeter= and
6756 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
6757 measured speed may be shown by 'networkctl status'.
6759 * "networkctl status" now displays MTU and queue lengths, and more
6760 detailed information about VXLAN and bridge devices.
6762 * systemd-networkd's .network and .link files gained a new Property=
6763 setting in the [Match] section, to match against devices with
6764 specific udev properties.
6766 * systemd-networkd's tunnel support gained a new option
6767 AssignToLoopback= for selecting whether to use the loopback device
6768 "lo" as underlying device.
6770 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
6771 been renamed to LinkLayerAddress=, and it now allows configuration of
6774 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
6775 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
6776 IPv6 configuration (static or DHCPv6) was found for a given
6777 interface. It will not touch the sysctl otherwise.
6779 * The order of entries is $PATH used by the user manager instance was
6780 changed to put bin/ entries before the corresponding sbin/ entries.
6781 It is recommended to not rely on this order, and only ever have one
6782 binary with a given name in the system paths under /usr.
6784 * A new tool systemd-network-generator has been added that may generate
6785 .network, .netdev and .link files from IP configuration specified on
6786 the kernel command line in the format used by Dracut.
6788 * The CriticalConnection= setting in .network files is now deprecated,
6789 and replaced by a new KeepConfiguration= setting which allows more
6790 detailed configuration of the IP configuration to keep in place.
6792 * systemd-analyze gained a few new verbs:
6794 - "systemd-analyze timestamp" parses and converts timestamps. This is
6795 similar to the existing "systemd-analyze calendar" command which
6796 does the same for recurring calendar events.
6798 - "systemd-analyze timespan" parses and converts timespans (i.e.
6799 durations as opposed to points in time).
6801 - "systemd-analyze condition" will parse and test ConditionXYZ=
6804 - "systemd-analyze exit-status" will parse and convert exit status
6805 codes to their names and back.
6807 - "systemd-analyze unit-files" will print a list of all unit
6808 file paths and unit aliases.
6810 * SuccessExitStatus=, RestartPreventExitStatus=, and
6811 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
6812 is equivalent to "65"). Those exit status name mappings may be
6813 displayed with the systemd-analyze exit-status verb describe above.
6815 * systemd-logind now exposes a per-session SetBrightness() bus call,
6816 which may be used to securely change the brightness of a kernel
6817 brightness device, if it belongs to the session's seat. By using this
6818 call unprivileged clients can make changes to "backlight" and "leds"
6819 devices securely with strict requirements on session membership.
6820 Desktop environments may use this to generically make brightness
6821 changes to such devices without shipping private SUID binaries or
6822 udev rules for that purpose.
6824 * "udevadm info" gained a --wait-for-initialization switch to wait for
6825 a device to be initialized.
6827 * systemd-hibernate-resume-generator will now look for resumeflags= on
6828 the kernel command line, which is similar to rootflags= and may be
6829 used to configure device timeout for the hibernation device.
6831 * sd-event learnt a new API call sd_event_source_disable_unref() for
6832 disabling and unref'ing an event source in a single function. A
6833 related call sd_event_source_disable_unrefp() has been added for use
6834 with gcc's cleanup extension.
6836 * The sd-id128.h public API gained a new definition
6837 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
6840 * "busctl introspect" gained a new switch --xml-interface for dumping
6841 XML introspection data unmodified.
6843 * PID 1 may now show the unit name instead of the unit description
6844 string in its status output during boot. This may be configured in
6845 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
6846 kernel command line option systemd.status_unit_format=.
6848 * PID 1 now understands a new option KExecWatchdogSec= in
6849 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
6850 Previously watchdog functionality was only available for regular
6851 reboots. The new setting defaults to off, because we don't know in
6852 the general case if the watchdog will be reset after kexec (some
6853 drivers do reset it, but not all), and the new userspace might not be
6854 configured to handle the watchdog.
6856 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
6857 RebootWatchdogSec= to more clearly communicate what it is about. The
6858 old name is still accepted for compatibility.
6860 * The systemd.debug_shell kernel command line option now optionally
6861 takes a tty name to spawn the debug shell on, which allows a
6862 different tty to be selected than the built-in default.
6864 * Service units gained a new ExecCondition= setting which will run
6865 before ExecStartPre= and either continue execution of the unit (for
6866 clean exit codes), stop execution without marking the unit failed
6867 (for exit codes 1 through 254), or stop execution and fail the unit
6868 (for exit code 255 or abnormal termination).
6870 * A new service systemd-pstore.service has been added that pulls data
6871 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
6874 * timedatectl gained new verbs for configuring per-interface NTP
6875 service configuration for systemd-timesyncd.
6877 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
6878 2019. (You can set non-UTF-8 locales though, if you know their name.)
6880 * If variable assignments in sysctl.d/ files are prefixed with "-" any
6881 failures to apply them are now ignored.
6883 * systemd-random-seed.service now optionally credits entropy when
6884 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
6885 true for the service to enable this behaviour, but please consult the
6886 documentation first, since this comes with a couple of caveats.
6888 * systemd-random-seed.service is now a synchronization point for full
6889 initialization of the kernel's entropy pool. Services that require
6890 /dev/urandom to be correctly initialized should be ordered after this
6893 * The systemd-boot boot loader has been updated to optionally maintain
6894 a random seed file in the EFI System Partition (ESP). During the boot
6895 phase, this random seed is read and updated with a new seed
6896 cryptographically derived from it. Another derived seed is passed to
6897 the OS. The latter seed is then credited to the kernel's entropy pool
6898 very early during userspace initialization (from PID 1). This allows
6899 systems to boot up with a fully initialized kernel entropy pool from
6900 earliest boot on, and thus entirely removes all entropy pool
6901 initialization delays from systems using systemd-boot. Special care
6902 is taken to ensure different seeds are derived on system images
6903 replicated to multiple systems. "bootctl status" will show whether
6904 a seed was received from the boot loader.
6906 * bootctl gained two new verbs:
6908 - "bootctl random-seed" will generate the file in ESP and an EFI
6909 variable to allow a random seed to be passed to the OS as described
6912 - "bootctl is-installed" checks whether systemd-boot is currently
6915 * bootctl will warn if it detects that boot entries are misconfigured
6916 (for example if the kernel image was removed without purging the
6919 * A new document has been added describing systemd's use and support
6920 for the kernel's entropy pool subsystem:
6922 https://systemd.io/RANDOM_SEEDS
6924 * When the system is hibernated the swap device to write the
6925 hibernation image to is now automatically picked from all available
6926 swap devices, preferring the swap device with the highest configured
6927 priority over all others, and picking the device with the most free
6928 space if there are multiple devices with the highest priority.
6930 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
6931 option that permits selecting the timeout how long to wait for a
6932 device with an encryption key before asking for the password.
6934 * IOWeight= has learnt to properly set the IO weight when using the
6935 BFQ scheduler officially found in kernels 5.0+.
6937 * A new mailing list has been created for reporting of security issues:
6938 systemd-security@redhat.com. For mode details, see
6939 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
6941 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
6942 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
6943 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
6944 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
6945 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
6946 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
6947 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
6948 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
6949 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
6950 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
6951 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
6952 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
6953 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
6954 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
6955 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
6956 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
6957 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
6958 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
6959 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
6960 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
6961 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
6962 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
6963 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
6964 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
6965 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
6966 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
6967 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
6968 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
6969 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
6970 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
6972 – Camerino, 2019-09-03
6976 * In .link files, MACAddressPolicy=persistent (the default) is changed
6977 to cover more devices. For devices like bridges, tun, tap, bond, and
6978 similar interfaces that do not have other identifying information,
6979 the interface name is used as the basis for persistent seed for MAC
6980 and IPv4LL addresses. The way that devices that were handled
6981 previously is not changed, and this change is about covering more
6982 devices then previously by the "persistent" policy.
6984 MACAddressPolicy=random may be used to force randomized MACs and
6985 IPv4LL addresses for a device if desired.
6987 Hint: the log output from udev (at debug level) was enhanced to
6988 clarify what policy is followed and which attributes are used.
6989 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
6990 may be used to view this.
6992 Hint: if a bridge interface is created without any slaves, and gains
6993 a slave later, then now the bridge does not inherit slave's MAC.
6994 To inherit slave's MAC, for example, create the following file:
6996 # /etc/systemd/network/98-bridge-inherit-mac.link
7001 MACAddressPolicy=none
7004 * The .device units generated by systemd-fstab-generator and other
7005 generators do not automatically pull in the corresponding .mount unit
7006 as a Wants= dependency. This means that simply plugging in the device
7007 will not cause the mount unit to be started automatically. But please
7008 note that the mount unit may be started for other reasons, in
7009 particular if it is part of local-fs.target, and any unit which
7010 (transitively) depends on local-fs.target is started.
7012 * networkctl list/status/lldp now accept globbing wildcards for network
7013 interface names to match against all existing interfaces.
7015 * The $PIDFILE environment variable is set to point the absolute path
7016 configured with PIDFile= for processes of that service.
7018 * The fallback DNS server list was augmented with Cloudflare public DNS
7019 servers. Use `-Ddns-servers=` to set a different fallback.
7021 * A new special target usb-gadget.target will be started automatically
7022 when a USB Device Controller is detected (which means that the system
7023 is a USB peripheral).
7025 * A new unit setting CPUQuotaPeriodSec= assigns the time period
7026 relatively to which the CPU time quota specified by CPUQuota= is
7029 * A new unit setting ProtectHostname= may be used to prevent services
7030 from modifying hostname information (even if they otherwise would
7031 have privileges to do so).
7033 * A new unit setting NetworkNamespacePath= may be used to specify a
7034 namespace for service or socket units through a path referring to a
7035 Linux network namespace pseudo-file.
7037 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
7038 have an effect on .socket units: when used the listening socket is
7039 created within the configured network namespace instead of the host
7042 * ExecStart= command lines in unit files may now be prefixed with ':'
7043 in which case environment variable substitution is
7044 disabled. (Supported for the other ExecXYZ= settings, too.)
7046 * .timer units gained two new boolean settings OnClockChange= and
7047 OnTimezoneChange= which may be used to also trigger a unit when the
7048 system clock is changed or the local timezone is
7049 modified. systemd-run has been updated to make these options easily
7050 accessible from the command line for transient timers.
7052 * Two new conditions for units have been added: ConditionMemory= may be
7053 used to conditionalize a unit based on installed system
7054 RAM. ConditionCPUs= may be used to conditionalize a unit based on
7055 installed CPU cores.
7057 * The @default system call filter group understood by SystemCallFilter=
7058 has been updated to include the new rseq() system call introduced in
7061 * A new time-set.target has been added that indicates that the system
7062 time has been set from a local source (possibly imprecise). The
7063 existing time-sync.target is stronger and indicates that the time has
7064 been synchronized with a precise external source. Services where
7065 approximate time is sufficient should use the new target.
7067 * "systemctl start" (and related commands) learnt a new
7068 --show-transaction option. If specified brief information about all
7069 jobs queued because of the requested operation is shown.
7071 * systemd-networkd recognizes a new operation state 'enslaved', used
7072 (instead of 'degraded' or 'carrier') for interfaces which form a
7073 bridge, bond, or similar, and an new 'degraded-carrier' operational
7074 state used for the bond or bridge master interface when one of the
7075 enslaved devices is not operational.
7077 * .network files learnt the new IgnoreCarrierLoss= option for leaving
7078 networks configured even if the carrier is lost.
7080 * The RequiredForOnline= setting in .network files may now specify a
7081 minimum operational state required for the interface to be considered
7082 "online" by systemd-networkd-wait-online. Related to this
7083 systemd-networkd-wait-online gained a new option --operational-state=
7084 to configure the same, and its --interface= option was updated to
7085 optionally also take an operational state specific for an interface.
7087 * systemd-networkd-wait-online gained a new setting --any for waiting
7088 for only one of the requested interfaces instead of all of them.
7090 * systemd-networkd now implements L2TP tunnels.
7092 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
7093 may be used to cause autonomous and onlink prefixes received in IPv6
7094 Router Advertisements to be ignored.
7096 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
7097 file settings may be used to tweak bridge behaviour.
7099 * The new TripleSampling= option in .network files may be used to
7100 configure CAN triple sampling.
7102 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
7103 used to point to private or preshared key for a WireGuard interface.
7105 * /etc/crypttab now supports the same-cpu-crypt and
7106 submit-from-crypt-cpus options to tweak encryption work scheduling
7109 * systemd-tmpfiles will now take a BSD file lock before operating on a
7110 contents of directory. This may be used to temporarily exclude
7111 directories from aging by taking the same lock (useful for example
7112 when extracting a tarball into /tmp or /var/tmp as a privileged user,
7113 which might create files with really old timestamps, which
7114 nevertheless should not be deleted). For further details, see:
7116 https://systemd.io/TEMPORARY_DIRECTORIES
7118 * systemd-tmpfiles' h line type gained support for the
7119 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
7120 controlling project quota inheritance.
7122 * sd-boot and bootctl now implement support for an Extended Boot Loader
7123 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
7124 addition to the ESP partition mounted to /efi or /boot/efi.
7125 Configuration file fragments, kernels, initrds and other EFI images
7126 to boot will be loaded from both the ESP and XBOOTLDR partitions.
7127 The XBOOTLDR partition was previously described by the Boot Loader
7128 Specification, but implementation was missing in sd-boot. Support for
7129 this concept allows using the sd-boot boot loader in more
7130 conservative scenarios where the boot loader itself is placed in the
7131 ESP but the kernels to boot (and their metadata) in a separate
7134 * A system may now be booted with systemd.volatile=overlay on the
7135 kernel command line, which causes the root file system to be set up
7136 an overlayfs mount combining the root-only root directory with a
7137 writable tmpfs. In this setup, the underlying root device is not
7138 modified, and any changes are lost at reboot.
7140 * Similar, systemd-nspawn can now boot containers with a volatile
7141 overlayfs root with the new --volatile=overlay switch.
7143 * systemd-nspawn can now consume OCI runtime bundles using a new
7144 --oci-bundle= option. This implementation is fully usable, with most
7145 features in the specification implemented, but since this a lot of
7146 new code and functionality, this feature should most likely not
7147 be used in production yet.
7149 * systemd-nspawn now supports various options described by the OCI
7150 runtime specification on the command-line and in .nspawn files:
7151 --inaccessible=/Inaccessible= may be used to mask parts of the file
7152 system tree, --console=/--pipe may be used to configure how standard
7153 input, output, and error are set up.
7155 * busctl learned the `emit` verb to generate D-Bus signals.
7157 * systemd-analyze cat-config may be used to gather and display
7158 configuration spread over multiple files, for example system and user
7159 presets, tmpfiles.d, sysusers.d, udev rules, etc.
7161 * systemd-analyze calendar now takes an optional new parameter
7162 --iterations= which may be used to show a maximum number of iterations
7163 the specified expression will elapse next.
7165 * The sd-bus C API gained support for naming method parameters in the
7168 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
7169 the reboot() system call expects.
7171 * journalctl learnt a new --cursor-file= option that points to a file
7172 from which a cursor should be loaded in the beginning and to which
7173 the updated cursor should be stored at the end.
7175 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
7176 detected by systemd-detect-virt (and may also be used in
7177 ConditionVirtualization=).
7179 * The behaviour of systemd-logind may now be modified with environment
7180 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
7181 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
7182 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
7183 skip the relevant operation completely (when set to false), or to
7184 create a flag file in /run/systemd (when set to true), instead of
7185 actually commencing the real operation when requested. The presence
7186 of /run/systemd/reboot-to-firmware-setup,
7187 /run/systemd/reboot-to-boot-loader-menu, and
7188 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
7189 boot loader implementations to replace some steps logind performs
7190 during reboot with their own operations.
7192 * systemctl can be used to request a reboot into the boot loader menu
7193 or a specific boot loader entry with the new --boot-load-menu= and
7194 --boot-loader-entry= options to a reboot command. (This requires a
7195 boot loader that supports this, for example sd-boot.)
7197 * kernel-install will no longer unconditionally create the output
7198 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
7199 snippets, but will do only if the machine-specific parent directory
7200 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
7201 to create this parent directory during sd-boot installation.
7203 This makes it easier to use kernel-install with plugins which support
7204 a different layout of the bootloader partitions (for example grub2).
7206 * During package installation (with `ninja install`), we would create
7207 symlinks for getty@tty1.service, systemd-networkd.service,
7208 systemd-networkd.socket, systemd-resolved.service,
7209 remote-cryptsetup.target, remote-fs.target,
7210 systemd-networkd-wait-online.service, and systemd-timesyncd.service
7211 in /etc, as if `systemctl enable` was called for those units, to make
7212 the system usable immediately after installation. Now this is not
7213 done anymore, and instead calling `systemctl preset-all` is
7214 recommended after the first installation of systemd.
7216 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
7217 is built on seccomp. When turned on creation of SUID/SGID files is
7220 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
7221 implied if DynamicUser= is turned on for a service. This hardens
7222 these services, so that they neither can benefit from nor create
7223 SUID/SGID executables. This is a minor compatibility breakage, given
7224 that when DynamicUser= was first introduced SUID/SGID behaviour was
7225 unaffected. However, the security benefit of these two options is
7226 substantial, and the setting is still relatively new, hence we opted
7227 to make it mandatory for services with dynamic users.
7229 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
7230 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
7231 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
7232 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
7233 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
7234 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
7235 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
7236 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
7237 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
7238 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
7239 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
7240 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
7241 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
7242 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
7243 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
7244 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
7245 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
7246 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
7248 — Warsaw, 2019-04-11
7252 * The default locale can now be configured at compile time. Otherwise,
7253 a suitable default will be selected automatically (one of C.UTF-8,
7254 en_US.UTF-8, and C).
7256 * The version string shown by systemd and other tools now includes the
7257 git commit hash when built from git. An override may be specified
7258 during compilation, which is intended to be used by distributions to
7259 include the package release information.
7261 * systemd-cat can now filter standard input and standard error streams
7262 for different syslog priorities using the new --stderr-priority=
7265 * systemd-journald and systemd-journal-remote reject entries which
7266 contain too many fields (CVE-2018-16865) and set limits on the
7267 process' command line length (CVE-2018-16864).
7269 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
7272 * A new network device NamePolicy "keep" is implemented for link files,
7273 and used by default in 99-default.link (the fallback configuration
7274 provided by systemd). With this policy, if the network device name
7275 was already set by userspace, the device will not be renamed again.
7276 This matches the naming scheme that was implemented before
7277 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
7278 is also enabled by default, even if not specified. Effectively, this
7279 means that if naming-scheme >= 240 is specified, network devices will
7280 be renamed according to the configuration, even if they have been
7281 renamed already, if "keep" is not specified as the naming policy in
7282 the .link file. The 99-default.link file provided by systemd includes
7283 "keep" for backwards compatibility, but it is recommended for user
7284 installed .link files to *not* include it.
7286 The "kernel" policy, which keeps kernel names declared to be
7287 "persistent", now works again as documented.
7289 * kernel-install script now optionally takes the paths to one or more
7290 initrd files, and passes them to all plugins.
7292 * The mincore() system call has been dropped from the @system-service
7293 system call filter group, as it is pretty exotic and may potentially
7294 used for side-channel attacks.
7296 * -fPIE is dropped from compiler and linker options. Please specify
7297 -Db_pie=true option to meson to build position-independent
7298 executables. Note that the meson option is supported since meson-0.49.
7300 * The fs.protected_regular and fs.protected_fifos sysctls, which were
7301 added in Linux 4.19 to make some data spoofing attacks harder, are
7302 now enabled by default. While this will hopefully improve the
7303 security of most installations, it is technically a backwards
7304 incompatible change; to disable these sysctls again, place the
7305 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
7307 fs.protected_regular = 0
7308 fs.protected_fifos = 0
7310 Note that the similar hardlink and symlink protection has been
7311 enabled since v199, and may be disabled likewise.
7313 * The files read from the EnvironmentFile= setting in unit files now
7314 parse backslashes inside quotes literally, matching the behaviour of
7317 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
7318 now automatically become NOPs when run in a chroot() environment.
7320 * The tmpfiles.d/ "C" line type will now copy directory trees not only
7321 when the destination is so far missing, but also if it already exists
7322 as a directory and is empty. This is useful to cater for systems
7323 where directory trees are put together from multiple separate mount
7324 points but otherwise empty.
7326 * A new function sd_bus_close_unref() (and the associated
7327 sd_bus_close_unrefp()) has been added to libsystemd, that combines
7328 sd_bus_close() and sd_bus_unref() in one.
7330 * udevadm control learnt a new option for --ping for testing whether a
7331 systemd-udevd instance is running and reacting.
7333 * udevadm trigger learnt a new option for --wait-daemon for waiting
7334 systemd-udevd daemon to be initialized.
7336 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
7337 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
7338 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
7339 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
7340 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
7341 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
7342 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
7343 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
7344 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
7345 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
7346 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
7347 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
7348 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
7349 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
7350 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
7351 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
7352 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
7354 — Berlin, 2019-02-14
7358 * NoNewPrivileges=yes has been set for all long-running services
7359 implemented by systemd. Previously, this was problematic due to
7360 SELinux (as this would also prohibit the transition from PID1's label
7361 to the service's label). This restriction has since been lifted, but
7362 an SELinux policy update is required.
7363 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
7365 * DynamicUser=yes is dropped from systemd-networkd.service,
7366 systemd-resolved.service and systemd-timesyncd.service, which was
7367 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
7368 and since v236 for systemd-timesyncd.service. The users and groups
7369 systemd-network, systemd-resolve and systemd-timesync are created
7370 by systemd-sysusers again. Distributors or system administrators
7371 may need to create these users and groups if they not exist (or need
7372 to re-enable DynamicUser= for those units) while upgrading systemd.
7373 Also, the clock file for systemd-timesyncd may need to move from
7374 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
7376 * When unit files are loaded from disk, previously systemd would
7377 sometimes (depending on the unit loading order) load units from the
7378 target path of symlinks in .wants/ or .requires/ directories of other
7379 units. This meant that unit could be loaded from different paths
7380 depending on whether the unit was requested explicitly or as a
7381 dependency of another unit, not honouring the priority of directories
7382 in search path. It also meant that it was possible to successfully
7383 load and start units which are not found in the unit search path, as
7384 long as they were requested as a dependency and linked to from
7385 .wants/ or .requires/. The target paths of those symlinks are not
7386 used for loading units anymore and the unit file must be found in
7389 * A new service type has been added: Type=exec. It's very similar to
7390 Type=simple but ensures the service manager will wait for both fork()
7391 and execve() of the main service binary to complete before proceeding
7392 with follow-up units. This is primarily useful so that the manager
7393 propagates any errors in the preparation phase of service execution
7394 back to the job that requested the unit to be started. For example,
7395 consider a service that has ExecStart= set to a file system binary
7396 that doesn't exist. With Type=simple starting the unit would be
7397 considered instantly successful, as only fork() has to complete
7398 successfully and the manager does not wait for execve(), and hence
7399 its failure is seen "too late". With the new Type=exec service type
7400 starting the unit will fail, as the manager will wait for the
7401 execve() and notice its failure, which is then propagated back to the
7404 NOTE: with the next release 241 of systemd we intend to change the
7405 systemd-run tool to default to Type=exec for transient services
7406 started by it. This should be mostly safe, but in specific corner
7407 cases might result in problems, as the systemd-run tool will then
7408 block on NSS calls (such as user name look-ups due to User=) done
7409 between the fork() and execve(), which under specific circumstances
7410 might cause problems. It is recommended to specify "-p Type=simple"
7411 explicitly in the few cases where this applies. For regular,
7412 non-transient services (i.e. those defined with unit files on disk)
7413 we will continue to default to Type=simple.
7415 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
7416 userspace processes is set to 1024 (soft) and 4096
7417 (hard). Previously, systemd passed this on unmodified to all
7418 processes it forked off. With this systemd release the hard limit
7419 systemd passes on is increased to 512K, overriding the kernel's
7420 defaults and substantially increasing the number of simultaneous file
7421 descriptors unprivileged userspace processes can allocate. Note that
7422 the soft limit remains at 1024 for compatibility reasons: the
7423 traditional UNIX select() call cannot deal with file descriptors >=
7424 1024 and increasing the soft limit globally might thus result in
7425 programs unexpectedly allocating a high file descriptor and thus
7426 failing abnormally when attempting to use it with select() (of
7427 course, programs shouldn't use select() anymore, and prefer
7428 poll()/epoll, but the call unfortunately remains undeservedly popular
7429 at this time). This change reflects the fact that file descriptor
7430 handling in the Linux kernel has been optimized in more recent
7431 kernels and allocating large numbers of them should be much cheaper
7432 both in memory and in performance than it used to be. Programs that
7433 want to take benefit of the increased limit have to "opt-in" into
7434 high file descriptors explicitly by raising their soft limit. Of
7435 course, when they do that they must acknowledge that they cannot use
7436 select() anymore (and neither can any shared library they use — or
7437 any shared library used by any shared library they use and so on).
7438 Which default hard limit is most appropriate is of course hard to
7439 decide. However, given reports that ~300K file descriptors are used
7440 in real-life applications we believe 512K is sufficiently high as new
7441 default for now. Note that there are also reports that using very
7442 high hard limits (e.g. 1G) is problematic: some software allocates
7443 large arrays with one element for each potential file descriptor
7444 (Java, …) — a high hard limit thus triggers excessively large memory
7445 allocations in these applications. Hopefully, the new default of 512K
7446 is a good middle ground: higher than what real-life applications
7447 currently need, and low enough for avoid triggering excessively large
7448 allocations in problematic software. (And yes, somebody should fix
7451 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
7452 to the highest possible values, as separate accounting of file
7453 descriptors is no longer necessary, as memcg tracks them correctly as
7454 part of the memory accounting anyway. Thus, from the four limits on
7455 file descriptors currently enforced (fs.file-max, fs.nr_open,
7456 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
7457 and keep only the latter two. A set of build-time options
7458 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
7459 has been added to revert this change in behaviour, which might be
7460 an option for systems that turn off memcg in the kernel.
7462 * When no /etc/locale.conf file exists (and hence no locale settings
7463 are in place), systemd will now use the "C.UTF-8" locale by default,
7464 and set LANG= to it. This locale is supported by various
7465 distributions including Fedora, with clear indications that upstream
7466 glibc is going to make it available too. This locale enables UTF-8
7467 mode by default, which appears appropriate for 2018.
7469 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
7470 default. This effectively switches the RFC3704 Reverse Path filtering
7471 from Strict mode to Loose mode. This is more appropriate for hosts
7472 that have multiple links with routes to the same networks (e.g.
7473 a client with a Wi-Fi and Ethernet both connected to the internet).
7475 Consult the kernel documentation for details on this sysctl:
7476 https://docs.kernel.org/networking/ip-sysctl.html
7478 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
7481 * CPUAccounting=yes no longer enables the CPU controller when using
7482 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
7483 statistics are now provided independently from the CPU controller.
7485 * Support for disabling a particular cgroup controller within a sub-tree
7486 has been added through the DisableControllers= directive.
7488 * cgroup_no_v1=all on the kernel command line now also implies
7489 using the unified cgroup hierarchy, unless one explicitly passes
7490 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
7492 * The new "MemoryMin=" unit file property may now be used to set the
7493 memory usage protection limit of processes invoked by the unit. This
7494 controls the cgroup v2 memory.min attribute. Similarly, the new
7495 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
7496 cgroup v2 io.latency cgroup property for configuring per-service I/O
7499 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
7500 to the cgroup v1 "devices" cgroup controller.
7502 * systemd-escape now is able to combine --unescape with --template. It
7503 also learnt a new option --instance for extracting and unescaping the
7504 instance part of a unit name.
7506 * sd-bus now provides the sd_bus_message_readv() which is similar to
7507 sd_bus_message_read() but takes a va_list object. The pair
7508 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
7509 has been added for configuring the default method call timeout to
7510 use. sd_bus_error_move() may be used to efficiently move the contents
7511 from one sd_bus_error structure to another, invalidating the
7512 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
7513 be used to control whether a bus connection object is automatically
7514 flushed when an sd-event loop is exited.
7516 * When processing classic BSD syslog log messages, journald will now
7517 save the original time-stamp string supplied in the new
7518 SYSLOG_TIMESTAMP= journal field. This permits consumers to
7519 reconstruct the original BSD syslog message more correctly.
7521 * StandardOutput=/StandardError= in service files gained support for
7522 new "append:…" parameters, for connecting STDOUT/STDERR of a service
7523 to a file, and appending to it.
7525 * The signal to use as last step of killing of unit processes is now
7526 configurable. Previously it was hard-coded to SIGKILL, which may now
7527 be overridden with the new KillSignal= setting. Note that this is the
7528 signal used when regular termination (i.e. SIGTERM) does not suffice.
7529 Similarly, the signal used when aborting a program in case of a
7530 watchdog timeout may now be configured too (WatchdogSignal=).
7532 * The XDG_SESSION_DESKTOP environment variable may now be configured in
7533 the pam_systemd argument line, using the new desktop= switch. This is
7534 useful to initialize it properly from a display manager without
7535 having to touch C code.
7537 * Most configuration options that previously accepted percentage values
7538 now also accept permille values with the '‰' suffix (instead of '%').
7540 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
7543 * systemd-resolved's configuration file resolved.conf gained a new
7544 option ReadEtcHosts= which may be used to turn off processing and
7545 honoring /etc/hosts entries.
7547 * The "--wait" switch may now be passed to "systemctl
7548 is-system-running", in which case the tool will synchronously wait
7549 until the system finished start-up.
7551 * hostnamed gained a new bus call to determine the DMI product UUID.
7553 * On x86-64 systemd will now prefer using the RDRAND processor
7554 instruction over /dev/urandom whenever it requires randomness that
7555 neither has to be crypto-grade nor should be reproducible. This
7556 should substantially reduce the amount of entropy systemd requests
7557 from the kernel during initialization on such systems, though not
7558 reduce it to zero. (Why not zero? systemd still needs to allocate
7559 UUIDs and such uniquely, which require high-quality randomness.)
7561 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
7562 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
7563 for forcing the "Other Information" bit in IPv6 RA messages. The
7564 bonding logic gained four new options AdActorSystemPriority=,
7565 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
7566 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
7567 shuffling of flows. The tunnel logic gained a new
7568 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
7569 Deployment. The policy rule logic gained four new options IPProtocol=,
7570 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
7571 support for the MulticastToUnicast= option. networkd also gained
7572 support for configuring static IPv4 ARP or IPv6 neighbor entries.
7574 * .preset files (as read by 'systemctl preset') may now be used to
7575 instantiate services.
7577 * /etc/crypttab now understands the sector-size= option to configure
7578 the sector size for an encrypted partition.
7580 * Key material for encrypted disks may now be placed on a formatted
7581 medium, and referenced from /etc/crypttab by the UUID of the file
7582 system, followed by "=" suffixed by the path to the key file.
7584 * The "collect" udev component has been removed without replacement, as
7585 it is neither used nor maintained.
7587 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
7588 LogsDirectory=, ConfigurationDirectory= settings are used in a
7589 service the executed processes will now receive a set of environment
7590 variables containing the full paths of these directories.
7591 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
7592 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
7593 are used. Note that these options may be used multiple times per
7594 service in which case the resulting paths will be concatenated and
7595 separated by colons.
7597 * Predictable interface naming has been extended to cover InfiniBand
7598 NICs. They will be exposed with an "ib" prefix.
7600 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
7601 which case the respective line failing is ignored.
7603 * .link files may now be used to configure the equivalent to the
7604 "ethtool advertise" commands.
7606 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
7607 alternative to libudev.h. Previously, the latter was just an internal
7608 wrapper around the former, but now these two APIs are exposed
7611 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
7612 which calculates an app-specific boot ID similar to how
7613 sd_id128_get_machine_app_specific() generates an app-specific machine
7616 * A new tool systemd-id128 has been added that can be used to determine
7617 and generate various 128-bit IDs.
7619 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
7622 * systemd-hibernate-resume-generator will now honor the "noresume"
7623 kernel command line option, in which case it will bypass resuming
7624 from any hibernated image.
7626 * The systemd-sleep.conf configuration file gained new options
7627 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
7628 AllowHybridSleep= for prohibiting specific sleep modes even if the
7629 kernel exports them.
7631 * portablectl is now officially supported and has thus moved to
7634 * bootctl learnt the two new commands "set-default" and "set-oneshot"
7635 for setting the default boot loader item to boot to (either
7636 persistently or only for the next boot). This is currently only
7637 compatible with sd-boot, but may be implemented on other boot loaders
7638 too, that follow the boot loader interface. The updated interface is
7639 now documented here:
7641 https://systemd.io/BOOT_LOADER_INTERFACE
7643 * A new kernel command line option systemd.early_core_pattern= is now
7644 understood which may be used to influence the core_pattern PID 1
7645 installs during early boot.
7647 * busctl learnt two new options -j and --json= for outputting method
7648 call replies, properties and monitoring output in JSON.
7650 * journalctl's JSON output now supports simple ANSI coloring as well as
7651 a new "json-seq" mode for generating RFC7464 output.
7653 * Unit files now support the %g/%G specifiers that resolve to the UNIX
7654 group/GID of the service manager runs as, similar to the existing
7655 %u/%U specifiers that resolve to the UNIX user/UID.
7657 * systemd-logind learnt a new global configuration option
7658 UserStopDelaySec= that may be set in logind.conf. It specifies how
7659 long the systemd --user instance shall remain started after a user
7660 logs out. This is useful to speed up repetitive re-connections of the
7661 same user, as it means the user's service manager doesn't have to be
7662 stopped/restarted on each iteration, but can be reused between
7663 subsequent options. This setting defaults to 10s. systemd-logind also
7664 exports two new properties on its Manager D-Bus objects indicating
7665 whether the system's lid is currently closed, and whether the system
7668 * systemd gained support for a generic boot counting logic, which
7669 generically permits automatic reverting to older boot loader entries
7670 if newer updated ones don't work. The boot loader side is implemented
7671 in sd-boot, but is kept open for other boot loaders too. For details
7674 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
7676 * The SuccessAction=/FailureAction= unit file settings now learnt two
7677 new parameters: "exit" and "exit-force", which result in immediate
7678 exiting of the service manager, and are only useful in systemd --user
7679 and container environments.
7681 * Unit files gained support for a pair of options
7682 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
7683 exit status to use as service manager exit status when
7684 SuccessAction=/FailureAction= is set to exit or exit-force.
7686 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
7687 options may now be used to configure the log rate limiting applied by
7688 journald per-service.
7690 * systemd-analyze gained a new verb "timespan" for parsing and
7691 normalizing time span values (i.e. strings like "5min 7s 8us").
7693 * systemd-analyze also gained a new verb "security" for analyzing the
7694 security and sand-boxing settings of services in order to determine an
7695 "exposure level" for them, indicating whether a service would benefit
7696 from more sand-boxing options turned on for them.
7698 * "systemd-analyze syscall-filter" will now also show system calls
7699 supported by the local kernel but not included in any of the defined
7702 * .nspawn files now understand the Ephemeral= setting, matching the
7703 --ephemeral command line switch.
7705 * sd-event gained the new APIs sd_event_source_get_floating() and
7706 sd_event_source_set_floating() for controlling whether a specific
7707 event source is "floating", i.e. destroyed along with the even loop
7710 * Unit objects on D-Bus gained a new "Refs" property that lists all
7711 clients that currently have a reference on the unit (to ensure it is
7714 * The JoinControllers= option in system.conf is no longer supported, as
7715 it didn't work correctly, is hard to support properly, is legacy (as
7716 the concept only exists on cgroup v1) and apparently wasn't used.
7718 * Journal messages that are generated whenever a unit enters the failed
7719 state are now tagged with a unique MESSAGE_ID. Similarly, messages
7720 generated whenever a service process exits are now made recognizable,
7721 too. A tagged message is also emitted whenever a unit enters the
7722 "dead" state on success.
7724 * systemd-run gained a new switch --working-directory= for configuring
7725 the working directory of the service to start. A shortcut -d is
7726 equivalent, setting the working directory of the service to the
7727 current working directory of the invoking program. The new --shell
7728 (or just -S) option has been added for invoking the $SHELL of the
7729 caller as a service, and implies --pty --same-dir --wait --collect
7730 --service-type=exec. Or in other words, "systemd-run -S" is now the
7731 quickest way to quickly get an interactive in a fully clean and
7732 well-defined system service context.
7734 * machinectl gained a new verb "import-fs" for importing an OS tree
7735 from a directory. Moreover, when a directory or tarball is imported
7736 and single top-level directory found with the OS itself below the OS
7737 tree is automatically mangled and moved one level up.
7739 * systemd-importd will no longer set up an implicit btrfs loop-back
7740 file system on /var/lib/machines. If one is already set up, it will
7741 continue to be used.
7743 * A new generator "systemd-run-generator" has been added. It will
7744 synthesize a unit from one or more program command lines included in
7745 the kernel command line. This is very useful in container managers
7748 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
7750 This will run "systemd-nspawn" on an image, invoke the specified
7751 command line and immediately shut down the container again, returning
7752 the command line's exit code.
7754 * The block device locking logic is now documented:
7756 https://systemd.io/BLOCK_DEVICE_LOCKING
7758 * loginctl and machinectl now optionally output the various tables in
7759 JSON using the --output= switch. It is our intention to add similar
7760 support to systemctl and all other commands.
7762 * udevadm's query and trigger verb now optionally take a .device unit
7765 * systemd-udevd's network naming logic now understands a new
7766 net.naming-scheme= kernel command line switch, which may be used to
7767 pick a specific version of the naming scheme. This helps stabilizing
7768 interface names even as systemd/udev are updated and the naming logic
7771 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
7772 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
7773 initialize one to all 0xFF.
7775 * After loading the SELinux policy systemd will now recursively relabel
7776 all files and directories listed in
7777 /run/systemd/relabel-extra.d/*.relabel (which should be simple
7778 newline separated lists of paths) in addition to the ones it already
7779 implicitly relabels in /run, /dev and /sys. After the relabelling is
7780 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
7781 removed. This is useful to permit initrds (i.e. code running before
7782 the SELinux policy is in effect) to generate files in the host
7783 filesystem safely and ensure that the correct label is applied during
7784 the transition to the host OS.
7786 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
7787 mknod() handling in user namespaces. Previously mknod() would always
7788 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
7789 but device nodes generated that way cannot be opened, and attempts to
7790 open them result in EPERM. This breaks the "graceful fallback" logic
7791 in systemd's PrivateDevices= sand-boxing option. This option is
7792 implemented defensively, so that when systemd detects it runs in a
7793 restricted environment (such as a user namespace, or an environment
7794 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
7795 where device nodes cannot be created the effect of PrivateDevices= is
7796 bypassed (following the logic that 2nd-level sand-boxing is not
7797 essential if the system systemd runs in is itself already sand-boxed
7798 as a whole). This logic breaks with 4.18 in container managers where
7799 user namespacing is used: suddenly PrivateDevices= succeeds setting
7800 up a private /dev/ file system containing devices nodes — but when
7801 these are opened they don't work.
7803 At this point it is recommended that container managers utilizing
7804 user namespaces that intend to run systemd in the payload explicitly
7805 block mknod() with seccomp or similar, so that the graceful fallback
7808 We are very sorry for the breakage and the requirement to change
7809 container configurations for newer kernels. It's purely caused by an
7810 incompatible kernel change. The relevant kernel developers have been
7811 notified about this userspace breakage quickly, but they chose to
7814 * PermissionsStartOnly= setting is deprecated (but is still supported
7815 for backwards compatibility). The same functionality is provided by
7816 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
7819 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
7820 pam_systemd anymore.
7822 * The naming scheme for network devices was changed to always rename
7823 devices, even if they were already renamed by userspace. The "kernel"
7824 policy was changed to only apply as a fallback, if no other naming
7827 * The requirements to build systemd is bumped to meson-0.46 and
7830 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
7831 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
7832 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
7833 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
7834 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
7835 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
7836 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
7837 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
7838 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
7839 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
7840 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
7841 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
7842 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
7843 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
7844 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
7845 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
7846 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
7847 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
7848 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
7849 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
7850 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
7851 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
7852 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
7853 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
7854 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
7855 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
7856 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
7857 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
7858 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
7859 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
7860 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
7861 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
7862 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
7863 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
7864 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
7865 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
7866 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
7867 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
7868 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
7869 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
7870 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
7871 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
7872 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
7873 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
7874 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
7876 — Warsaw, 2018-12-21
7880 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
7881 builtin will name network interfaces differently than in previous
7882 versions for virtual network interfaces created with SR-IOV and NPAR
7883 and for devices where the PCI network controller device does not have
7884 a slot number associated.
7886 SR-IOV virtual devices are now named based on the name of the parent
7887 interface, with a suffix of "v<N>", where <N> is the virtual device
7888 number. Previously those virtual devices were named as if completely
7891 The ninth and later NPAR virtual devices will be named following the
7892 scheme used for the first eight NPAR partitions. Previously those
7893 devices were not renamed and the kernel default (eth<n>) was used.
7895 "net_id" will also generate names for PCI devices where the PCI
7896 network controller device does not have an associated slot number
7897 itself, but one of its parents does. Previously those devices were
7898 not renamed and the kernel default (eth<n>) was used.
7900 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
7901 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
7902 the unit. So, it is expected that the default behavior of
7903 systemd-logind is not changed. However, if distribution packagers or
7904 administrators disabled or modified IPAddressDeny= setting by a
7905 drop-in config file, then it may be necessary to update the file to
7906 re-enable AF_INET and AF_INET6 to support network user name services,
7909 * When the RestrictNamespaces= unit property is specified multiple
7910 times, then the specified types are merged now. Previously, only the
7911 last assignment was used. So, if distribution packagers or
7912 administrators modified the setting by a drop-in config file, then it
7913 may be necessary to update the file.
7915 * When OnFailure= is used in combination with Restart= on a service
7916 unit, then the specified units will no longer be triggered on
7917 failures that result in restarting. Previously, the specified units
7918 would be activated each time the unit failed, even when the unit was
7919 going to be restarted automatically. This behaviour contradicted the
7920 documentation. With this release the code is adjusted to match the
7923 * systemd-tmpfiles will now print a notice whenever it encounters
7924 tmpfiles.d/ lines referencing the /var/run/ directory. It will
7925 recommend reworking them to use the /run/ directory instead (for
7926 which /var/run/ is simply a symlinked compatibility alias). This way
7927 systemd-tmpfiles can properly detect line conflicts and merge lines
7928 referencing the same file by two paths, without having to access
7931 * systemctl disable/unmask/preset/preset-all cannot be used with
7932 --runtime. Previously this was allowed, but resulted in unintuitive
7933 behaviour that wasn't useful. systemctl disable/unmask will now undo
7934 both runtime and persistent enablement/masking, i.e. it will remove
7935 any relevant symlinks both in /run and /etc.
7937 * Note that all long-running system services shipped with systemd will
7938 now default to a system call allow list (rather than a deny list, as
7939 before). In particular, systemd-udevd will now enforce one too. For
7940 most cases this should be safe, however downstream distributions
7941 which disabled sandboxing of systemd-udevd (specifically the
7942 MountFlags= setting), might want to disable this security feature
7943 too, as the default allow-listing will prohibit all mount, swap,
7944 reboot and clock changing operations from udev rules.
7946 * sd-boot acquired new loader configuration settings to optionally turn
7947 off Windows and MacOS boot partition discovery as well as
7948 reboot-into-firmware menu items. It is also able to pick a better
7949 screen resolution for HiDPI systems, and now provides loader
7950 configuration settings to change the resolution explicitly.
7952 * systemd-resolved now supports DNS-over-TLS. It's still
7953 turned off by default, use DNSOverTLS=opportunistic to turn it on in
7954 resolved.conf. We intend to make this the default as soon as couple
7955 of additional techniques for optimizing the initial latency caused by
7956 establishing a TLS/TCP connection are implemented.
7958 * systemd-resolved.service and systemd-networkd.service now set
7959 DynamicUser=yes. The users systemd-resolve and systemd-network are
7960 not created by systemd-sysusers anymore.
7962 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
7963 that embed a network facing module into any process using getpwuid()
7964 or related call: the dynamic allocation of the user ID for
7965 systemd-resolved.service means the service manager has to check NSS
7966 if the user name is already taken when forking off the service. Since
7967 the user in the common case won't be defined in /etc/passwd the
7968 lookup is likely to trigger nss-ldap which in turn might use NSS to
7969 ask systemd-resolved for hostname lookups. This will hence result in
7970 a deadlock: a user name lookup in order to start
7971 systemd-resolved.service will result in a hostname lookup for which
7972 systemd-resolved.service needs to be started already. There are
7973 multiple ways to work around this problem: pre-allocate the
7974 "systemd-resolve" user on such systems, so that nss-ldap won't be
7975 triggered; or use a different NSS package that doesn't do networking
7976 in-process but provides a local asynchronous name cache; or configure
7977 the NSS package to avoid lookups for UIDs in the range `pkg-config
7978 systemd --variable=dynamicuidmin` … `pkg-config systemd
7979 --variable=dynamicuidmax`, so that it does not consider itself
7980 authoritative for the same UID range systemd allocates dynamic users
7983 * The systemd-resolve tool has been renamed to resolvectl (it also
7984 remains available under the old name, for compatibility), and its
7985 interface is now verb-based, similar in style to the other <xyz>ctl
7986 tools, such as systemctl or loginctl.
7988 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
7989 compatibility. It may be symlinked under the 'resolvconf' name, in
7990 which case it will take arguments and input compatible with the
7991 Debian and FreeBSD resolvconf tool.
7993 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
7994 where the system initially suspends, and after a timeout resumes and
7997 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
7998 set the client will only send a DUID as client identifier. (EDIT: the
7999 option was broken, and was dropped in v255.)
8001 * The nss-systemd glibc NSS module will now enumerate dynamic users and
8002 groups in effect. Previously, it could resolve UIDs/GIDs to user
8003 names/groups and vice versa, but did not support enumeration.
8005 * journald's Compress= configuration setting now optionally accepts a
8006 byte threshold value. All journal objects larger than this threshold
8007 will be compressed, smaller ones will not. Previously this threshold
8008 was not configurable and set to 512.
8010 * A new system.conf setting NoNewPrivileges= is now available which may
8011 be used to turn off acquisition of new privileges system-wide
8012 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
8013 for all its children). Note that turning this option on means setuid
8014 binaries and file system capabilities lose their special powers.
8015 While turning on this option is a big step towards a more secure
8016 system, doing so is likely to break numerous pre-existing UNIX tools,
8017 in particular su and sudo.
8019 * A new service systemd-time-sync-wait.service has been added. If
8020 enabled it will delay the time-sync.target unit at boot until time
8021 synchronization has been received from the network. This
8022 functionality is useful on systems lacking a local RTC or where it is
8023 acceptable that the boot process shall be delayed by external network
8026 * When hibernating, systemd will now inform the kernel of the image
8027 write offset, on kernels new enough to support this. This means swap
8028 files should work for hibernation now.
8030 * When loading unit files, systemd will now look for drop-in unit files
8031 extensions in additional places. Previously, for a unit file name
8032 "foo-bar-baz.service" it would look for dropin files in
8033 "foo-bar-baz.service.d/*.conf". Now, it will also look in
8034 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
8035 service name truncated after all inner dashes. This scheme allows
8036 writing drop-ins easily that apply to a whole set of unit files at
8037 once. It's particularly useful for mount and slice units (as their
8038 naming is prefix based), but is also useful for service and other
8039 units, for packages that install multiple unit files at once,
8040 following a strict naming regime of beginning the unit file name with
8041 the package's name. Two new specifiers are now supported in unit
8042 files to match this: %j and %J are replaced by the part of the unit
8043 name following the last dash.
8045 * Unit files and other configuration files that support specifier
8046 expansion now understand another three new specifiers: %T and %V will
8047 resolve to /tmp and /var/tmp respectively, or whatever temporary
8048 directory has been set for the calling user. %E will expand to either
8049 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
8051 * The ExecStart= lines of unit files are no longer required to
8052 reference absolute paths. If non-absolute paths are specified the
8053 specified binary name is searched within the service manager's
8054 built-in $PATH, which may be queried with 'systemd-path
8055 search-binaries-default'. It's generally recommended to continue to
8056 use absolute paths for all binaries specified in unit files.
8058 * Units gained a new load state "bad-setting", which is used when a
8059 unit file was loaded, but contained fatal errors which prevent it
8060 from being started (for example, a service unit has been defined
8061 lacking both ExecStart= and ExecStop= lines).
8063 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
8064 support alternative debuggers, for example lldb. The old name
8065 continues to be available however, for compatibility reasons. Use the
8066 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
8067 to pick an alternative debugger instead of the default gdb.
8069 * systemctl and the other tools will now output escape sequences that
8070 generate proper clickable hyperlinks in various terminal emulators
8071 where useful (for example, in the "systemctl status" output you can
8072 now click on the unit file name to quickly open it in the
8073 editor/viewer of your choice). Note that not all terminal emulators
8074 support this functionality yet, but many do. Unfortunately, the
8075 "less" pager doesn't support this yet, hence this functionality is
8076 currently automatically turned off when a pager is started (which
8077 happens quite often due to auto-paging). We hope to remove this
8078 limitation as soon as "less" learns these escape sequences. This new
8079 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
8080 environment variable. For details on these escape sequences see:
8081 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
8083 * networkd's .network files now support a new IPv6MTUBytes= option for
8084 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
8085 option in the [Route] section to configure the MTU to use for
8086 specific routes. It also gained support for configuration of the DHCP
8087 "UserClass" option through the new UserClass= setting. It gained
8088 three new options in the new [CAN] section for configuring CAN
8089 networks. The MULTICAST and ALLMULTI interface flags may now be
8090 controlled explicitly with the new Multicast= and AllMulticast=
8093 * networkd will now automatically make use of the kernel's route
8094 expiration feature, if it is available.
8096 * udevd's .link files now support setting the number of receive and
8097 transmit channels, using the RxChannels=, TxChannels=,
8098 OtherChannels=, CombinedChannels= settings.
8100 * Support for UDPSegmentationOffload= has been removed, given its
8101 limited support in hardware, and waning software support.
8103 * networkd's .netdev files now support creating "netdevsim" interfaces.
8105 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
8106 to query the unit belonging to a specific kernel control group.
8108 * systemd-analyze gained a new verb "cat-config", which may be used to
8109 dump the contents of any configuration file, with all its matching
8110 drop-in files added in, and honouring the usual search and masking
8111 logic applied to systemd configuration files. For example use
8112 "systemd-analyze cat-config systemd/system.conf" to get the complete
8113 system configuration file of systemd how it would be loaded by PID 1
8114 itself. Similar to this, various tools such as systemd-tmpfiles or
8115 systemd-sysusers, gained a new option "--cat-config", which does the
8116 corresponding operation for their own configuration settings. For
8117 example, "systemd-tmpfiles --cat-config" will now output the full
8118 list of tmpfiles.d/ lines in place.
8120 * timedatectl gained three new verbs: "show" shows bus properties of
8121 systemd-timedated, "timesync-status" shows the current NTP
8122 synchronization state of systemd-timesyncd, and "show-timesync"
8123 shows bus properties of systemd-timesyncd.
8125 * systemd-timesyncd gained a bus interface on which it exposes details
8128 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
8129 understood by systemd-timedated. It takes a colon-separated list of
8130 unit names of NTP client services. The list is used by
8131 "timedatectl set-ntp".
8133 * systemd-nspawn gained a new --rlimit= switch for setting initial
8134 resource limits for the container payload. There's a new switch
8135 --hostname= to explicitly override the container's hostname. A new
8136 --no-new-privileges= switch may be used to control the
8137 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
8138 --oom-score-adjust= switch controls the OOM scoring adjustment value
8139 for the payload. The new --cpu-affinity= switch controls the CPU
8140 affinity of the container payload. The new --resolv-conf= switch
8141 allows more detailed control of /etc/resolv.conf handling of the
8142 container. Similarly, the new --timezone= switch allows more detailed
8143 control of /etc/localtime handling of the container.
8145 * systemd-detect-virt gained a new --list switch, which will print a
8146 list of all currently known VM and container environments.
8148 * Support for "Portable Services" has been added, see
8149 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
8150 experimental, but this is expected to change soon. Reflecting this
8151 experimental state, the "portablectl" binary is not installed into
8152 /usr/bin yet. The binary has to be called with the full path
8153 /usr/lib/systemd/portablectl instead.
8155 * journalctl's and systemctl's -o switch now knows a new log output
8156 mode "with-unit". The output it generates is very similar to the
8157 regular "short" mode, but displays the unit name instead of the
8158 syslog tag for each log line. Also, the date is shown with timezone
8159 information. This mode is probably more useful than the classic
8160 "short" output mode for most purposes, except where pixel-perfect
8161 compatibility with classic /var/log/messages formatting is required.
8163 * A new --dump-bus-properties switch has been added to the systemd
8164 binary, which may be used to dump all supported D-Bus properties.
8165 (Options which are still supported, but are deprecated, are *not*
8168 * sd-bus gained a set of new calls:
8169 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
8170 enable/disable the "floating" state of a bus slot object,
8171 i.e. whether the slot object pins the bus it is allocated for into
8172 memory or if the bus slot object gets disconnected when the bus goes
8173 away. sd_bus_open_with_description(),
8174 sd_bus_open_user_with_description(),
8175 sd_bus_open_system_with_description() may be used to allocate bus
8176 objects and set their description string already during allocation.
8178 * sd-event gained support for watching inotify events from the event
8179 loop, in an efficient way, sharing inotify handles between multiple
8180 users. For this a new function sd_event_add_inotify() has been added.
8182 * sd-event and sd-bus gained support for calling special user-supplied
8183 destructor functions for userdata pointers associated with
8184 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
8185 functions sd_bus_slot_set_destroy_callback,
8186 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
8187 sd_bus_track_get_destroy_callback,
8188 sd_event_source_set_destroy_callback,
8189 sd_event_source_get_destroy_callback have been added.
8191 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
8193 * PID 1 will now automatically reschedule .timer units whenever the
8194 local timezone changes. (They previously got rescheduled
8195 automatically when the system clock changed.)
8197 * New documentation has been added to document cgroups delegation,
8198 portable services and the various code quality tools we have set up:
8200 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
8201 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
8202 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
8204 * The Boot Loader Specification has been added to the source tree.
8206 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
8208 While moving it into our source tree we have updated it and further
8209 changes are now accepted through the usual github PR workflow.
8211 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
8212 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
8213 earlier PAM modules. The data in these fields is used to initialize
8214 the session scope's resource properties. Thus external PAM modules
8215 may now configure per-session limits, for example sourced from
8216 external user databases.
8218 * socket units with Accept=yes will now maintain a "refused" counter in
8219 addition to the existing "accepted" counter, counting connections
8220 refused due to the enforced limits.
8222 * The "systemd-path search-binaries-default" command may now be use to
8223 query the default, built-in $PATH PID 1 will pass to the services it
8226 * A new unit file setting PrivateMounts= has been added. It's a boolean
8227 option. If enabled the unit's processes are invoked in their own file
8228 system namespace. Note that this behaviour is also implied if any
8229 other file system namespacing options (such as PrivateTmp=,
8230 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
8231 primarily useful for services that do not use any of the other file
8232 system namespacing options. One such service is systemd-udevd.service
8233 where this is now used by default.
8235 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
8236 when the system is booted in UEFI "secure mode".
8238 * A new unit "system-update-pre.target" is added, which defines an
8239 optional synchronization point for offline system updates, as
8240 implemented by the pre-existing "system-update.target" unit. It
8241 allows ordering services before the service that executes the actual
8242 update process in a generic way.
8244 * Systemd now emits warnings whenever .include syntax is used.
8246 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
8247 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
8248 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
8249 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
8250 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
8251 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
8252 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
8253 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
8254 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
8255 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
8256 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
8257 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
8258 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
8259 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
8260 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
8261 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
8262 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
8263 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
8264 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
8265 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
8266 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
8267 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
8268 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
8269 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
8270 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
8271 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
8272 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
8273 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
8274 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8276 — Berlin, 2018-06-22
8280 * The MemoryAccounting= unit property now defaults to on. After
8281 discussions with the upstream control group maintainers we learnt
8282 that the negative impact of cgroup memory accounting on current
8283 kernels is finally relatively minimal, so that it should be safe to
8284 enable this by default without affecting system performance. Besides
8285 memory accounting only task accounting is turned on by default, all
8286 other forms of resource accounting (CPU, IO, IP) remain off for now,
8287 because it's not clear yet that their impact is small enough to move
8288 from opt-in to opt-out. We recommend downstreams to leave memory
8289 accounting on by default if kernel 4.14 or higher is primarily
8290 used. On very resource constrained systems or when support for old
8291 kernels is a necessity, -Dmemory-accounting-default=false can be used
8292 to revert this change.
8294 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
8295 %udev_rules_update) and the journal catalog (%journal_catalog_update)
8296 from the upgrade scriptlets of individual packages now do nothing.
8297 Transfiletriggers have been added which will perform those updates
8298 once at the end of the transaction.
8300 Similar transfiletriggers have been added to execute any sysctl.d
8301 and binfmt.d rules. Thus, it should be unnecessary to provide any
8302 scriptlets to execute this configuration from package installation
8305 * systemd-sysusers gained a mode where the configuration to execute is
8306 specified on the command line, but this configuration is not executed
8307 directly, but instead it is merged with the configuration on disk,
8308 and the result is executed. This is useful for package installation
8309 scripts which want to create the user before installing any files on
8310 disk (in case some of those files are owned by that user), while
8311 still allowing local admin overrides.
8313 This functionality is exposed to rpm scriptlets through a new
8314 %sysusers_create_package macro. Old %sysusers_create and
8315 %sysusers_create_inline macros are deprecated.
8317 A transfiletrigger for sysusers.d configuration is now installed,
8318 which means that it should be unnecessary to call systemd-sysusers from
8319 package installation scripts, unless the package installs any files
8320 owned by those newly-created users, in which case
8321 %sysusers_create_package should be used.
8323 * Analogous change has been done for systemd-tmpfiles: it gained a mode
8324 where the command-line configuration is merged with the configuration
8325 on disk. This is exposed as the new %tmpfiles_create_package macro,
8326 and %tmpfiles_create is deprecated. A transfiletrigger is installed
8327 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
8328 from package installation scripts.
8330 * sysusers.d configuration for a user may now also specify the group
8331 number, in addition to the user number ("u username 123:456"), or
8332 without the user number ("u username -:456").
8334 * Configution items for systemd-sysusers can now be specified as
8335 positional arguments when the new --inline switch is used.
8337 * The login shell of users created through sysusers.d may now be
8338 specified (previously, it was always /bin/sh for root and
8339 /sbin/nologin for other users).
8341 * systemd-analyze gained a new --global switch to look at global user
8342 configuration. It also gained a unit-paths verb to list the unit load
8343 paths that are compiled into systemd (which can be used with
8344 --systemd, --user, or --global).
8346 * udevadm trigger gained a new --settle/-w option to wait for any
8347 triggered events to finish (but just those, and not any other events
8348 which are triggered meanwhile).
8350 * The action that systemd-logind takes when the lid is closed and the
8351 machine is connected to external power can now be configured using
8352 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
8353 was determined by HandleLidSwitch=, and, for backwards compatibility,
8354 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
8356 * journalctl will periodically call sd_journal_process() to make it
8357 resilient against inotify queue overruns when journal files are
8358 rotated very quickly.
8360 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
8361 sd_bus_get_n_queued_write — may be used to check the number of
8362 pending bus messages.
8364 * systemd gained a new
8365 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
8366 which can be used to migrate foreign processes to scope and service
8367 units. The primary user for this new API is systemd itself: the
8368 systemd --user instance uses this call of the systemd --system
8369 instance to migrate processes if it itself gets the request to
8370 migrate processes and the kernel refuses this due to access
8371 restrictions. Thanks to this "systemd-run --scope --user …" works
8372 again in pure cgroup v2 environments when invoked from the user
8375 * A new TemporaryFileSystem= setting can be used to mask out part of
8376 the real file system tree with tmpfs mounts. This may be combined
8377 with BindPaths= and BindReadOnlyPaths= to hide files or directories
8378 not relevant to the unit, while still allowing some paths lower in
8379 the tree to be accessed.
8381 ProtectHome=tmpfs may now be used to hide user home and runtime
8382 directories from units, in a way that is mostly equivalent to
8383 "TemporaryFileSystem=/home /run/user /root".
8385 * Non-service units are now started with KeyringMode=shared by default.
8386 This means that mount and swapon and other mount tools have access
8387 to keys in the main keyring.
8389 * /sys/fs/bpf is now mounted automatically.
8391 * QNX virtualization is now detected by systemd-detect-virt and may
8392 be used in ConditionVirtualization=.
8394 * IPAccounting= may now be enabled also for slice units.
8396 * A new -Dsplit-bin= build configuration switch may be used to specify
8397 whether bin and sbin directories are merged, or if they should be
8398 included separately in $PATH and various listings of executable
8399 directories. The build configuration scripts will try to autodetect
8400 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
8401 system, but distributions are encouraged to configure this
8404 * A new -Dok-color= build configuration switch may be used to change
8405 the colour of "OK" status messages.
8407 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
8408 PrivateNetwork=yes was buggy in previous versions of systemd. This
8409 means that after the upgrade and daemon-reexec, any such units must
8412 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
8413 will not exclude read-only files owned by root from cleanup.
8415 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
8416 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
8417 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
8418 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
8419 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
8420 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
8421 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
8422 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
8423 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
8424 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
8425 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
8426 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
8427 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
8428 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
8429 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
8430 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
8432 — Warsaw, 2018-03-05
8436 * Some keyboards come with a zoom see-saw or rocker which until now got
8437 mapped to the Linux "zoomin/out" keys in hwdb. However, these
8438 keycodes are not recognized by any major desktop. They now produce
8439 Up/Down key events so that they can be used for scrolling.
8441 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
8442 slightly: previously, if an argument was specified for lines of this
8443 type (i.e. the right-most column was set) this string was appended to
8444 existing files each time systemd-tmpfiles was run. This behaviour was
8445 different from what the documentation said, and not particularly
8446 useful, as repeated systemd-tmpfiles invocations would not be
8447 idempotent and grow such files without bounds. With this release
8448 behaviour has been altered to match what the documentation says:
8449 lines of this type only have an effect if the indicated files don't
8450 exist yet, and only then the argument string is written to the file.
8452 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
8453 systemd-tmpfiles behaviour: previously, read-only files owned by root
8454 were always excluded from the file "aging" algorithm (i.e. the
8455 automatic clean-up of directories like /tmp based on
8456 atime/mtime/ctime). We intend to drop this restriction, and age files
8457 by default even when owned by root and read-only. This behaviour was
8458 inherited from older tools, but there have been requests to remove
8459 it, and it's not obvious why this restriction was made in the first
8460 place. Please speak up now, if you are aware of software that requires
8461 this behaviour, otherwise we'll remove the restriction in v238.
8463 * A new environment variable $SYSTEMD_OFFLINE is now understood by
8464 systemctl. It takes a boolean argument. If on, systemctl assumes it
8465 operates on an "offline" OS tree, and will not attempt to talk to the
8466 service manager. Previously, this mode was implicitly enabled if a
8467 chroot() environment was detected, and this new environment variable
8468 now provides explicit control.
8470 * .path and .socket units may now be created transiently, too.
8471 Previously only service, mount, automount and timer units were
8472 supported as transient units. The systemd-run tool has been updated
8473 to expose this new functionality, you may hence use it now to bind
8474 arbitrary commands to path or socket activation on-the-fly from the
8475 command line. Moreover, almost all properties are now exposed for the
8476 unit types that already supported transient operation.
8478 * The systemd-mount command gained support for a new --owner= parameter
8479 which takes a user name, which is then resolved and included in uid=
8480 and gid= mount options string of the file system to mount.
8482 * A new unit condition ConditionControlGroupController= has been added
8483 that checks whether a specific cgroup controller is available.
8485 * Unit files, udev's .link files, and systemd-networkd's .netdev and
8486 .network files all gained support for a new condition
8487 ConditionKernelVersion= for checking against specific kernel
8490 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
8491 support for configuring device flags in the Flags= setting. In the
8492 same files, the [Tunnel] section gained support for configuring
8493 AllowLocalRemote=. The [Route] section in .network files gained
8494 support for configuring InitialCongestionWindow=,
8495 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
8496 understands RapidCommit=.
8498 * systemd-networkd's DHCPv6 support gained support for Prefix
8501 * sd-bus gained support for a new "watch-bind" feature. When this
8502 feature is enabled, an sd_bus connection may be set up to connect to
8503 an AF_UNIX socket in the file system as soon as it is created. This
8504 functionality is useful for writing early-boot services that
8505 automatically connect to the system bus as soon as it is started,
8506 without ugly time-based polling. systemd-networkd and
8507 systemd-resolved have been updated to make use of this
8508 functionality. busctl exposes this functionality in a new
8509 --watch-bind= command line switch.
8511 * sd-bus will now optionally synthesize a local "Connected" signal as
8512 soon as a D-Bus connection is set up fully. This message mirrors the
8513 already existing "Disconnected" signal which is synthesized when the
8514 connection is terminated. This signal is generally useful but
8515 particularly handy in combination with the "watch-bind" feature
8516 described above. Synthesizing of this message has to be requested
8517 explicitly through the new API call sd_bus_set_connected_signal(). In
8518 addition a new call sd_bus_is_ready() has been added that checks
8519 whether a connection is fully set up (i.e. between the "Connected" and
8520 "Disconnected" signals).
8522 * sd-bus gained two new calls sd_bus_request_name_async() and
8523 sd_bus_release_name_async() for asynchronously registering bus
8524 names. Similar, there is now sd_bus_add_match_async() for installing
8525 a signal match asynchronously. All of systemd's own services have
8526 been updated to make use of these calls. Doing these operations
8527 asynchronously has two benefits: it reduces the risk of deadlocks in
8528 case of cyclic dependencies between bus services, and it speeds up
8529 service initialization since synchronization points for bus
8530 round-trips are removed.
8532 * sd-bus gained two new calls sd_bus_match_signal() and
8533 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
8534 and sd_bus_add_match_async() but instead of taking a D-Bus match
8535 string take match fields as normal function parameters.
8537 * sd-bus gained two new calls sd_bus_set_sender() and
8538 sd_bus_message_set_sender() for setting the sender name of outgoing
8539 messages (either for all outgoing messages or for just one specific
8540 one). These calls are only useful in direct connections as on
8541 brokered connections the broker fills in the sender anyway,
8542 overwriting whatever the client filled in.
8544 * sd-event gained a new pseudo-handle that may be specified on all API
8545 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
8546 used this refers to the default event loop object of the calling
8547 thread. Note however that this does not implicitly allocate one —
8548 which has to be done prior by using sd_event_default(). Similarly
8549 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
8550 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
8551 to the default bus of the specified type of the calling thread. Here
8552 too this does not implicitly allocate bus connection objects, this
8553 has to be done prior with sd_bus_default() and friends.
8555 * sd-event gained a new call pair
8556 sd_event_source_{get|set}_io_fd_own(). This may be used to request
8557 automatic closure of the file descriptor an IO event source watches
8558 when the event source is destroyed.
8560 * systemd-networkd gained support for natively configuring WireGuard
8563 * In previous versions systemd synthesized user records both for the
8564 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
8565 internally. In order to simplify distribution-wide renames of the
8566 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
8567 new transitional flag file has been added: if
8568 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
8569 user and group record within the systemd codebase is disabled.
8571 * systemd-notify gained a new --uid= option for selecting the source
8572 user/UID to use for notification messages sent to the service
8575 * journalctl gained a new --grep= option to list only entries in which
8576 the message matches a certain pattern. By default matching is case
8577 insensitive if the pattern is lowercase, and case sensitive
8578 otherwise. Option --case-sensitive=yes|no can be used to override
8579 this an specify case sensitivity or case insensitivity.
8581 * There's now a "systemd-analyze service-watchdogs" command for printing
8582 the current state of the service runtime watchdog, and optionally
8583 enabling or disabling the per-service watchdogs system-wide if given a
8584 boolean argument (i.e. the concept you configure in WatchdogSec=), for
8585 debugging purposes. There's also a kernel command line option
8586 systemd.service_watchdogs= for controlling the same.
8588 * Two new "log-level" and "log-target" options for systemd-analyze were
8589 added that merge the now deprecated get-log-level, set-log-level and
8590 get-log-target, set-log-target pairs. The deprecated options are still
8591 understood for backwards compatibility. The two new options print the
8592 current value when no arguments are given, and set them when a
8593 level/target is given as an argument.
8595 * sysusers.d's "u" lines now optionally accept both a UID and a GID
8596 specification, separated by a ":" character, in order to create users
8597 where UID and GID do not match.
8599 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
8600 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
8601 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
8602 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
8603 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
8604 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
8605 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
8606 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
8607 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
8608 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
8609 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
8610 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
8611 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
8612 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
8613 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
8614 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
8615 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
8616 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
8617 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
8618 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
8625 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
8626 in v235 has been extended to also set the dummy.ko module option
8627 numdummies=0, preventing the kernel from automatically creating
8628 dummy0. All dummy interfaces must now be explicitly created.
8630 * Unknown '%' specifiers in configuration files are now rejected. This
8631 applies to units and tmpfiles.d configuration. Any percent characters
8632 that are followed by a letter or digit that are not supposed to be
8633 interpreted as the beginning of a specifier should be escaped by
8634 doubling ("%%"). (So "size=5%" is still accepted, as well as
8635 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
8636 valid specifiers today.)
8638 * systemd-resolved now maintains a new dynamic
8639 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
8640 recommended to make /etc/resolv.conf a symlink to it. This file
8641 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
8642 includes dynamically acquired search domains, achieving more correct
8643 DNS resolution by software that bypasses local DNS APIs such as NSS.
8645 * The "uaccess" udev tag has been dropped from /dev/kvm and
8646 /dev/dri/renderD*. These devices now have the 0666 permissions by
8647 default (but this may be changed at build-time). /dev/dri/renderD*
8648 will now be owned by the "render" group along with /dev/kfd.
8650 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
8651 systemd-journal-gatewayd.service and
8652 systemd-journal-upload.service. This means "nss-systemd" must be
8653 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
8654 services are resolved properly.
8656 * In /etc/fstab two new mount options are now understood:
8657 x-systemd.makefs and x-systemd.growfs. The former has the effect that
8658 the configured file system is formatted before it is mounted, the
8659 latter that the file system is resized to the full block device size
8660 after it is mounted (i.e. if the file system is smaller than the
8661 partition it resides on, it's grown). This is similar to the fsck
8662 logic in /etc/fstab, and pulls in systemd-makefs@.service and
8663 systemd-growfs@.service as necessary, similar to
8664 systemd-fsck@.service. Resizing is currently only supported on ext4
8667 * In systemd-networkd, the IPv6 RA logic now optionally may announce
8668 DNS server and domain information.
8670 * Support for the LUKS2 on-disk format for encrypted partitions has
8671 been added. This requires libcryptsetup2 during compilation and
8674 * The systemd --user instance will now signal "readiness" when its
8675 basic.target unit has been reached, instead of when the run queue ran
8676 empty for the first time.
8678 * Tmpfiles.d with user configuration are now also supported.
8679 systemd-tmpfiles gained a new --user switch, and snippets placed in
8680 ~/.config/user-tmpfiles.d/ and corresponding directories will be
8681 executed by systemd-tmpfiles --user running in the new
8682 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
8683 running in the user session.
8685 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
8686 %S resolves to the top-level state directory (/var/lib for the system
8687 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
8688 top-level cache directory (/var/cache for the system instance,
8689 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
8690 logs directory (/var/log for the system instance,
8691 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
8692 existing %t specifier, that resolves to the top-level runtime
8693 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
8696 * journalctl learnt a new parameter --output-fields= for limiting the
8697 set of journal fields to output in verbose and JSON output modes.
8699 * systemd-timesyncd's configuration file gained a new option
8700 RootDistanceMaxSec= for setting the maximum root distance of servers
8701 it'll use, as well as the new options PollIntervalMinSec= and
8702 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
8704 * bootctl gained a new command "list" for listing all available boot
8705 menu items on systems that follow the boot loader specification.
8707 * systemctl gained a new --dry-run switch that shows what would be done
8708 instead of doing it, and is currently supported by the shutdown and
8711 * ConditionSecurity= can now detect the TOMOYO security module.
8713 * Unit file [Install] sections are now also respected in unit drop-in
8714 files. This is intended to be used by drop-ins under /usr/lib/.
8716 * systemd-firstboot may now also set the initial keyboard mapping.
8718 * Udev "changed" events for devices which are exposed as systemd
8719 .device units are now propagated to units specified in
8720 ReloadPropagatedFrom= as reload requests.
8722 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
8723 unit template name (i.e. a name in the form of 'foobar@.service',
8724 without the instance component between the '@' and - the '.'), then
8725 the escaped sysfs path of the device is automatically used as the
8728 * SystemCallFilter= in unit files has been extended so that an "errno"
8729 can be specified individually for each system call. Example:
8730 SystemCallFilter=~uname:EILSEQ.
8732 * The cgroup delegation logic has been substantially updated. Delegate=
8733 now optionally takes a list of controllers (instead of a boolean, as
8734 before), which lists the controllers to delegate at least.
8736 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
8738 * A new LogLevelMax= setting configures the maximum log level any
8739 process of the service may log at (i.e. anything with a lesser
8740 priority than what is specified is automatically dropped). A new
8741 LogExtraFields= setting allows configuration of additional journal
8742 fields to attach to all log records generated by any of the unit's
8745 * New StandardInputData= and StandardInputText= settings along with the
8746 new option StandardInput=data may be used to configure textual or
8747 binary data that shall be passed to the executed service process via
8748 standard input, encoded in-line in the unit file.
8750 * StandardInput=, StandardOutput= and StandardError= may now be used to
8751 connect stdin/stdout/stderr of executed processes directly with a
8752 file or AF_UNIX socket in the file system, using the new "file:" option.
8754 * A new unit file option CollectMode= has been added, that allows
8755 tweaking the garbage collection logic for units. It may be used to
8756 tell systemd to garbage collect units that have failed automatically
8757 (normally it only GCs units that exited successfully). systemd-run
8758 and systemd-mount expose this new functionality with a new -G option.
8760 * "machinectl bind" may now be used to bind mount non-directories
8761 (i.e. regularfiles, devices, fifos, sockets).
8763 * systemd-analyze gained a new verb "calendar" for validating and
8764 testing calendar time specifications to use for OnCalendar= in timer
8765 units. Besides validating the expression it will calculate the next
8766 time the specified expression would elapse.
8768 * In addition to the pre-existing FailureAction= unit file setting
8769 there's now SuccessAction=, for configuring a shutdown action to
8770 execute when a unit completes successfully. This is useful in
8771 particular inside containers that shall terminate after some workload
8772 has been completed. Also, both options are now supported for all unit
8773 types, not just services.
8775 * networkds's IP rule support gained two new options
8776 IncomingInterface= and OutgoingInterface= for configuring the incoming
8777 and outgoing interfaces of configured rules. systemd-networkd also
8778 gained support for "vxcan" network devices.
8780 * networkd gained a new setting RequiredForOnline=, taking a
8781 boolean. If set, systemd-wait-online will take it into consideration
8782 when determining that the system is up, otherwise it will ignore the
8783 interface for this purpose.
8785 * The sd_notify() protocol gained support for a new operation: with
8786 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
8787 store again, ahead of POLLHUP or POLLERR when they are removed
8790 * A new document doc/UIDS-GIDS.md has been added to the source tree,
8791 that documents the UID/GID range and assignment assumptions and
8792 requirements of systemd.
8794 * The watchdog device PID 1 will ping may now be configured through the
8795 WatchdogDevice= configuration file setting, or by setting the
8796 systemd.watchdog_service= kernel command line option.
8798 * systemd-resolved's gained support for registering DNS-SD services on
8799 the local network using MulticastDNS. Services may either be
8800 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
8801 the same dir below /run, /usr/lib), or through its D-Bus API.
8803 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
8804 extend the effective start, runtime, and stop time. The service must
8805 continue to send EXTEND_TIMEOUT_USEC within the period specified to
8806 prevent the service manager from making the service as timedout.
8808 * systemd-resolved's DNSSEC support gained support for RFC 8080
8809 (Ed25519 keys and signatures).
8811 * The systemd-resolve command line tool gained a new set of options
8812 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
8813 --set-nta= and --revert to configure per-interface DNS configuration
8814 dynamically during runtime. It's useful for pushing DNS information
8815 into systemd-resolved from DNS hook scripts that various interface
8816 managing software supports (such as pppd).
8818 * systemd-nspawn gained a new --network-namespace-path= command line
8819 option, which may be used to make a container join an existing
8820 network namespace, by specifying a path to a "netns" file.
8822 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
8823 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
8824 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
8825 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
8826 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
8827 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
8828 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
8829 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
8830 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
8831 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
8832 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
8833 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
8834 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
8835 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
8836 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
8837 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
8838 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
8839 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
8840 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
8841 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
8842 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
8843 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
8844 Jędrzejewski-Szmek, Zeal Jagannatha
8846 — Berlin, 2017-12-14
8850 * INCOMPATIBILITY: systemd-logind.service and other long-running
8851 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
8852 communication with the outside. This generally improves security of
8853 the system, and is in almost all cases a safe and good choice, as
8854 these services do not and should not provide any network-facing
8855 functionality. However, systemd-logind uses the glibc NSS API to
8856 query the user database. This creates problems on systems where NSS
8857 is set up to directly consult network services for user database
8858 lookups. In particular, this creates incompatibilities with the
8859 "nss-nis" module, which attempts to directly contact the NIS/YP
8860 network servers it is configured for, and will now consistently
8861 fail. In such cases, it is possible to turn off IP sandboxing for
8862 systemd-logind.service (set IPAddressDeny= in its [Service] section
8863 to the empty string, via a .d/ unit file drop-in). Downstream
8864 distributions might want to update their nss-nis packaging to include
8865 such a drop-in snippet, accordingly, to hide this incompatibility
8866 from the user. Another option is to make use of glibc's nscd service
8867 to proxy such network requests through a privilege-separated, minimal
8868 local caching daemon, or to switch to more modern technologies such
8869 sssd, whose NSS hook-ups generally do not involve direct network
8870 access. In general, we think it's definitely time to question the
8871 implementation choices of nss-nis, i.e. whether it's a good idea
8872 today to embed a network-facing loadable module into all local
8873 processes that need to query the user database, including the most
8874 trivial and benign ones, such as "ls". For more details about
8875 IPAddressDeny= see below.
8877 * A new modprobe.d drop-in is now shipped by default that sets the
8878 bonding module option max_bonds=0. This overrides the kernel default,
8879 to avoid conflicts and ambiguity as to whether or not bond0 should be
8880 managed by systemd-networkd or not. This resolves multiple issues
8881 with bond0 properties not being applied, when bond0 is configured
8882 with systemd-networkd. Distributors may choose to not package this,
8883 however in that case users will be prevented from correctly managing
8884 bond0 interface using systemd-networkd.
8886 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
8887 which print the logging level and target of the system manager. They
8888 complement the existing "set-log-level" and "set-log-target" verbs
8889 used to change those values.
8891 * journald.conf gained a new boolean setting ReadKMsg= which defaults
8892 to on. If turned off kernel log messages will not be read by
8893 systemd-journald or included in the logs. It also gained a new
8894 setting LineMax= for configuring the maximum line length in
8895 STDOUT/STDERR log streams. The new default for this value is 48K, up
8896 from the previous hardcoded 2048.
8898 * A new unit setting RuntimeDirectoryPreserve= has been added, which
8899 allows more detailed control of what to do with a runtime directory
8900 configured with RuntimeDirectory= (i.e. a directory below /run or
8901 $XDG_RUNTIME_DIR) after a unit is stopped.
8903 * The RuntimeDirectory= setting for units gained support for creating
8904 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
8905 one top-level directory.
8907 * Units gained new options StateDirectory=, CacheDirectory=,
8908 LogsDirectory= and ConfigurationDirectory= which are closely related
8909 to RuntimeDirectory= but manage per-service directories below
8910 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
8911 possible to write unit files which when activated automatically gain
8912 properly owned service specific directories in these locations, thus
8913 making unit files self-contained and increasing compatibility with
8914 stateless systems and factory reset where /etc or /var are
8915 unpopulated at boot. Matching these new settings there's also
8916 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
8917 ConfigurationDirectoryMode= for configuring the access mode of these
8918 directories. These settings are particularly useful in combination
8919 with DynamicUser=yes as they provide secure, properly-owned,
8920 writable, and stateful locations for storage, excluded from the
8921 sandbox that such services live in otherwise.
8923 * Automake support has been removed from this release. systemd is now
8926 * systemd-journald will now aggressively cache client metadata during
8927 runtime, speeding up log write performance under pressure. This comes
8928 at a small price though: as much of the metadata is read
8929 asynchronously from /proc/ (and isn't implicitly attached to log
8930 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
8931 metadata stored alongside a log entry might be slightly
8932 out-of-date. Previously it could only be slightly newer than the log
8933 message. The time window is small however, and given that the kernel
8934 is unlikely to be improved anytime soon in this regard, this appears
8937 * nss-myhostname/systemd-resolved will now by default synthesize an
8938 A/AAAA resource record for the "_gateway" hostname, pointing to the
8939 current default IP gateway. Previously it did that for the "gateway"
8940 name, hampering adoption, as some distributions wanted to leave that
8941 hostname open for local use. The old behaviour may still be
8942 requested at build time.
8944 * systemd-networkd's [Address] section in .network files gained a new
8945 Scope= setting for configuring the IP address scope. The [Network]
8946 section gained a new boolean setting ConfigureWithoutCarrier= that
8947 tells systemd-networkd to ignore link sensing when configuring the
8948 device. The [DHCP] section gained a new Anonymize= boolean option for
8949 turning on a number of options suggested in RFC 7844. A new
8950 [RoutingPolicyRule] section has been added for configuring the IP
8951 routing policy. The [Route] section has gained support for a new
8952 Type= setting which permits configuring
8953 blackhole/unreachable/prohibit routes.
8955 * The [VRF] section in .netdev files gained a new Table= setting for
8956 configuring the routing table to use. The [Tunnel] section gained a
8957 new Independent= boolean field for configuring tunnels independent of
8958 an underlying network interface. The [Bridge] section gained a new
8959 GroupForwardMask= option for configuration of propagation of link
8960 local frames between bridge ports.
8962 * The WakeOnLan= setting in .link files gained support for a number of
8963 new modes. A new TCP6SegmentationOffload= setting has been added for
8964 configuring TCP/IPv6 hardware segmentation offload.
8966 * The IPv6 RA sender implementation may now optionally send out RDNSS
8967 and RDNSSL records to supply DNS configuration to peers.
8969 * systemd-nspawn gained support for a new --system-call-filter= command
8970 line option for adding and removing entries in the default system
8971 call filter it applies. Moreover systemd-nspawn has been changed to
8972 implement a system call allow list instead of a deny list.
8974 * systemd-run gained support for a new --pipe command line option. If
8975 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
8976 are directly passed on to the activated transient service
8977 executable. This allows invoking arbitrary processes as systemd
8978 services (for example to take benefit of dependency management,
8979 accounting management, resource management or log management that is
8980 done automatically for services) — while still allowing them to be
8981 integrated in a classic UNIX shell pipeline.
8983 * When a service sends RELOAD=1 via sd_notify() and reload propagation
8984 using ReloadPropagationTo= is configured, a reload is now propagated
8985 to configured units. (Previously this was only done on explicitly
8986 requested reloads, using "systemctl reload" or an equivalent
8989 * For each service unit a restart counter is now kept: it is increased
8990 each time the service is restarted due to Restart=, and may be
8991 queried using "systemctl show -p NRestarts …".
8993 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
8994 @signal and @timer have been added, for usage with SystemCallFilter=
8995 in unit files and the new --system-call-filter= command line option
8996 of systemd-nspawn (see above).
8998 * ExecStart= lines in unit files gained two new modifiers: when a
8999 command line is prefixed with "!" the command will be executed as
9000 configured, except for the credentials applied by
9001 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
9002 "+", but does still apply namespacing options unlike "+". There's
9003 also "!!" now, which is mostly identical, but becomes a NOP on
9004 systems that support ambient capabilities. This is useful to write
9005 unit files that work with ambient capabilities where possible but
9006 automatically fall back to traditional privilege dropping mechanisms
9007 on systems where this is not supported.
9009 * ListenNetlink= settings in socket units now support RDMA netlink
9012 * A new unit file setting LockPersonality= has been added which permits
9013 locking down the chosen execution domain ("personality") of a service
9016 * A new special target "getty-pre.target" has been added, which is
9017 ordered before all text logins, and may be used to order services
9018 before textual logins acquire access to the console.
9020 * systemd will now attempt to load the virtio-rng.ko kernel module very
9021 early on if a VM environment supporting this is detected. This should
9022 improve entropy during early boot in virtualized environments.
9024 * A _netdev option is now supported in /etc/crypttab that operates in a
9025 similar way as the same option in /etc/fstab: it permits configuring
9026 encrypted devices that need to be ordered after the network is up.
9027 Following this logic, two new special targets
9028 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
9029 added that are to cryptsetup.target what remote-fs.target and
9030 remote-fs-pre.target are to local-fs.target.
9032 * Service units gained a new UnsetEnvironment= setting which permits
9033 unsetting specific environment variables for services that are
9034 normally passed to it (for example in order to mask out locale
9035 settings for specific services that can't deal with it).
9037 * Units acquired a new boolean option IPAccounting=. When turned on, IP
9038 traffic accounting (packet count as well as byte count) is done for
9039 the service, and shown as part of "systemctl status" or "systemd-run
9042 * Service units acquired two new options IPAddressAllow= and
9043 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
9044 for configuring a simple IP access control list for all sockets of
9045 the unit. These options are available also on .slice and .socket
9046 units, permitting flexible access list configuration for individual
9047 services as well as groups of services (as defined by a slice unit),
9048 including system-wide. Note that IP ACLs configured this way are
9049 enforced on every single IPv4 and IPv6 socket created by any process
9050 of the service unit, and apply to ingress as well as egress traffic.
9052 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
9053 structured log message is generated each time the unit is stopped,
9054 containing information about the consumed resources of this
9057 * A new setting KeyringMode= has been added to unit files, which may be
9058 used to control how the kernel keyring is set up for executed
9061 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
9062 "systemctl kexec" and "systemctl exit" are now always asynchronous in
9063 behaviour (that is: these commands return immediately after the
9064 operation was enqueued instead of waiting for the operation to
9065 complete). Previously, "systemctl poweroff" and "systemctl reboot"
9066 were asynchronous on systems using systemd-logind (i.e. almost
9067 always, and like they were on sysvinit), and the other three commands
9068 were unconditionally synchronous. With this release this is cleaned
9069 up, and callers will see the same asynchronous behaviour on all
9070 systems for all five operations.
9072 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
9075 * .timer units now accept calendar specifications in other timezones
9076 than UTC or the local timezone.
9078 * The tmpfiles snippet var.conf has been changed to create
9079 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
9080 the "utmp" group already, and it appears to be generally understood
9081 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
9082 databases. Previously this was implemented correctly for all these
9083 databases excepts btmp, which has been opened up like this now
9084 too. Note that while the other databases are world-readable
9085 (i.e. 0644), btmp is not and remains more restrictive.
9087 * The systemd-resolve tool gained a new --reset-server-features
9088 switch. When invoked like this systemd-resolved will forget
9089 everything it learnt about the features supported by the configured
9090 upstream DNS servers, and restarts the feature probing logic on the
9091 next resolver look-up for them at the highest feature level
9094 * The status dump systemd-resolved sends to the logs upon receiving
9095 SIGUSR1 now also includes information about all DNS servers it is
9096 configured to use, and the features levels it probed for them.
9098 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
9099 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
9100 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
9101 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
9102 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
9103 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
9104 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
9105 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
9106 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
9107 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
9108 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
9109 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
9110 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
9111 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
9112 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
9113 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
9114 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
9115 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
9116 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
9117 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9119 — Berlin, 2017-10-06
9123 * Meson is now supported as build system in addition to Automake. It is
9124 our plan to remove Automake in one of our next releases, so that
9125 Meson becomes our exclusive build system. Hence, please start using
9126 the Meson build system in your downstream packaging. There's plenty
9127 of documentation around how to use Meson, the extremely brief
9130 ./autogen.sh && ./configure && make && sudo make install
9134 meson build && ninja -C build && sudo ninja -C build install
9136 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
9137 which permits configuring a timeout on the time a job is
9138 running. This is particularly useful for setting timeouts on jobs for
9141 * Unit files gained two new options ConditionUser= and ConditionGroup=
9142 for conditionalizing units based on the identity of the user/group
9143 running a systemd user instance.
9145 * systemd-networkd now understands a new FlowLabel= setting in the
9146 [VXLAN] section of .network files, as well as a Priority= in
9147 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
9148 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
9149 gained support for configuration of GENEVE links, and IPv6 address
9150 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
9152 * .link files now understand a new Port= setting.
9154 * systemd-networkd's DHCP support gained support for DHCP option 119
9155 (domain search list).
9157 * systemd-networkd gained support for serving IPv6 address ranges using
9158 the Router Advertisement protocol. The new .network configuration
9159 section [IPv6Prefix] may be used to configure the ranges to
9160 serve. This is implemented based on a new, minimal, native server
9161 implementation of RA.
9163 * journalctl's --output= switch gained support for a new parameter
9164 "short-iso-precise" for a mode where timestamps are shown as precise
9167 * systemd-udevd's "net_id" builtin may now generate stable network
9168 interface names from IBM PowerVM VIO devices as well as ACPI platform
9171 * MulticastDNS support in systemd-resolved may now be explicitly
9172 enabled/disabled using the new MulticastDNS= configuration file
9175 * systemd-resolved may now optionally use libidn2 instead of the libidn
9176 for processing internationalized domain names. Support for libidn2
9177 should be considered experimental and should not be enabled by
9180 * "machinectl pull-tar" and related call may now do verification of
9181 downloaded images using SUSE-style .sha256 checksum files in addition
9182 to the already existing support for validating using Ubuntu-style
9185 * sd-bus gained support for a new sd_bus_message_appendv() call which
9186 is va_list equivalent of sd_bus_message_append().
9188 * sd-boot gained support for validating images using SHIM/MOK.
9190 * The SMACK code learnt support for "onlycap".
9192 * systemd-mount --umount is now much smarter in figuring out how to
9193 properly unmount a device given its mount or device path.
9195 * The code to call libnss_dns as a fallback from libnss_resolve when
9196 the communication with systemd-resolved fails was removed. This
9197 fallback was redundant and interfered with the [!UNAVAIL=return]
9198 suffix. See nss-resolve(8) for the recommended configuration.
9200 * systemd-logind may now be restarted without losing state. It stores
9201 the file descriptors for devices it manages in the system manager
9202 using the FDSTORE= mechanism. Please note that further changes in
9203 other components may be required to make use of this (for example
9204 Xorg has code to listen for stops of systemd-logind and terminate
9205 itself when logind is stopped or restarted, in order to avoid using
9206 stale file descriptors for graphical devices, which is now
9207 counterproductive and must be reverted in order for restarts of
9208 systemd-logind to be safe. See
9209 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
9211 * All kernel-install plugins are called with the environment variable
9212 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
9213 /etc/machine-id. If the machine ID could not be determined,
9214 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
9215 anything in the entry directory (passed as the second argument) if
9216 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
9217 temporary directory is passed as the entry directory and removed
9218 after all the plugins exit.
9220 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
9221 will now use its value as the machine ID instead of the machine ID
9222 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
9223 /etc/machine-info and no machine ID is set in /etc/machine-id,
9224 kernel-install will try to store the current machine ID there as
9225 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
9226 will generate a new UUID, store it in /etc/machine-info as
9227 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
9229 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
9230 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
9231 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
9232 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
9233 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
9234 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
9235 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
9236 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
9237 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
9238 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
9239 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
9240 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
9241 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
9242 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
9243 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
9244 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
9245 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
9246 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
9247 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
9248 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
9249 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
9250 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
9251 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
9252 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
9253 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
9254 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
9255 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
9256 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
9257 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
9260 — Berlin, 2017-07-12
9264 * The "hybrid" control group mode has been modified to improve
9265 compatibility with "legacy" cgroups-v1 setups. Specifically, the
9266 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
9267 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
9268 cgroups-v1 hierarchy), the only externally visible change being that
9269 the cgroups-v2 hierarchy is also mounted, to
9270 /sys/fs/cgroup/unified. This should provide a large degree of
9271 compatibility with "legacy" cgroups-v1, while taking benefit of the
9272 better management capabilities of cgroups-v2.
9274 * The default control group setup mode may be selected both a boot-time
9275 via a set of kernel command line parameters (specifically:
9276 systemd.unified_cgroup_hierarchy= and
9277 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
9278 default selected on the configure command line
9279 (--with-default-hierarchy=). The upstream default is "hybrid"
9280 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
9281 this will change in a future systemd version to be "unified" (pure
9282 cgroups-v2 mode). The third option for the compile time option is
9283 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
9284 distributions to default to "hybrid" mode for release distributions,
9285 starting with v233. We recommend "unified" for development
9286 distributions (specifically: distributions such as Fedora's rawhide)
9287 as that's where things are headed in the long run. Use "legacy" for
9288 greatest stability and compatibility only.
9290 * Note one current limitation of "unified" and "hybrid" control group
9291 setup modes: the kernel currently does not permit the systemd --user
9292 instance (i.e. unprivileged code) to migrate processes between two
9293 disconnected cgroup subtrees, even if both are managed and owned by
9294 the user. This effectively means "systemd-run --user --scope" doesn't
9295 work when invoked from outside of any "systemd --user" service or
9296 scope. Specifically, it is not supported from session scopes. We are
9297 working on fixing this in a future systemd version. (See #3388 for
9298 further details about this.)
9300 * DBus policy files are now installed into /usr rather than /etc. Make
9301 sure your system has dbus >= 1.9.18 running before upgrading to this
9302 version, or override the install path with --with-dbuspolicydir= .
9304 * All python scripts shipped with systemd (specifically: the various
9305 tests written in Python) now require Python 3.
9307 * systemd unit tests can now run standalone (without the source or
9308 build directories), and can be installed into /usr/lib/systemd/tests/
9309 with 'make install-tests'.
9311 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
9312 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
9315 * Support for the %c, %r, %R specifiers in unit files has been
9316 removed. Specifiers are not supposed to be dependent on configuration
9317 in the unit file itself (so that they resolve the same regardless
9318 where used in the unit files), but these specifiers were influenced
9319 by the Slice= option.
9321 * The shell invoked by debug-shell.service now defaults to /bin/sh in
9322 all cases. If distributions want to use a different shell for this
9323 purpose (for example Fedora's /sbin/sushell) they need to specify
9324 this explicitly at configure time using --with-debug-shell=.
9326 * The confirmation spawn prompt has been reworked to offer the
9329 (c)ontinue, proceed without asking anymore
9330 (D)ump, show the state of the unit
9331 (f)ail, don't execute the command and pretend it failed
9333 (i)nfo, show a short summary of the unit
9334 (j)obs, show jobs that are in progress
9335 (s)kip, don't execute the command and pretend it succeeded
9336 (y)es, execute the command
9338 The 'n' choice for the confirmation spawn prompt has been removed,
9339 because its meaning was confusing.
9341 The prompt may now also be redirected to an alternative console by
9342 specifying the console as parameter to systemd.confirm_spawn=.
9344 * Services of Type=notify require a READY=1 notification to be sent
9345 during startup. If no such message is sent, the service now fails,
9346 even if the main process exited with a successful exit code.
9348 * Services that fail to start up correctly now always have their
9349 ExecStopPost= commands executed. Previously, they'd enter "failed"
9350 state directly, without executing these commands.
9352 * The option MulticastDNS= of network configuration files has acquired
9353 an actual implementation. With MulticastDNS=yes a host can resolve
9354 names of remote hosts and reply to mDNS A and AAAA requests.
9356 * When units are about to be started an additional check is now done to
9357 ensure that all dependencies of type BindsTo= (when used in
9358 combination with After=) have been started.
9360 * systemd-analyze gained a new verb "syscall-filter" which shows which
9361 system call groups are defined for the SystemCallFilter= unit file
9362 setting, and which system calls they contain.
9364 * A new system call filter group "@filesystem" has been added,
9365 consisting of various file system related system calls. Group
9366 "@reboot" has been added, covering reboot, kexec and shutdown related
9367 calls. Finally, group "@swap" has been added covering swap
9368 configuration related calls.
9370 * A new unit file option RestrictNamespaces= has been added that may be
9371 used to restrict access to the various process namespace types the
9372 Linux kernel provides. Specifically, it may be used to take away the
9373 right for a service unit to create additional file system, network,
9374 user, and other namespaces. This sandboxing option is particularly
9375 relevant due to the high amount of recently discovered namespacing
9376 related vulnerabilities in the kernel.
9378 * systemd-udev's .link files gained support for a new AutoNegotiation=
9379 setting for configuring Ethernet auto-negotiation.
9381 * systemd-networkd's .network files gained support for a new
9382 ListenPort= setting in the [DHCP] section to explicitly configure the
9383 UDP client port the DHCP client shall listen on.
9385 * .network files gained a new Unmanaged= boolean setting for explicitly
9386 excluding one or more interfaces from management by systemd-networkd.
9388 * The systemd-networkd ProxyARP= option has been renamed to
9389 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
9390 renamed to ReduceARPProxy=. The old names continue to be available
9393 * systemd-networkd gained support for configuring IPv6 Proxy NDP
9394 addresses via the new IPv6ProxyNDPAddress= .network file setting.
9396 * systemd-networkd's bonding device support gained support for two new
9397 configuration options ActiveSlave= and PrimarySlave=.
9399 * The various options in the [Match] section of .network files gained
9400 support for negative matching.
9402 * New systemd-specific mount options are now understood in /etc/fstab:
9404 x-systemd.mount-timeout= may be used to configure the maximum
9405 permitted runtime of the mount command.
9407 x-systemd.device-bound may be set to bind a mount point to its
9408 backing device unit, in order to automatically remove a mount point
9409 if its backing device is unplugged. This option may also be
9410 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
9411 on the block device, which is now automatically set for all CDROM
9412 drives, so that mounted CDs are automatically unmounted when they are
9413 removed from the drive.
9415 x-systemd.after= and x-systemd.before= may be used to explicitly
9416 order a mount after or before another unit or mount point.
9418 * Enqueued start jobs for device units are now automatically garbage
9419 collected if there are no jobs waiting for them anymore.
9421 * systemctl list-jobs gained two new switches: with --after, for every
9422 queued job the jobs it's waiting for are shown; with --before the
9423 jobs which it's blocking are shown.
9425 * systemd-nspawn gained support for ephemeral boots from disk images
9426 (or in other words: --ephemeral and --image= may now be
9427 combined). Moreover, ephemeral boots are now supported for normal
9428 directories, even if the backing file system is not btrfs. Of course,
9429 if the file system does not support file system snapshots or
9430 reflinks, the initial copy operation will be relatively expensive, but
9431 this should still be suitable for many use cases.
9433 * Calendar time specifications in .timer units now support
9434 specifications relative to the end of a month by using "~" instead of
9435 "-" as separator between month and day. For example, "*-02~03" means
9436 "the third last day in February". In addition a new syntax for
9437 repeated events has been added using the "/" character. For example,
9438 "9..17/2:00" means "every two hours from 9am to 5pm".
9440 * systemd-socket-proxyd gained a new parameter --connections-max= for
9441 configuring the maximum number of concurrent connections.
9443 * sd-id128 gained a new API for generating unique IDs for the host in a
9444 way that does not leak the machine ID. Specifically,
9445 sd_id128_get_machine_app_specific() derives an ID based on the
9446 machine ID in a well-defined, non-reversible, stable way. This is
9447 useful whenever an identifier for the host is needed but where the
9448 identifier shall not be useful to identify the system beyond the
9449 scope of the application itself. (Internally this uses HMAC-SHA256 as
9450 keyed hash function using the machine ID as input.)
9452 * NotifyAccess= gained a new supported value "exec". When set
9453 notifications are accepted from all processes systemd itself invoked,
9454 including all control processes.
9456 * .nspawn files gained support for defining overlay mounts using the
9457 Overlay= and OverlayReadOnly= options. Previously this functionality
9458 was only available on the systemd-nspawn command line.
9460 * systemd-nspawn's --bind= and --overlay= options gained support for
9461 bind/overlay mounts whose source lies within the container tree by
9462 prefixing the source path with "+".
9464 * systemd-nspawn's --bind= and --overlay= options gained support for
9465 automatically allocating a temporary source directory in /var/tmp
9466 that is removed when the container dies. Specifically, if the source
9467 directory is specified as empty string this mechanism is selected. An
9468 example usage is --overlay=+/var::/var, which creates an overlay
9469 mount based on the original /var contained in the image, overlaid
9470 with a temporary directory in the host's /var/tmp. This way changes
9471 to /var are automatically flushed when the container shuts down.
9473 * systemd-nspawn --image= option does now permit raw file system block
9474 devices (in addition to images containing partition tables, as
9477 * The disk image dissection logic in systemd-nspawn gained support for
9478 automatically setting up LUKS encrypted as well as Verity protected
9479 partitions. When a container is booted from an encrypted image the
9480 passphrase is queried at start-up time. When a container with Verity
9481 data is started, the root hash is search in a ".roothash" file
9482 accompanying the disk image (alternatively, pass the root hash via
9483 the new --root-hash= command line option).
9485 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
9486 be used to dissect disk images the same way as systemd-nspawn does
9487 it, following the Bootable Partition Specification. It may even be
9488 used to mount disk images with complex partition setups (including
9489 LUKS and Verity partitions) to a local host directory, in order to
9490 inspect them. This tool is not considered public API (yet), and is
9491 thus not installed into /usr/bin. Please do not rely on its
9492 existence, since it might go away or be changed in later systemd
9495 * A new generator "systemd-verity-generator" has been added, similar in
9496 style to "systemd-cryptsetup-generator", permitting automatic setup of
9497 Verity root partitions when systemd boots up. In order to make use of
9498 this your partition setup should follow the Discoverable Partitions
9499 Specification, and the GPT partition ID of the root file system
9500 partition should be identical to the upper 128-bit of the Verity root
9501 hash. The GPT partition ID of the Verity partition protecting it
9502 should be the lower 128-bit of the Verity root hash. If the partition
9503 image follows this model it is sufficient to specify a single
9504 "roothash=" kernel command line argument to both configure which root
9505 image and verity partition to use as well as the root hash for
9506 it. Note that systemd-nspawn's Verity support follows the same
9507 semantics, meaning that disk images with proper Verity data in place
9508 may be booted in containers with systemd-nspawn as well as on
9509 physical systems via the verity generator. Also note that the "mkosi"
9510 tool available at https://github.com/systemd/mkosi has been updated
9511 to generate Verity protected disk images following this scheme. In
9512 fact, it has been updated to generate disk images that optionally
9513 implement a complete UEFI SecureBoot trust chain, involving a signed
9514 kernel and initrd image that incorporates such a root hash as well as
9515 a Verity-enabled root partition.
9517 * The hardware database (hwdb) udev supports has been updated to carry
9518 accelerometer quirks.
9520 * All system services are now run with a fresh kernel keyring set up
9521 for them. The invocation ID is stored by default in it, thus
9522 providing a safe, non-overridable way to determine the invocation
9525 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
9526 options for bind mounting arbitrary paths in a service-specific
9527 way. When these options are used, arbitrary host or service files and
9528 directories may be mounted to arbitrary locations in the service's
9531 * Documentation has been added that lists all of systemd's low-level
9532 environment variables:
9534 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
9536 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
9537 whether a specific socket file descriptor matches a specified socket
9540 * systemd-firstboot has been updated to check for the
9541 systemd.firstboot= kernel command line option. It accepts a boolean
9542 and when set to false the first boot questions are skipped.
9544 * systemd-fstab-generator has been updated to check for the
9545 systemd.volatile= kernel command line option, which either takes an
9546 optional boolean parameter or the special value "state". If used the
9547 system may be booted in a "volatile" boot mode. Specifically,
9548 "systemd.volatile" is used, the root directory will be mounted as
9549 tmpfs, and only /usr is mounted from the actual root file system. If
9550 "systemd.volatile=state" is used, the root directory will be mounted
9551 as usual, but /var is mounted as tmpfs. This concept provides similar
9552 functionality as systemd-nspawn's --volatile= option, but provides it
9553 on physical boots. Use this option for implementing stateless
9554 systems, or testing systems with all state and/or configuration reset
9555 to the defaults. (Note though that many distributions are not
9556 prepared to boot up without a populated /etc or /var, though.)
9558 * systemd-gpt-auto-generator gained support for LUKS encrypted root
9559 partitions. Previously it only supported LUKS encrypted partitions
9560 for all other uses, except for the root partition itself.
9562 * Socket units gained support for listening on AF_VSOCK sockets for
9563 communication in virtualized QEMU environments.
9565 * The "configure" script gained a new option --with-fallback-hostname=
9566 for specifying the fallback hostname to use if none is configured in
9567 /etc/hostname. For example, by specifying
9568 --with-fallback-hostname=fedora it is possible to default to a
9569 hostname of "fedora" on pristine installations.
9571 * systemd-cgls gained support for a new --unit= switch for listing only
9572 the control groups of a specific unit. Similar --user-unit= has been
9573 added for listing only the control groups of a specific user unit.
9575 * systemd-mount gained a new --umount switch for unmounting a mount or
9576 automount point (and all mount/automount points below it).
9578 * systemd will now refuse full configuration reloads (via systemctl
9579 daemon-reload and related calls) unless at least 16MiB of free space
9580 are available in /run. This is a safety precaution in order to ensure
9581 that generators can safely operate after the reload completed.
9583 * A new unit file option RootImage= has been added, which has a similar
9584 effect as RootDirectory= but mounts the service's root directory from
9585 a disk image instead of plain directory. This logic reuses the same
9586 image dissection and mount logic that systemd-nspawn already uses,
9587 and hence supports any disk images systemd-nspawn supports, including
9588 those following the Discoverable Partition Specification, as well as
9589 Verity enabled images. This option enables systemd to run system
9590 services directly off disk images acting as resource bundles,
9591 possibly even including full integrity data.
9593 * A new MountAPIVFS= unit file option has been added, taking a boolean
9594 argument. If enabled /proc, /sys and /dev (collectively called the
9595 "API VFS") will be mounted for the service. This is only relevant if
9596 RootDirectory= or RootImage= is used for the service, as these mounts
9597 are of course in place in the host mount namespace anyway.
9599 * systemd-nspawn gained support for a new --pivot-root= switch. If
9600 specified the root directory within the container image is pivoted to
9601 the specified mount point, while the original root disk is moved to a
9602 different place. This option enables booting of ostree images
9603 directly with systemd-nspawn.
9605 * The systemd build scripts will no longer complain if the NTP server
9606 addresses are not changed from the defaults. Google now supports
9607 these NTP servers officially. We still recommend downstreams to
9608 properly register an NTP pool with the NTP pool project though.
9610 * coredumpctl gained a new "--reverse" option for printing the list
9611 of coredumps in reverse order.
9613 * coredumpctl will now show additional information about truncated and
9614 inaccessible coredumps, as well as coredumps that are still being
9615 processed. It also gained a new --quiet switch for suppressing
9616 additional informational message in its output.
9618 * coredumpctl gained support for only showing coredumps newer and/or
9619 older than specific timestamps, using the new --since= and --until=
9620 options, reminiscent of journalctl's options by the same name.
9622 * The systemd-coredump logic has been improved so that it may be reused
9623 to collect backtraces in non-compiled languages, for example in
9624 scripting languages such as Python.
9626 * machinectl will now show the UID shift of local containers, if user
9627 namespacing is enabled for them.
9629 * systemd will now optionally run "environment generator" binaries at
9630 configuration load time. They may be used to add environment
9631 variables to the environment block passed to services invoked. One
9632 user environment generator is shipped by default that sets up
9633 environment variables based on files dropped into /etc/environment.d
9634 and ~/.config/environment.d/.
9636 * systemd-resolved now includes the new, recently published 2017 DNSSEC
9639 * hostnamed has been updated to report a new chassis type of
9640 "convertible" to cover "foldable" laptops that can both act as a
9641 tablet and as a laptop, such as various Lenovo Yoga devices.
9643 Contributions from: Adrián López, Alexander Galanin, Alexander
9644 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
9645 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
9646 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
9647 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
9648 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
9649 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
9650 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
9651 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
9652 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
9653 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
9654 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
9655 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
9656 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
9657 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
9658 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
9659 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
9660 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
9661 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
9662 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
9663 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
9664 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
9665 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
9666 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
9667 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
9668 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
9669 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
9670 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
9673 — Berlin, 2017-03-01
9677 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
9678 RestrictAddressFamilies= enabled. These sandboxing options should
9679 generally be compatible with the various external udev call-out
9680 binaries we are aware of, however there may be exceptions, in
9681 particular when exotic languages for these call-outs are used. In
9682 this case, consider turning off these settings locally.
9684 * The new RemoveIPC= option can be used to remove IPC objects owned by
9685 the user or group of a service when that service exits.
9687 * The new ProtectKernelModules= option can be used to disable explicit
9688 load and unload operations of kernel modules by a service. In
9689 addition access to /usr/lib/modules is removed if this option is set.
9691 * ProtectSystem= option gained a new value "strict", which causes the
9692 whole file system tree with the exception of /dev, /proc, and /sys,
9693 to be remounted read-only for a service.
9695 * The new ProtectKernelTunables= option can be used to disable
9696 modification of configuration files in /sys and /proc by a service.
9697 Various directories and files are remounted read-only, so access is
9698 restricted even if the file permissions would allow it.
9700 * The new ProtectControlGroups= option can be used to disable write
9701 access by a service to /sys/fs/cgroup.
9703 * Various systemd services have been hardened with
9704 ProtectKernelTunables=yes, ProtectControlGroups=yes,
9705 RestrictAddressFamilies=.
9707 * Support for dynamically creating users for the lifetime of a service
9708 has been added. If DynamicUser=yes is specified, user and group IDs
9709 will be allocated from the range 61184…65519 for the lifetime of the
9710 service. They can be resolved using the new nss-systemd.so NSS
9711 module. The module must be enabled in /etc/nsswitch.conf. Services
9712 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
9713 any resources allocated by the service will be cleaned up when the
9714 service exits. They also have ProtectHome=read-only and
9715 ProtectSystem=strict enabled, so they are not able to make any
9716 permanent modifications to the system.
9718 * The nss-systemd module also always resolves root and nobody, making
9719 it possible to have no /etc/passwd or /etc/group files in minimal
9720 container or chroot environments.
9722 * Services may be started with their own user namespace using the new
9723 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
9724 under which the service is running are mapped. All other users are
9727 * Support for the cgroup namespace has been added to systemd-nspawn. If
9728 supported by kernel, the container system started by systemd-nspawn
9729 will have its own view of the cgroup hierarchy. This new behaviour
9730 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
9732 * The new MemorySwapMax= option can be used to limit the maximum swap
9733 usage under the unified cgroup hierarchy.
9735 * Support for the CPU controller in the unified cgroup hierarchy has
9736 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
9737 options. This controller requires out-of-tree patches for the kernel
9738 and the support is provisional.
9740 * Mount and automount units may now be created transiently
9741 (i.e. dynamically at runtime via the bus API, instead of requiring
9742 unit files in the file system).
9744 * systemd-mount is a new tool which may mount file systems – much like
9745 mount(8), optionally pulling in additional dependencies through
9746 transient .mount and .automount units. For example, this tool
9747 automatically runs fsck on a backing block device before mounting,
9748 and allows the automount logic to be used dynamically from the
9749 command line for establishing mount points. This tool is particularly
9750 useful when dealing with removable media, as it will ensure fsck is
9751 run – if necessary – before the first access and that the file system
9752 is quickly unmounted after each access by utilizing the automount
9753 logic. This maximizes the chance that the file system on the
9754 removable media stays in a clean state, and if it isn't in a clean
9755 state is fixed automatically.
9757 * LazyUnmount=yes option for mount units has been added to expose the
9758 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
9761 * /efi will be used as the mount point of the EFI boot partition, if
9762 the directory is present, and the mount point was not configured
9763 through other means (e.g. fstab). If /efi directory does not exist,
9764 /boot will be used as before. This makes it easier to automatically
9765 mount the EFI partition on systems where /boot is used for something
9768 * When operating on GPT disk images for containers, systemd-nspawn will
9769 now mount the ESP to /boot or /efi according to the same rules as PID
9770 1 running on a host. This allows tools like "bootctl" to operate
9771 correctly within such containers, in order to make container images
9772 bootable on physical systems.
9774 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
9776 * Two new user session targets have been added to support running
9777 graphical sessions under the systemd --user instance:
9778 graphical-session.target and graphical-session-pre.target. See
9779 systemd.special(7) for a description of how those targets should be
9782 * The vconsole initialization code has been significantly reworked to
9783 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
9784 support unicode keymaps. Font and keymap configuration will now be
9785 copied to all allocated virtual consoles.
9787 * FreeBSD's bhyve virtualization is now detected.
9789 * Information recorded in the journal for core dumps now includes the
9790 contents of /proc/mountinfo and the command line of the process at
9791 the top of the process hierarchy (which is usually the init process
9794 * systemd-journal-gatewayd learned the --directory= option to serve
9795 files from the specified location.
9797 * journalctl --root=… can be used to peruse the journal in the
9798 /var/log/ directories inside of a container tree. This is similar to
9799 the existing --machine= option, but does not require the container to
9802 * The hardware database has been extended to support
9803 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
9806 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
9807 specify the click rate for mice which include a horizontal wheel with
9808 a click rate that is different than the one for the vertical wheel.
9810 * systemd-run gained a new --wait option that makes service execution
9811 synchronous. (Specifically, the command will not return until the
9812 specified service binary exited.)
9814 * systemctl gained a new --wait option that causes the start command to
9815 wait until the units being started have terminated again.
9817 * A new journal output mode "short-full" has been added which displays
9818 timestamps with abbreviated English day names and adds a timezone
9819 suffix. Those timestamps include more information than the default
9820 "short" output mode, and can be passed directly to journalctl's
9821 --since= and --until= options.
9823 * /etc/resolv.conf will be bind-mounted into containers started by
9824 systemd-nspawn, if possible, so any changes to resolv.conf contents
9825 are automatically propagated to the container.
9827 * The number of instances for socket-activated services originating
9828 from a single IP address can be limited with
9829 MaxConnectionsPerSource=, extending the existing setting of
9832 * systemd-networkd gained support for vcan ("Virtual CAN") interface
9835 * .netdev and .network configuration can now be extended through
9838 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
9839 Segmentation Offload, Generic Receive Offload, Large Receive Offload
9840 can be enabled and disabled using the new UDPSegmentationOffload=,
9841 TCPSegmentationOffload=, GenericSegmentationOffload=,
9842 GenericReceiveOffload=, LargeReceiveOffload= options in the
9843 [Link] section of .link files.
9845 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
9846 Port VLAN ID can be configured for bridge devices using the new STP=,
9847 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
9848 section of .netdev files.
9850 * The route table to which routes received over DHCP or RA should be
9851 added can be configured with the new RouteTable= option in the [DHCP]
9852 and [IPv6AcceptRA] sections of .network files.
9854 * The Address Resolution Protocol can be disabled on links managed by
9855 systemd-networkd using the ARP=no setting in the [Link] section of
9858 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
9859 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
9860 encode information about the result and exit codes of the current
9861 service runtime cycle.
9863 * systemd-sysctl will now configure kernel parameters in the order
9864 they occur in the configuration files. This matches what sysctl
9865 has been traditionally doing.
9867 * kernel-install "plugins" that are executed to perform various
9868 tasks after a new kernel is added and before an old one is removed
9869 can now return a special value to terminate the procedure and
9870 prevent any later plugins from running.
9872 * Journald's SplitMode=login setting has been deprecated. It has been
9873 removed from documentation, and its use is discouraged. In a future
9874 release it will be completely removed, and made equivalent to current
9875 default of SplitMode=uid.
9877 * Storage=both option setting in /etc/systemd/coredump.conf has been
9878 removed. With fast LZ4 compression storing the core dump twice is not
9881 * The --share-system systemd-nspawn option has been replaced with an
9882 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
9883 this functionality is discouraged. In addition the variables
9884 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
9885 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
9886 individual namespaces.
9888 * "machinectl list" now shows the IP address of running containers in
9889 the output, as well as OS release information.
9891 * "loginctl list" now shows the TTY of each session in the output.
9893 * sd-bus gained new API calls sd_bus_track_set_recursive(),
9894 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
9895 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
9896 tracking objects in a "recursive" mode, where a single client can be
9897 counted multiple times, if it takes multiple references.
9899 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
9900 sd_bus_get_exit_on_disconnect(). They may be used to make a
9901 process using sd-bus automatically exit if the bus connection is
9904 * Bus clients of the service manager may now "pin" loaded units into
9905 memory, by taking an explicit reference on them. This is useful to
9906 ensure the client can retrieve runtime data about the service even
9907 after the service completed execution. Taking such a reference is
9908 available only for privileged clients and should be helpful to watch
9909 running services in a race-free manner, and in particular collect
9910 information about exit statuses and results.
9912 * The nss-resolve module has been changed to strictly return UNAVAIL
9913 when communication via D-Bus with resolved failed, and NOTFOUND when
9914 a lookup completed but was negative. This means it is now possible to
9915 neatly configure fallbacks using nsswitch.conf result checking
9916 expressions. Taking benefit of this, the new recommended
9917 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
9919 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
9921 * A new setting CtrlAltDelBurstAction= has been added to
9922 /etc/systemd/system.conf which may be used to configure the precise
9923 behaviour if the user on the console presses Ctrl-Alt-Del more often
9924 than 7 times in 2s. Previously this would unconditionally result in
9925 an expedited, immediate reboot. With this new setting the precise
9926 operation may be configured in more detail, and also turned off
9929 * In .netdev files two new settings RemoteChecksumTx= and
9930 RemoteChecksumRx= are now understood that permit configuring the
9931 remote checksumming logic for VXLAN networks.
9933 * The service manager learnt a new "invocation ID" concept for invoked
9934 services. Each runtime cycle of a service will get a new invocation
9935 ID (a 128-bit random UUID) assigned that identifies the current
9936 run of the service uniquely and globally. A new invocation ID
9937 is generated each time a service starts up. The journal will store
9938 the invocation ID of a service along with any logged messages, thus
9939 making the invocation ID useful for matching the online runtime of a
9940 service with the offline log data it generated in a safe way without
9941 relying on synchronized timestamps. In many ways this new service
9942 invocation ID concept is similar to the kernel's boot ID concept that
9943 uniquely and globally identifies the runtime of each boot. The
9944 invocation ID of a service is passed to the service itself via an
9945 environment variable ($INVOCATION_ID). A new bus call
9946 GetUnitByInvocationID() has been added that is similar to GetUnit()
9947 but instead of retrieving the bus path for a unit by its name
9948 retrieves it by its invocation ID. The returned path is valid only as
9949 long as the passed invocation ID is current.
9951 * systemd-resolved gained a new "DNSStubListener" setting in
9952 resolved.conf. It either takes a boolean value or the special values
9953 "udp" and "tcp", and configures whether to enable the stub DNS
9954 listener on 127.0.0.53:53.
9956 * IP addresses configured via networkd may now carry additional
9957 configuration settings supported by the kernel. New options include:
9958 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
9959 PrefixRoute=, AutoJoin=.
9961 * The PAM configuration fragment file for "user@.service" shipped with
9962 systemd (i.e. the --user instance of systemd) has been stripped to
9963 the minimum necessary to make the system boot. Previously, it
9964 contained Fedora-specific stanzas that did not apply to other
9965 distributions. It is expected that downstream distributions add
9966 additional configuration lines, matching their needs to this file,
9967 using it only as rough template of what systemd itself needs. Note
9968 that this reduced fragment does not even include an invocation of
9969 pam_limits which most distributions probably want to add, even though
9970 systemd itself does not need it. (There's also the new build time
9971 option --with-pamconfdir=no to disable installation of the PAM
9974 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
9975 capability is now also dropped from its set (in addition to
9976 CAP_SYS_MKNOD as before).
9978 * In service unit files it is now possible to connect a specific named
9979 file descriptor with stdin/stdout/stdout of an executed service. The
9980 name may be specified in matching .socket units using the
9981 FileDescriptorName= setting.
9983 * A number of journal settings may now be configured on the kernel
9984 command line. Specifically, the following options are now understood:
9985 systemd.journald.max_level_console=,
9986 systemd.journald.max_level_store=,
9987 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
9988 systemd.journald.max_level_wall=.
9990 * "systemctl is-enabled --full" will now show by which symlinks a unit
9991 file is enabled in the unit dependency tree.
9993 * Support for VeraCrypt encrypted partitions has been added to the
9994 "cryptsetup" logic and /etc/crypttab.
9996 * systemd-detect-virt gained support for a new --private-users switch
9997 that checks whether the invoking processes are running inside a user
9998 namespace. Similar, a new special value "private-users" for the
9999 existing ConditionVirtualization= setting has been added, permitting
10000 skipping of specific units in user namespace environments.
10002 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
10003 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
10004 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
10005 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
10006 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
10007 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
10008 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
10009 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
10010 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
10011 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
10012 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
10013 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
10014 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
10015 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
10016 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
10017 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
10018 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
10019 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
10020 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
10021 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
10022 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
10023 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
10024 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
10025 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
10026 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
10027 Jędrzejewski-Szmek, Zeal Jagannatha
10029 — Santa Fe, 2016-11-03
10033 * In service units the various ExecXYZ= settings have been extended
10034 with an additional special character as first argument of the
10035 assigned value: if the character '+' is used the specified command
10036 line it will be run with full privileges, regardless of User=,
10037 Group=, CapabilityBoundingSet= and similar options. The effect is
10038 similar to the existing PermissionsStartOnly= option, but allows
10039 configuration of this concept for each executed command line
10042 * Services may now alter the service watchdog timeout at runtime by
10043 sending a WATCHDOG_USEC= message via sd_notify().
10045 * MemoryLimit= and related unit settings now optionally take percentage
10046 specifications. The percentage is taken relative to the amount of
10047 physical memory in the system (or in case of containers, the assigned
10048 amount of memory). This allows scaling service resources neatly with
10049 the amount of RAM available on the system. Similarly, systemd-logind's
10050 RuntimeDirectorySize= option now also optionally takes percentage
10053 * In similar fashion TasksMax= takes percentage values now, too. The
10054 value is taken relative to the configured maximum number of processes
10055 on the system. The per-service task maximum has been changed to 15%
10056 using this functionality. (Effectively this is an increase of 512 →
10057 4915 for service units, given the kernel's default pid_max setting.)
10059 * Calendar time specifications in .timer units now understand a ".."
10060 syntax for time ranges. Example: "4..7:10" may now be used for
10061 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
10064 * The InaccessableDirectories=, ReadOnlyDirectories= and
10065 ReadWriteDirectories= unit file settings have been renamed to
10066 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
10067 applied to all kinds of file nodes, and not just directories, with
10068 the exception of symlinks. Specifically these settings may now be
10069 used on block and character device nodes, UNIX sockets and FIFOS as
10070 well as regular files. The old names of these settings remain
10071 available for compatibility.
10073 * systemd will now log about all service processes it kills forcibly
10074 (using SIGKILL) because they remained after the clean shutdown phase
10075 of the service completed. This should help identifying services that
10076 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
10077 systemd-logind's configuration a similar log message is generated for
10078 processes killed at the end of each session due to this setting.
10080 * systemd will now set the $JOURNAL_STREAM environment variable for all
10081 services whose stdout/stderr are connected to the Journal (which
10082 effectively means by default: all services). The variable contains
10083 the device and inode number of the file descriptor used for
10084 stdout/stderr. This may be used by invoked programs to detect whether
10085 their stdout/stderr is connected to the Journal, in which case they
10086 can switch over to direct Journal communication, thus being able to
10087 pass extended, structured metadata along with their log messages. As
10088 one example, this is now used by glib's logging primitives.
10090 * When using systemd's default tmp.mount unit for /tmp, the mount point
10091 will now be established with the "nosuid" and "nodev" options. This
10092 avoids privilege escalation attacks that put traps and exploits into
10093 /tmp. However, this might cause problems if you e.g. put container
10094 images or overlays into /tmp; if you need this, override tmp.mount's
10095 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
10098 * systemd now supports the "memory" cgroup controller also on
10101 * The systemd-cgtop tool now optionally takes a control group path as
10102 command line argument. If specified, the control group list shown is
10103 limited to subgroups of that group.
10105 * The SystemCallFilter= unit file setting gained support for
10106 pre-defined, named system call filter sets. For example
10107 SystemCallFilter=@clock is now an effective way to make all clock
10108 changing-related system calls unavailable to a service. A number of
10109 similar pre-defined groups are defined. Writing system call filters
10110 for system services is simplified substantially with this new
10111 concept. Accordingly, all of systemd's own, long-running services now
10112 enable system call filtering based on this, by default.
10114 * A new service setting MemoryDenyWriteExecute= has been added, taking
10115 a boolean value. If turned on, a service may no longer create memory
10116 mappings that are writable and executable at the same time. This
10117 enhances security for services where this is enabled as it becomes
10118 harder to dynamically write and then execute memory in exploited
10119 service processes. This option has been enabled for all of systemd's
10120 own long-running services.
10122 * A new RestrictRealtime= service setting has been added, taking a
10123 boolean argument. If set the service's processes may no longer
10124 acquire realtime scheduling. This improves security as realtime
10125 scheduling may otherwise be used to easily freeze the system.
10127 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
10128 value. This may be used for requesting that the system manager inside
10129 of the container reports start-up completion to nspawn which then
10130 propagates this notification further to the service manager
10131 supervising nspawn itself. A related option NotifyReady= in .nspawn
10132 files has been added too. This functionality allows ordering of the
10133 start-up of multiple containers using the usual systemd ordering
10136 * machinectl gained a new command "stop" that is an alias for
10139 * systemd-resolved gained support for contacting DNS servers on
10140 link-local IPv6 addresses.
10142 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
10143 its caches. A method call for requesting the same operation has been
10144 added to the bus API too, and is made available via "systemd-resolve
10147 * systemd-resolve gained a new --status switch. If passed a brief
10148 summary of the used DNS configuration with per-interface information
10151 * resolved.conf gained a new Cache= boolean option, defaulting to
10152 on. If turned off local DNS caching is disabled. This comes with a
10153 performance penalty in particular when DNSSEC is enabled. Note that
10154 resolved disables its internal caching implicitly anyway, when the
10155 configured DNS server is on a host-local IP address such as ::1 or
10156 127.0.0.1, thus automatically avoiding double local caching.
10158 * systemd-resolved now listens on the local IP address 127.0.0.53:53
10159 for DNS requests. This improves compatibility with local programs
10160 that do not use the libc NSS or systemd-resolved's bus APIs for name
10161 resolution. This minimal DNS service is only available to local
10162 programs and does not implement the full DNS protocol, but enough to
10163 cover local DNS clients. A new, static resolv.conf file, listing just
10164 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
10165 now recommended to make /etc/resolv.conf a symlink to this file in
10166 order to route all DNS lookups to systemd-resolved, regardless if
10167 done via NSS, the bus API or raw DNS packets. Note that this local
10168 DNS service is not as fully featured as the libc NSS or
10169 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
10170 used to deliver link-local address information (as this implies
10171 sending a local interface index along), LLMNR/mDNS support via this
10172 interface is severely restricted. It is thus strongly recommended for
10173 all applications to use the libc NSS API or native systemd-resolved
10176 * systemd-networkd's bridge support learned a new setting
10177 VLANFiltering= for controlling VLAN filtering. Moreover a new section
10178 in .network files has been added for configuring VLAN bridging in
10179 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
10181 * systemd-networkd's IPv6 Router Advertisement code now makes use of
10182 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
10183 now be acquired without relying on DHCPv6. Two new options
10184 UseDomains= and UseDNS= have been added to configure this behaviour.
10186 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
10187 renamed IPv6AcceptRA=, without altering its behaviour. The old
10188 setting name remains available for compatibility reasons.
10190 * The systemd-networkd VTI/VTI6 tunneling support gained new options
10191 Key=, InputKey= and OutputKey=.
10193 * systemd-networkd gained support for VRF ("Virtual Routing Function")
10194 interface configuration.
10196 * "systemctl edit" may now be used to create new unit files by
10197 specifying the --force switch.
10199 * sd-event gained a new function sd_event_get_iteration() for
10200 requesting the current iteration counter of the event loop. It starts
10201 at zero and is increased by one with each event loop iteration.
10203 * A new rpm macro %systemd_ordering is provided by the macros.systemd
10204 file. It can be used in lieu of %systemd_requires in packages which
10205 don't use any systemd functionality and are intended to be installed
10206 in minimal containers without systemd present. This macro provides
10207 ordering dependencies to ensure that if the package is installed in
10208 the same rpm transaction as systemd, systemd will be installed before
10209 the scriptlets for the package are executed, allowing unit presets
10212 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
10213 been added to simplify packaging of generators.
10215 * The os-release file gained VERSION_CODENAME field for the
10216 distribution nickname (e.g. VERSION_CODENAME=woody).
10218 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
10219 can be set to disable parsing of metadata and the creation
10220 of persistent symlinks for that device.
10222 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
10223 to make them available to logged-in users has been reverted.
10225 * Much of the common code of the various systemd components is now
10226 built into an internal shared library libsystemd-shared-231.so
10227 (incorporating the systemd version number in the name, to be updated
10228 with future releases) that the components link to. This should
10229 decrease systemd footprint both in memory during runtime and on
10230 disk. Note that the shared library is not for public use, and is
10231 neither API nor ABI stable, but is likely to change with every new
10232 released update. Packagers need to make sure that binaries
10233 linking to libsystemd-shared.so are updated in step with the
10236 * Configuration for "mkosi" is now part of the systemd
10237 repository. mkosi is a tool to easily build legacy-free OS images,
10238 and is available on github: https://github.com/systemd/mkosi. If
10239 "mkosi" is invoked in the build tree a new raw OS image is generated
10240 incorporating the systemd sources currently being worked on and a
10241 clean, fresh distribution installation. The generated OS image may be
10242 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
10243 UEFI PC. This functionality is particularly useful to easily test
10244 local changes made to systemd in a pristine, defined environment. See
10245 doc/HACKING for details.
10247 * configure learned the --with-support-url= option to specify the
10248 distribution's bugtracker.
10250 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
10251 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
10252 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
10253 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
10254 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
10255 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
10256 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
10257 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
10258 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
10259 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
10260 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
10261 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
10262 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
10263 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
10264 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
10265 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
10266 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
10267 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
10268 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
10270 — Berlin, 2016-07-25
10274 * DNSSEC is now turned on by default in systemd-resolved (in
10275 "allow-downgrade" mode), but may be turned off during compile time by
10276 passing "--with-default-dnssec=no" to "configure" (and of course,
10277 during runtime with DNSSEC= in resolved.conf). We recommend
10278 downstreams to leave this on at least during development cycles and
10279 report any issues with the DNSSEC logic upstream. We are very
10280 interested in collecting feedback about the DNSSEC validator and its
10281 limitations in the wild. Note however, that DNSSEC support is
10282 probably nothing downstreams should turn on in stable distros just
10283 yet, as it might create incompatibilities with a few DNS servers and
10284 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
10285 automatically whenever we detect such incompatible setups, but there
10286 might be systems we do not cover yet. Hence: please help us testing
10287 the DNSSEC code, leave this on where you can, report back, but then
10288 again don't consider turning this on in your stable, LTS or
10289 production release just yet. (Note that you have to enable
10290 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
10291 and its DNSSEC mode for hostname resolution from local
10294 * systemd-resolve conveniently resolves DANE records with the --tlsa
10295 option and OPENPGPKEY records with the --openpgp option. It also
10296 supports dumping raw DNS record data via the new --raw= switch.
10298 * systemd-logind will now by default terminate user processes that are
10299 part of the user session scope unit (session-XX.scope) when the user
10300 logs out. This behavior is controlled by the KillUserProcesses=
10301 setting in logind.conf, and the previous default of "no" is now
10302 changed to "yes". This means that user sessions will be properly
10303 cleaned up after, but additional steps are necessary to allow
10304 intentionally long-running processes to survive logout.
10306 While the user is logged in at least once, user@.service is running,
10307 and any service that should survive the end of any individual login
10308 session can be started at a user service or scope using systemd-run.
10309 systemd-run(1) man page has been extended with an example which shows
10310 how to run screen in a scope unit underneath user@.service. The same
10311 command works for tmux.
10313 After the user logs out of all sessions, user@.service will be
10314 terminated too, by default, unless the user has "lingering" enabled.
10315 To effectively allow users to run long-term tasks even if they are
10316 logged out, lingering must be enabled for them. See loginctl(1) for
10317 details. The default polkit policy was modified to allow users to
10318 set lingering for themselves without authentication.
10320 Previous defaults can be restored at compile time by the
10321 --without-kill-user-processes option to "configure".
10323 * systemd-logind gained new configuration settings SessionsMax= and
10324 InhibitorsMax=, both with a default of 8192. It will not register new
10325 user sessions or inhibitors above this limit.
10327 * systemd-logind will now reload configuration on SIGHUP.
10329 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
10330 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
10331 enable. Also, support for the "io" cgroup controller in the unified
10332 hierarchy has been added, so that the "memory", "pids" and "io" are
10333 now the controllers that are supported on the unified hierarchy.
10335 WARNING: it is not possible to use previous systemd versions with
10336 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
10337 is necessary to also update systemd in the initramfs if using the
10338 unified hierarchy. An updated SELinux policy is also required.
10340 * LLDP support has been extended, and both passive (receive-only) and
10341 active (sender) modes are supported. Passive mode ("routers-only") is
10342 enabled by default in systemd-networkd. Active LLDP mode is enabled
10343 by default for containers on the internal network. The "networkctl
10344 lldp" command may be used to list information gathered. "networkctl
10345 status" will also show basic LLDP information on connected peers now.
10347 * The IAID and DUID unique identifier sent in DHCP requests may now be
10348 configured for the system and each .network file managed by
10349 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
10351 * systemd-networkd gained support for configuring proxy ARP support for
10352 each interface, via the ProxyArp= setting in .network files. It also
10353 gained support for configuring the multicast querier feature of
10354 bridge devices, via the new MulticastQuerier= setting in .netdev
10355 files. Similarly, snooping on the IGMP traffic can be controlled
10356 via the new setting MulticastSnooping=.
10358 A new setting PreferredLifetime= has been added for addresses
10359 configured in .network file to configure the lifetime intended for an
10362 The systemd-networkd DHCP server gained the option EmitRouter=, which
10363 defaults to yes, to configure whether the DHCP Option 3 (Router)
10366 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
10367 systemd-socket-activate and installed into /usr/bin. It is now fully
10370 * systemd-journald now uses separate threads to flush changes to disk
10371 when closing journal files, thus reducing impact of slow disk I/O on
10372 logging performance.
10374 * The sd-journal API gained two new calls
10375 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
10376 can be used to open journal files using file descriptors instead of
10377 file or directory paths. sd_journal_open_container() has been
10378 deprecated, sd_journal_open_directory_fd() should be used instead
10379 with the flag SD_JOURNAL_OS_ROOT.
10381 * journalctl learned a new output mode "-o short-unix" that outputs log
10382 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
10383 UTC). It also gained support for a new --no-hostname setting to
10384 suppress the hostname column in the family of "short" output modes.
10386 * systemd-ask-password now optionally skips printing of the password to
10387 stdout with --no-output which can be useful in scripts.
10389 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
10390 (devices tagged with ID_MAKER_TOOL) are now tagged with
10391 "uaccess" and are available to logged in users.
10393 * The DeviceAllow= unit setting now supports specifiers (with "%").
10395 * "systemctl show" gained a new --value switch, which allows print a
10396 only the contents of a specific unit property, without also printing
10397 the property's name. Similar support was added to "show*" verbs
10398 of loginctl and machinectl that output "key=value" lists.
10400 * A new unit type "generated" was added for files dynamically generated
10401 by generator tools. Similarly, a new unit type "transient" is used
10402 for unit files created using the runtime API. "systemctl enable" will
10403 refuse to operate on such files.
10405 * A new command "systemctl revert" has been added that may be used to
10406 revert to the vendor version of a unit file, in case local changes
10407 have been made by adding drop-ins or overriding the unit file.
10409 * "machinectl clean" gained a new verb to automatically remove all or
10410 just hidden container images.
10412 * systemd-tmpfiles gained support for a new line type "e" for emptying
10413 directories, if they exist, without creating them if they don't.
10415 * systemd-nspawn gained support for automatically patching the UID/GIDs
10416 of the owners and the ACLs of all files and directories in a
10417 container tree to match the UID/GID user namespacing range selected
10418 for the container invocation. This mode is enabled via the new
10419 --private-users-chown switch. It also gained support for
10420 automatically choosing a free, previously unused UID/GID range when
10421 starting a container, via the new --private-users=pick setting (which
10422 implies --private-users-chown). Together, these options for the first
10423 time make user namespacing for nspawn containers fully automatic and
10424 thus deployable. The systemd-nspawn@.service template unit file has
10425 been changed to use this functionality by default.
10427 * systemd-nspawn gained a new --network-zone= switch, that allows
10428 creating ad-hoc virtual Ethernet links between multiple containers,
10429 that only exist as long as at least one container referencing them is
10430 running. This allows easy connecting of multiple containers with a
10431 common link that implements an Ethernet broadcast domain. Each of
10432 these network "zones" may be named relatively freely by the user, and
10433 may be referenced by any number of containers, but each container may
10434 only reference one of these "zones". On the lower level, this is
10435 implemented by an automatically managed bridge network interface for
10436 each zone, that is created when the first container referencing its
10437 zone is created and removed when the last one referencing its zone
10440 * The default start timeout may now be configured on the kernel command
10441 line via systemd.default_timeout_start_sec=. It was already
10442 configurable via the DefaultTimeoutStartSec= option in
10443 /etc/systemd/system.conf.
10445 * Socket units gained a new TriggerLimitIntervalSec= and
10446 TriggerLimitBurst= setting to configure a limit on the activation
10447 rate of the socket unit.
10449 * The LimitNICE= setting now optionally takes normal UNIX nice values
10450 in addition to the raw integer limit value. If the specified
10451 parameter is prefixed with "+" or "-" and is in the range -20…19 the
10452 value is understood as UNIX nice value. If not prefixed like this it
10453 is understood as raw RLIMIT_NICE limit.
10455 * Note that the effect of the PrivateDevices= unit file setting changed
10456 slightly with this release: the per-device /dev file system will be
10457 mounted read-only from this version on, and will have "noexec"
10458 set. This (minor) change of behavior might cause some (exceptional)
10459 legacy software to break, when PrivateDevices=yes is set for its
10460 service. Please leave PrivateDevices= off if you run into problems
10463 * systemd-bootchart has been split out to a separate repository:
10464 https://github.com/systemd/systemd-bootchart
10466 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
10467 merged into the kernel in its current form.
10469 * The compatibility libraries libsystemd-daemon.so,
10470 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
10471 which have been deprecated since systemd-209 have been removed along
10472 with the corresponding pkg-config files. All symbols provided by
10473 those libraries are provided by libsystemd.so.
10475 * The Capabilities= unit file setting has been removed (it is ignored
10476 for backwards compatibility). AmbientCapabilities= and
10477 CapabilityBoundingSet= should be used instead.
10479 * A new special target has been added, initrd-root-device.target,
10480 which creates a synchronization point for dependencies of the root
10481 device in early userspace. Initramfs builders must ensure that this
10482 target is now included in early userspace.
10484 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
10485 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
10486 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
10487 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
10488 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
10489 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
10490 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
10491 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
10492 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
10493 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
10494 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
10495 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
10496 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
10497 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
10498 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
10499 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
10500 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
10501 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
10502 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
10503 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
10504 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
10505 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
10506 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
10507 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
10508 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
10511 — Fairfax, 2016-05-21
10515 * The systemd-resolved DNS resolver service has gained a substantial
10516 set of new features, most prominently it may now act as a DNSSEC
10517 validating stub resolver. DNSSEC mode is currently turned off by
10518 default, but is expected to be turned on by default in one of the
10519 next releases. For now, we invite everybody to test the DNSSEC logic
10520 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
10521 service also gained a full set of D-Bus interfaces, including calls
10522 to configure DNS and DNSSEC settings per link (for use by external
10523 network management software). systemd-resolved and systemd-networkd
10524 now distinguish between "search" and "routing" domains. The former
10525 are used to qualify single-label names, the latter are used purely
10526 for routing lookups within certain domains to specific links.
10527 resolved now also synthesizes RRs for all entries from /etc/hosts.
10529 * The systemd-resolve tool (which is a client utility for
10530 systemd-resolved) has been improved considerably and is now fully
10531 supported and documented. Hence it has moved from /usr/lib/systemd to
10534 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
10537 * The coredump collection logic has been reworked: when a coredump is
10538 collected it is now written to disk, compressed and processed
10539 (including stacktrace extraction) from a new instantiated service
10540 systemd-coredump@.service, instead of directly from the
10541 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
10542 processing large coredumps can take up a substantial amount of
10543 resources and time, and this previously happened entirely outside of
10544 systemd's service supervision. With the new logic the core_pattern
10545 hook only does minimal metadata collection before passing off control
10546 to the new instantiated service, which is configured with a time
10547 limit, a nice level and other settings to minimize negative impact on
10548 the rest of the system. Also note that the new logic will honour the
10549 RLIMIT_CORE setting of the crashed process, which now allows users
10550 and processes to turn off coredumping for their processes by setting
10553 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
10554 and all forked processes by default. Previously, PID 1 would leave
10555 the setting at "0" for all processes, as set by the kernel. Note that
10556 the resource limit traditionally has no effect on the generated
10557 coredumps on the system if the /proc/sys/kernel/core_pattern hook
10558 logic is used. Since the limit is now honoured (see above) its
10559 default has been changed so that the coredumping logic is enabled by
10560 default for all processes, while allowing specific opt-out.
10562 * When the stacktrace is extracted from processes of system users, this
10563 is now done as "systemd-coredump" user, in order to sandbox this
10564 potentially security sensitive parsing operation. (Note that when
10565 processing coredumps of normal users this is done under the user ID
10566 of process that crashed, as before.) Packagers should take notice
10567 that it is now necessary to create the "systemd-coredump" system user
10568 and group at package installation time.
10570 * The systemd-activate socket activation testing tool gained support
10571 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
10572 and --seqpacket switches. It also has been extended to support both
10573 new-style and inetd-style file descriptor passing. Use the new
10574 --inetd switch to request inetd-style file descriptor passing.
10576 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
10577 variable, which takes a boolean value. If set to false, ANSI color
10578 output is disabled in the tools even when run on a terminal that
10581 * The VXLAN support in networkd now supports two new settings
10582 DestinationPort= and PortRange=.
10584 * A new systemd.machine_id= kernel command line switch has been added,
10585 that may be used to set the machine ID in /etc/machine-id if it is
10586 not initialized yet. This command line option has no effect if the
10587 file is already initialized.
10589 * systemd-nspawn gained a new --as-pid2 switch that invokes any
10590 specified command line as PID 2 rather than PID 1 in the
10591 container. In this mode PID 1 is a minimal stub init process that
10592 implements the special POSIX and Linux semantics of PID 1 regarding
10593 signal and child process management. Note that this stub init process
10594 is implemented in nspawn itself and requires no support from the
10595 container image. This new logic is useful to support running
10596 arbitrary commands in the container, as normal processes are
10597 generally not prepared to run as PID 1.
10599 * systemd-nspawn gained a new --chdir= switch for setting the current
10600 working directory for the process started in the container.
10602 * "journalctl /dev/sda" will now output all kernel log messages for
10603 specified device from the current boot, in addition to all devices
10604 that are parents of it. This should make log output about devices
10605 pretty useful, as long as kernel drivers attach enough metadata to
10606 the log messages. (The usual SATA drivers do.)
10608 * The sd-journal API gained two new calls
10609 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
10610 that report whether log data from /run or /var has been found.
10612 * journalctl gained a new switch "--fields" that prints all journal
10613 record field names currently in use in the journal. This is backed
10614 by two new sd-journal API calls sd_journal_enumerate_fields() and
10615 sd_journal_restart_fields().
10617 * Most configurable timeouts in systemd now expect an argument of
10618 "infinity" to turn them off, instead of "0" as before. The semantics
10619 from now on is that a timeout of "0" means "now", and "infinity"
10620 means "never". To maintain backwards compatibility, "0" continues to
10621 turn off previously existing timeout settings.
10623 * "systemctl reload-or-try-restart" has been renamed to "systemctl
10624 try-reload-or-restart" to clarify what it actually does: the "try"
10625 logic applies to both reloading and restarting, not just restarting.
10626 The old name continues to be accepted for compatibility.
10628 * On boot-up, when PID 1 detects that the system clock is behind the
10629 release date of the systemd version in use, the clock is now set
10630 to the latter. Previously, this was already done in timesyncd, in order
10631 to avoid running with clocks set to the various clock epochs such as
10632 1902, 1938 or 1970. With this change the logic is now done in PID 1
10633 in addition to timesyncd during early boot-up, so that it is enforced
10634 before the first process is spawned by systemd. Note that the logic
10635 in timesyncd remains, as it is more comprehensive and ensures
10636 clock monotonicity by maintaining a persistent timestamp file in
10637 /var. Since /var is generally not available in earliest boot or the
10638 initrd, this part of the logic remains in timesyncd, and is not done
10641 * Support for tweaking details in net_cls.class_id through the
10642 NetClass= configuration directive has been removed, as the kernel
10643 people have decided to deprecate that controller in cgroup v2.
10644 Userspace tools such as nftables are moving over to setting rules
10645 that are specific to the full cgroup path of a task, which obsoletes
10646 these controllers anyway. The NetClass= directive is kept around for
10647 legacy compatibility reasons. For a more in-depth description of the
10648 kernel change, please refer to the respective upstream commit:
10650 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
10652 * A new service setting RuntimeMaxSec= has been added that may be used
10653 to specify a maximum runtime for a service. If the timeout is hit, the
10654 service is terminated and put into a failure state.
10656 * A new service setting AmbientCapabilities= has been added. It allows
10657 configuration of additional Linux process capabilities that are
10658 passed to the activated processes. This is only available on very
10661 * The process resource limit settings in service units may now be used
10662 to configure hard and soft limits individually.
10664 * The various libsystemd APIs such as sd-bus or sd-event now publicly
10665 expose support for gcc's __attribute__((cleanup())) C extension.
10666 Specifically, for many object destructor functions alternative
10667 versions have been added that have names suffixed with "p" and take a
10668 pointer to a pointer to the object to destroy, instead of just a
10669 pointer to the object itself. This is useful because these destructor
10670 functions may be used directly as parameters to the cleanup
10671 construct. Internally, systemd has been a heavy user of this GCC
10672 extension for a long time, and with this change similar support is
10673 now available to consumers of the library outside of systemd. Note
10674 that by using this extension in your sources compatibility with old
10675 and strictly ANSI compatible C compilers is lost. However, all gcc or
10676 LLVM versions of recent years support this extension.
10678 * Timer units gained support for a new setting RandomizedDelaySec= that
10679 allows configuring some additional randomized delay to the configured
10680 time. This is useful to spread out timer events to avoid load peaks in
10681 clusters or larger setups.
10683 * Calendar time specifications now support sub-second accuracy.
10685 * Socket units now support listening on SCTP and UDP-lite protocol
10688 * The sd-event API now comes with a full set of man pages.
10690 * Older versions of systemd contained experimental support for
10691 compressing journal files and coredumps with the LZ4 compressor that
10692 was not compatible with the lz4 binary (due to API limitations of the
10693 lz4 library). This support has been removed; only support for files
10694 compatible with the lz4 binary remains. This LZ4 logic is now
10695 officially supported and no longer considered experimental.
10697 * The dkr image import logic has been removed again from importd. dkr's
10698 micro-services focus doesn't fit into the machine image focus of
10699 importd, and quickly got out of date with the upstream dkr API.
10701 * Creation of the /run/lock/lockdev/ directory was dropped from
10702 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
10703 been available for many years. If you still need this, you need to
10704 create your own tmpfiles.d config file with:
10706 d /run/lock/lockdev 0775 root lock -
10708 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
10709 and RebootArgument= have been moved from the [Service] section of
10710 unit files to [Unit], and they are now supported on all unit types,
10711 not just service units. Of course, systemd will continue to
10712 understand these settings also at the old location, in order to
10713 maintain compatibility.
10715 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
10716 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
10717 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
10718 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
10719 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
10720 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
10721 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
10722 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
10723 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
10724 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
10725 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
10726 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
10727 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
10728 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
10729 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
10730 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
10731 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
10732 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
10733 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
10735 — Berlin, 2016-02-11
10739 * A number of properties previously only settable in unit
10740 files are now also available as properties to set when
10741 creating transient units programmatically via the bus, as it
10742 is exposed with systemd-run's --property=
10743 setting. Specifically, these are: SyslogIdentifier=,
10744 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
10745 EnvironmentFile=, ReadWriteDirectories=,
10746 ReadOnlyDirectories=, InaccessibleDirectories=,
10747 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
10749 * When creating transient services via the bus API it is now
10750 possible to pass in a set of file descriptors to use as
10751 STDIN/STDOUT/STDERR for the invoked process.
10753 * Slice units may now be created transiently via the bus APIs,
10754 similar to the way service and scope units may already be
10755 created transiently.
10757 * Wherever systemd expects a calendar timestamp specification
10758 (like in journalctl's --since= and --until= switches) UTC
10759 timestamps are now supported. Timestamps suffixed with "UTC"
10760 are now considered to be in Universal Time Coordinated
10761 instead of the local timezone. Also, timestamps may now
10762 optionally be specified with sub-second accuracy. Both of
10763 these additions also apply to recurring calendar event
10764 specification, such as OnCalendar= in timer units.
10766 * journalctl gained a new "--sync" switch that asks the
10767 journal daemon to write all so far unwritten log messages to
10768 disk and sync the files, before returning.
10770 * systemd-tmpfiles learned two new line types "q" and "Q" that
10771 operate like "v", but also set up a basic btrfs quota
10772 hierarchy when used on a btrfs file system with quota
10775 * tmpfiles' "v", "q" and "Q" will now create a plain directory
10776 instead of a subvolume (even on a btrfs file system) if the
10777 root directory is a plain directory, and not a
10778 subvolume. This should simplify things with certain chroot()
10779 environments which are not aware of the concept of btrfs
10782 * systemd-detect-virt gained a new --chroot switch to detect
10783 whether execution takes place in a chroot() environment.
10785 * CPUAffinity= now takes CPU index ranges in addition to
10786 individual indexes.
10788 * The various memory-related resource limit settings (such as
10789 LimitAS=) now understand the usual K, M, G, … suffixes to
10790 the base of 1024 (IEC). Similar, the time-related resource
10791 limit settings understand the usual min, h, day, … suffixes
10794 * There's a new system.conf setting DefaultTasksMax= to
10795 control the default TasksMax= setting for services and
10796 scopes running on the system. (TasksMax= is the primary
10797 setting that exposes the "pids" cgroup controller on systemd
10798 and was introduced in the previous systemd release.) The
10799 setting now defaults to 512, which means services that are
10800 not explicitly configured otherwise will only be able to
10801 create 512 processes or threads at maximum, from this
10802 version on. Note that this means that thread- or
10803 process-heavy services might need to be reconfigured to set
10804 TasksMax= to a higher value. It is sufficient to set
10805 TasksMax= in these specific unit files to a higher value, or
10806 even "infinity". Similar, there's now a logind.conf setting
10807 UserTasksMax= that defaults to 4096 and limits the total
10808 number of processes or tasks each user may own
10809 concurrently. nspawn containers also have the TasksMax=
10810 value set by default now, to 8192. Note that all of this
10811 only has an effect if the "pids" cgroup controller is
10812 enabled in the kernel. The general benefit of these changes
10813 should be a more robust and safer system, that provides a
10814 certain amount of per-service fork() bomb protection.
10816 * systemd-nspawn gained the new --network-veth-extra= switch
10817 to define additional and arbitrarily-named virtual Ethernet
10818 links between the host and the container.
10820 * A new service execution setting PassEnvironment= has been
10821 added that allows importing select environment variables
10822 from PID1's environment block into the environment block of
10825 * Timer units gained support for a new RemainAfterElapse=
10826 setting which takes a boolean argument. It defaults to on,
10827 exposing behaviour unchanged to previous releases. If set to
10828 off, timer units are unloaded after they elapsed if they
10829 cannot elapse again. This is particularly useful for
10830 transient timer units, which shall not stay around longer
10831 than until they first elapse.
10833 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
10834 default now (the kernel default is 16). This is beneficial
10835 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
10836 allows substantially larger numbers of queued
10837 datagrams. This should increase the capability of systemd to
10838 parallelize boot-up, as logging and sd_notify() are unlikely
10839 to stall execution anymore. If you need to change the value
10840 from the new defaults, use the usual sysctl.d/ snippets.
10842 * The compression framing format used by the journal or
10843 coredump processing has changed to be in line with what the
10844 official LZ4 tools generate. LZ4 compression support in
10845 systemd was considered unsupported previously, as the format
10846 was not compatible with the normal tools. With this release
10847 this has changed now, and it is hence safe for downstream
10848 distributions to turn it on. While not compressing as well
10849 as the XZ, LZ4 is substantially faster, which makes
10850 it a good default choice for the compression logic in the
10851 journal and in coredump handling.
10853 * Any reference to /etc/mtab has been dropped from
10854 systemd. The file has been obsolete since a while, but
10855 systemd refused to work on systems where it was incorrectly
10856 set up (it should be a symlink or non-existent). Please make
10857 sure to update to util-linux 2.27.1 or newer in conjunction
10858 with this systemd release, which also drops any reference to
10859 /etc/mtab. If you maintain a distribution make sure that no
10860 software you package still references it, as this is a
10861 likely source of bugs. There's also a glibc bug pending,
10862 asking for removal of any reference to this obsolete file:
10864 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
10866 Note that only util-linux versions built with
10867 --enable-libmount-force-mountinfo are supported.
10869 * Support for the ".snapshot" unit type has been removed. This
10870 feature turned out to be little useful and little used, and
10871 has now been removed from the core and from systemctl.
10873 * The dependency types RequiresOverridable= and
10874 RequisiteOverridable= have been removed from systemd. They
10875 have been used only very sparingly to our knowledge and
10876 other options that provide a similar effect (such as
10877 systemctl --mode=ignore-dependencies) are much more useful
10878 and commonly used. Moreover, they were only half-way
10879 implemented as the option to control behaviour regarding
10880 these dependencies was never added to systemctl. By removing
10881 these dependency types the execution engine becomes a bit
10882 simpler. Unit files that use these dependencies should be
10883 changed to use the non-Overridable dependency types
10884 instead. In fact, when parsing unit files with these
10885 options, that's what systemd will automatically convert them
10886 too, but it will also warn, asking users to fix the unit
10887 files accordingly. Removal of these dependency types should
10888 only affect a negligible number of unit files in the wild.
10890 * Behaviour of networkd's IPForward= option changed
10891 (again). It will no longer maintain a per-interface setting,
10892 but propagate one way from interfaces where this is enabled
10893 to the global kernel setting. The global setting will be
10894 enabled when requested by a network that is set up, but
10895 never be disabled again. This change was made to make sure
10896 IPv4 and IPv6 behaviour regarding packet forwarding is
10897 similar (as the Linux IPv6 stack does not support
10898 per-interface control of this setting) and to minimize
10901 * In unit files the behaviour of %u, %U, %h, %s has
10902 changed. These specifiers will now unconditionally resolve
10903 to the various user database fields of the user that the
10904 systemd instance is running as, instead of the user
10905 configured in the specific unit via User=. Note that this
10906 effectively doesn't change much, as resolving of these
10907 specifiers was already turned off in the --system instance
10908 of systemd, as we cannot do NSS lookups from PID 1. In the
10909 --user instance of systemd these specifiers where correctly
10910 resolved, but hardly made any sense, since the user instance
10911 lacks privileges to do user switches anyway, and User= is
10912 hence useless. Moreover, even in the --user instance of
10913 systemd behaviour was awkward as it would only take settings
10914 from User= assignment placed before the specifier into
10915 account. In order to unify and simplify the logic around
10916 this the specifiers will now always resolve to the
10917 credentials of the user invoking the manager (which in case
10918 of PID 1 is the root user).
10920 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
10921 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
10922 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
10923 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
10924 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
10925 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
10926 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
10927 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
10928 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
10929 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
10930 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
10931 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
10932 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
10933 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
10936 — Berlin, 2015-11-18
10940 * systemd now depends on util-linux v2.27. More specifically,
10941 the newly added mount monitor feature in libmount now
10942 replaces systemd's former own implementation.
10944 * libmount mandates /etc/mtab not to be regular file, and
10945 systemd now enforces this condition at early boot.
10946 /etc/mtab has been deprecated and warned about for a very
10947 long time, so systems running systemd should already have
10948 stopped having this file around as anything else than a
10949 symlink to /proc/self/mounts.
10951 * Support for the "pids" cgroup controller has been added. It
10952 allows accounting the number of tasks in a cgroup and
10953 enforcing limits on it. This adds two new setting
10954 TasksAccounting= and TasksMax= to each unit, as well as a
10955 global option DefaultTasksAccounting=.
10957 * Support for the "net_cls" cgroup controller has been added.
10958 It allows assigning a net class ID to each task in the
10959 cgroup, which can then be used in firewall rules and traffic
10960 shaping configurations. Note that the kernel netfilter net
10961 class code does not currently work reliably for ingress
10962 packets on unestablished sockets.
10964 This adds a new config directive called NetClass= to CGroup
10965 enabled units. Allowed values are positive numbers for fixed
10966 assignments and "auto" for picking a free value
10969 * 'systemctl is-system-running' now returns 'offline' if the
10970 system is not booted with systemd. This command can now be
10971 used as a substitute for 'systemd-notify --booted'.
10973 * Watchdog timeouts have been increased to 3 minutes for all
10974 in-tree service files. Apparently, disk IO issues are more
10975 frequent than we hoped, and user reported >1 minute waiting
10978 * 'machine-id-commit' functionality has been merged into
10979 'machine-id-setup --commit'. The separate binary has been
10982 * The WorkingDirectory= directive in unit files may now be set
10983 to the special value '~'. In this case, the working
10984 directory is set to the home directory of the user
10985 configured in User=.
10987 * "machinectl shell" will now open the shell in the home
10988 directory of the selected user by default.
10990 * The CrashChVT= configuration file setting is renamed to
10991 CrashChangeVT=, following our usual logic of not
10992 abbreviating unnecessarily. The old directive is still
10993 supported for compat reasons. Also, this directive now takes
10994 an integer value between 1 and 63, or a boolean value. The
10995 formerly supported '-1' value for disabling stays around for
10998 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
10999 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
11000 RootDirectory= properties can now be set for transient
11003 * The systemd-analyze tool gained a new "set-log-target" verb
11004 to change the logging target the system manager logs to
11005 dynamically during runtime. This is similar to how
11006 "systemd-analyze set-log-level" already changes the log
11009 * In nspawn /sys is now mounted as tmpfs, with only a selected
11010 set of subdirectories mounted in from the real sysfs. This
11011 enhances security slightly, and is useful for ensuring user
11012 namespaces work correctly.
11014 * Support for USB FunctionFS activation has been added. This
11015 allows implementation of USB gadget services that are
11016 activated as soon as they are requested, so that they don't
11017 have to run continuously, similar to classic socket
11020 * The "systemctl exit" command now optionally takes an
11021 additional parameter that sets the exit code to return from
11022 the systemd manager when exiting. This is only relevant when
11023 running the systemd user instance, or when running the
11024 system instance in a container.
11026 * sd-bus gained the new API calls sd_bus_path_encode_many()
11027 and sd_bus_path_decode_many() that allow easy encoding and
11028 decoding of multiple identifier strings inside a D-Bus
11029 object path. Another new call sd_bus_default_flush_close()
11030 has been added to flush and close per-thread default
11033 * systemd-cgtop gained support for a -M/--machine= switch to
11034 show the control groups within a certain container only.
11036 * "systemctl kill" gained support for an optional --fail
11037 switch. If specified the requested operation will fail of no
11038 processes have been killed, because the unit had no
11039 processes attached, or similar.
11041 * A new systemd.crash_reboot=1 kernel command line option has
11042 been added that triggers a reboot after crashing. This can
11043 also be set through CrashReboot= in systemd.conf.
11045 * The RuntimeDirectory= setting now understands unit
11046 specifiers like %i or %f.
11048 * A new (still internal) library API sd-ipv4acd has been added,
11049 that implements address conflict detection for IPv4. It's
11050 based on code from sd-ipv4ll, and will be useful for
11051 detecting DHCP address conflicts.
11053 * File descriptors passed during socket activation may now be
11054 named. A new API sd_listen_fds_with_names() is added to
11055 access the names. The default names may be overridden,
11056 either in the .socket file using the FileDescriptorName=
11057 parameter, or by passing FDNAME= when storing the file
11058 descriptors using sd_notify().
11060 * systemd-networkd gained support for:
11062 - Setting the IPv6 Router Advertisement settings via
11063 IPv6AcceptRouterAdvertisements= in .network files.
11065 - Configuring the HelloTimeSec=, MaxAgeSec= and
11066 ForwardDelaySec= bridge parameters in .netdev files.
11068 - Configuring PreferredSource= for static routes in
11071 * The "ask-password" framework used to query for LUKS harddisk
11072 passwords or SSL passwords during boot gained support for
11073 caching passwords in the kernel keyring, if it is
11074 available. This makes sure that the user only has to type in
11075 a passphrase once if there are multiple objects to unlock
11076 with the same one. Previously, such password caching was
11077 available only when Plymouth was used; this moves the
11078 caching logic into the systemd codebase itself. The
11079 "systemd-ask-password" utility gained a new --keyname=
11080 switch to control which kernel keyring key to use for
11081 caching a password in. This functionality is also useful for
11082 enabling display managers such as gdm to automatically
11083 unlock the user's GNOME keyring if its passphrase, the
11084 user's password and the harddisk password are the same, if
11085 gdm-autologin is used.
11087 * When downloading tar or raw images using "machinectl
11088 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
11089 file is now also downloaded, if it is available and stored
11090 next to the image file.
11092 * Units of type ".socket" gained a new boolean setting
11093 Writable= which is only useful in conjunction with
11094 ListenSpecial=. If true, enables opening the specified
11095 special file in O_RDWR mode rather than O_RDONLY mode.
11097 * systemd-rfkill has been reworked to become a singleton
11098 service that is activated through /dev/rfkill on each rfkill
11099 state change and saves the settings to disk. This way,
11100 systemd-rfkill is now compatible with devices that exist
11101 only intermittendly, and even restores state if the previous
11102 system shutdown was abrupt rather than clean.
11104 * The journal daemon gained support for vacuuming old journal
11105 files controlled by the number of files that shall remain,
11106 in addition to the already existing control by size and by
11107 date. This is useful as journal interleaving performance
11108 degrades with too many separate journal files, and allows
11109 putting an effective limit on them. The new setting defaults
11110 to 100, but this may be changed by setting SystemMaxFiles=
11111 and RuntimeMaxFiles= in journald.conf. Also, the
11112 "journalctl" tool gained the new --vacuum-files= switch to
11113 manually vacuum journal files to leave only the specified
11114 number of files in place.
11116 * udev will now create /dev/disk/by-path links for ATA devices
11117 on kernels where that is supported.
11119 * Galician, Serbian, Turkish and Korean translations were added.
11121 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
11122 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
11123 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
11124 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
11125 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
11126 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
11127 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
11128 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
11129 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
11130 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
11131 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
11132 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
11133 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
11134 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
11135 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
11136 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
11137 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
11138 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
11140 — Berlin, 2015-10-07
11144 * The DHCP implementation of systemd-networkd gained a set of
11147 - The DHCP server now supports emitting DNS and NTP
11148 information. It may be enabled and configured via
11149 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
11150 and NTP information is enabled, but no servers are
11151 configured, the corresponding uplink information (if there
11152 is any) is propagated.
11154 - Server and client now support transmission and reception
11155 of timezone information. It can be configured via the
11156 newly introduced network options UseTimezone=,
11157 EmitTimezone=, and Timezone=. Transmission of timezone
11158 information is enabled between host and containers by
11159 default now: the container will change its local timezone
11160 to what the host has set.
11162 - Lease timeouts can now be configured via
11163 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
11165 - The DHCP server improved on the stability of
11166 leases. Clients are more likely to get the same lease
11167 information back, even if the server loses state.
11169 - The DHCP server supports two new configuration options to
11170 control the lease address pool metrics, PoolOffset= and
11173 * The encapsulation limit of tunnels in systemd-networkd may
11174 now be configured via 'EncapsulationLimit='. It allows
11175 modifying the maximum additional levels of encapsulation
11176 that are permitted to be prepended to a packet.
11178 * systemd now supports the concept of user buses replacing
11179 session buses, if used with dbus-1.10 (and enabled via dbus
11180 --enable-user-session). It previously only supported this on
11181 kdbus-enabled systems, and this release expands this to
11182 'dbus-daemon' systems.
11184 * systemd-networkd now supports predictable interface names
11185 for virtio devices.
11187 * systemd now optionally supports the new Linux kernel
11188 "unified" control group hierarchy. If enabled via the kernel
11189 command-line option 'systemd.unified_cgroup_hierarchy=1',
11190 systemd will try to mount the unified cgroup hierarchy
11191 directly on /sys/fs/cgroup. If not enabled, or not
11192 available, systemd will fall back to the legacy cgroup
11193 hierarchy setup, as before. Host system and containers can
11194 mix and match legacy and unified hierarchies as they
11195 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
11196 environment variable to individually select the hierarchy to
11197 use for executed containers. By default, nspawn will use the
11198 unified hierarchy for the containers if the host uses the
11199 unified hierarchy, and the legacy hierarchy otherwise.
11200 Please note that at this point the unified hierarchy is an
11201 experimental kernel feature and is likely to change in one
11202 of the next kernel releases. Therefore, it should not be
11203 enabled by default in downstream distributions yet. The
11204 minimum required kernel version for the unified hierarchy to
11205 work is 4.2. Note that when the unified hierarchy is used
11206 for the first time delegated access to controllers is
11207 safe. Because of this systemd-nspawn containers will get
11208 access to controllers now, as will systemd user
11209 sessions. This means containers and user sessions may now
11210 manage their own resources, partitioning up what the system
11213 * A new special scope unit "init.scope" has been introduced
11214 that encapsulates PID 1 of the system. It may be used to
11215 determine resource usage and enforce resource limits on PID
11216 1 itself. PID 1 hence moved out of the root of the control
11219 * The cgtop tool gained support for filtering out kernel
11220 threads when counting tasks in a control group. Also, the
11221 count of processes is now recursively summed up by
11222 default. Two options -k and --recursive= have been added to
11223 revert to old behaviour. The tool has also been updated to
11224 work correctly in containers now.
11226 * systemd-nspawn's --bind= and --bind-ro= options have been
11227 extended to allow creation of non-recursive bind mounts.
11229 * libsystemd gained two new calls sd_pid_get_cgroup() and
11230 sd_peer_get_cgroup() which return the control group path of
11231 a process or peer of a connected AF_UNIX socket. This
11232 function call is particularly useful when implementing
11233 delegated subtrees support in the control group hierarchy.
11235 * The "sd-event" event loop API of libsystemd now supports
11236 correct dequeuing of real-time signals, without losing
11239 * When systemd requests a polkit decision when managing units it
11240 will now add additional fields to the request, including unit
11241 name and desired operation. This enables more powerful polkit
11242 policies, that make decisions depending on these parameters.
11244 * nspawn learnt support for .nspawn settings files, that may
11245 accompany the image files or directories of containers, and
11246 may contain additional settings for the container. This is
11247 an alternative to configuring container parameters via the
11248 nspawn command line.
11250 Contributions from: Cristian Rodríguez, Daniel Mack, David
11251 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
11252 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
11253 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
11254 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
11255 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
11256 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
11257 Andersen, Tom Gundersen, Torstein Husebø
11259 — Berlin, 2015-09-08
11263 * machinectl gained a new verb 'shell' which opens a fresh
11264 shell on the target container or the host. It is similar to
11265 the existing 'login' command of machinectl, but spawns the
11266 shell directly without prompting for username or
11267 password. The pseudo machine '.host' now refers to the local
11268 host and is used by default. Hence, 'machinectl shell' can
11269 be used as replacement for 'su -' which spawns a session as
11270 a fresh systemd unit in a way that is fully isolated from
11271 the originating session.
11273 * systemd-networkd learned to cope with private-zone DHCP
11274 options and allows other programs to query the values.
11276 * SELinux access control when enabling/disabling units is no
11277 longer enforced with this release. The previous implementation
11278 was incorrect, and a new corrected implementation is not yet
11279 available. As unit file operations are still protected via
11280 polkit and D-Bus policy this is not a security problem. Yet,
11281 distributions which care about optimal SELinux support should
11282 probably not stabilize on this release.
11284 * sd-bus gained support for matches of type "arg0has=", that
11285 test for membership of strings in string arrays sent in bus
11288 * systemd-resolved now dumps the contents of its DNS and LLMNR
11289 caches to the logs on reception of the SIGUSR1 signal. This
11290 is useful to debug DNS behaviour.
11292 * The coredumpctl tool gained a new --directory= option to
11293 operate on journal files in a specific directory.
11295 * "systemctl reboot" and related commands gained a new
11296 "--message=" option which may be used to set a free-text
11297 wall message when shutting down or rebooting the
11298 system. This message is also logged, which is useful for
11299 figuring out the reason for a reboot or shutdown a
11302 * The "systemd-resolve-host" tool's -i switch now takes
11303 network interface numbers as alternative to interface names.
11305 * A new unit file setting for services has been introduced:
11306 UtmpMode= allows configuration of how precisely systemd
11307 handles utmp and wtmp entries for the service if this is
11308 enabled. This allows writing services that appear similar to
11309 user sessions in the output of the "w", "who", "last" and
11312 * systemd-resolved will now locally synthesize DNS resource
11313 records for the "localhost" and "gateway" domains as well as
11314 the local hostname. This should ensure that clients querying
11315 RRs via resolved will get similar results as those going via
11316 NSS, if nss-myhostname is enabled.
11318 Contributions from: Alastair Hughes, Alex Crawford, Daniel
11319 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
11320 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
11321 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
11322 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
11323 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
11324 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
11325 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
11326 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
11327 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
11328 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
11329 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11331 — Berlin, 2015-08-27
11335 * The systemd-efi-boot-generator functionality was merged into
11336 systemd-gpt-auto-generator.
11338 * systemd-networkd now supports Group Policy for vxlan
11339 devices. It can be enabled via the new boolean configuration
11340 option called 'GroupPolicyExtension='.
11342 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
11343 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
11344 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
11346 — Berlin, 2015-07-31
11350 * The python-systemd code has been removed from the systemd repository.
11351 A new repository has been created which accommodates the code from
11352 now on, and we kindly ask distributions to create a separate package
11353 for this: https://github.com/systemd/python-systemd
11355 * The systemd daemon will now reload its main configuration
11356 (/etc/systemd/system.conf) on daemon-reload.
11358 * sd-dhcp now exposes vendor specific extensions via
11359 sd_dhcp_lease_get_vendor_specific().
11361 * systemd-networkd gained a number of new configuration options.
11363 - A new boolean configuration option for TAP devices called
11364 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
11365 device, thus allowing to send and receive GSO packets.
11367 - A new tunnel configuration option called 'CopyDSCP='.
11368 If enabled, the DSCP field of ip6 tunnels is copied into the
11369 decapsulated packet.
11371 - A set of boolean bridge configuration options were added.
11372 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
11373 and 'UnicastFlood=' are now parsed by networkd and applied to the
11374 respective bridge link device via the respective IFLA_BRPORT_*
11377 - A new string configuration option to override the hostname sent
11378 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
11379 is true, networkd will use the configured hostname instead of the
11380 system hostname when sending DHCP requests.
11382 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
11383 networkd will configure the IPv6 flow-label of the tunnel device
11384 according to RFC2460.
11386 - The 'macvtap' virtual network devices are now supported, similar to
11387 the already supported 'macvlan' devices.
11389 * systemd-resolved now implements RFC5452 to improve resilience against
11390 cache poisoning. Additionally, source port randomization is enabled
11391 by default to further protect against DNS spoofing attacks.
11393 * nss-mymachines now supports translating UIDs and GIDs of running
11394 containers with user-namespaces enabled. If a container 'foo'
11395 translates a host uid 'UID' to the container uid 'TUID', then
11396 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
11397 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
11398 mapped as 'vg-foo-TGID'.
11400 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
11401 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
11402 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
11403 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
11404 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
11405 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
11406 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
11407 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
11408 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
11409 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
11411 — Berlin, 2015-07-29
11415 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
11416 There are no known issues with current sysfs, and udev does not need
11417 or should be used to work around such bugs.
11419 * udev does no longer enable USB HID power management. Several reports
11420 indicate, that some devices cannot handle that setting.
11422 * The udev accelerometer helper was removed. The functionality
11423 is now fully included in iio-sensor-proxy. But this means,
11424 older iio-sensor-proxy versions will no longer provide
11425 accelerometer/orientation data with this systemd version.
11426 Please upgrade iio-sensor-proxy to version 1.0.
11428 * networkd gained a new configuration option IPv6PrivacyExtensions=
11429 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
11430 for Stateless Address") on selected networks.
11432 * For the sake of fewer build-time dependencies and less code in the
11433 main repository, the python bindings are about to be removed in the
11434 next release. A new repository has been created which accommodates
11435 the code from now on, and we kindly ask distributions to create a
11436 separate package for this. The removal will take place in v223.
11438 https://github.com/systemd/python-systemd
11440 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
11441 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
11442 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
11443 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
11444 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
11445 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
11446 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
11447 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
11448 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
11449 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
11451 — Berlin, 2015-07-07
11455 * The sd-bus.h and sd-event.h APIs have now been declared
11456 stable and have been added to the official interface of
11457 libsystemd.so. sd-bus implements an alternative D-Bus client
11458 library, that is relatively easy to use, very efficient and
11459 supports both classic D-Bus as well as kdbus as transport
11460 backend. sd-event is a generic event loop abstraction that
11461 is built around Linux epoll, but adds features such as event
11462 prioritization or efficient timer handling. Both APIs are good
11463 choices for C programs looking for a bus and/or event loop
11464 implementation that is minimal and does not have to be
11465 portable to other kernels.
11467 * kdbus support is no longer compile-time optional. It is now
11468 always built-in. However, it can still be disabled at
11469 runtime using the kdbus=0 kernel command line setting, and
11470 that setting may be changed to default to off, by specifying
11471 --disable-kdbus at build-time. Note though that the kernel
11472 command line setting has no effect if the kdbus.ko kernel
11473 module is not installed, in which case kdbus is (obviously)
11474 also disabled. We encourage all downstream distributions to
11475 begin testing kdbus by adding it to the kernel images in the
11476 development distributions, and leaving kdbus support in
11479 * The minimal required util-linux version has been bumped to
11482 * Support for chkconfig (--enable-chkconfig) was removed in
11483 favor of calling an abstraction tool
11484 /lib/systemd/systemd-sysv-install. This needs to be
11485 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
11486 in README for details.
11488 * If there's a systemd unit and a SysV init script for the
11489 same service name, and the user executes "systemctl enable"
11490 for it (or a related call), then this will now enable both
11491 (or execute the related operation on both), not just the
11494 * The libudev API documentation has been converted from gtkdoc
11497 * gudev has been removed from the systemd tree, it is now an
11500 * The systemd-cgtop tool learnt a new --raw switch to generate
11501 "raw" (machine parsable) output.
11503 * networkd's IPForwarding= .network file setting learnt the
11504 new setting "kernel", which ensures that networkd does not
11505 change the IP forwarding sysctl from the default kernel
11508 * The systemd-logind bus API now exposes a new boolean
11509 property "Docked" that reports whether logind considers the
11510 system "docked", i.e. connected to a docking station or not.
11512 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
11513 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
11514 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
11515 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
11516 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
11517 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
11518 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
11519 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
11520 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
11521 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
11522 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
11523 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
11524 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
11525 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
11526 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
11527 Fink, Zbigniew Jędrzejewski-Szmek
11529 — Berlin, 2015-06-19
11533 * The gudev library has been extracted into a separate repository
11534 available at: https://git.gnome.org/browse/libgudev/
11535 It is now managed as part of the Gnome project. Distributions
11536 are recommended to pass --disable-gudev to systemd and use
11537 gudev from the Gnome project instead. gudev is still included
11538 in systemd, for now. It will be removed soon, though. Please
11539 also see the announcement-thread on systemd-devel:
11540 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
11542 * systemd now exposes a CPUUsageNSec= property for each
11543 service unit on the bus, that contains the overall consumed
11544 CPU time of a service (the sum of what each process of the
11545 service consumed). This value is only available if
11546 CPUAccounting= is turned on for a service, and is then shown
11547 in the "systemctl status" output.
11549 * Support for configuring alternative mappings of the old SysV
11550 runlevels to systemd targets has been removed. They are now
11551 hardcoded in a way that runlevels 2, 3, 4 all map to
11552 multi-user.target and 5 to graphical.target (which
11553 previously was already the default behaviour).
11555 * The auto-mounter logic gained support for mount point
11556 expiry, using a new TimeoutIdleSec= setting in .automount
11557 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
11559 * The EFI System Partition (ESP) as mounted to /boot by
11560 systemd-efi-boot-generator will now be unmounted
11561 automatically after 2 minutes of not being used. This should
11562 minimize the risk of ESP corruptions.
11564 * New /etc/fstab options x-systemd.requires= and
11565 x-systemd.requires-mounts-for= are now supported to express
11566 additional dependencies for mounts. This is useful for
11567 journaling file systems that support external journal
11568 devices or overlay file systems that require underlying file
11569 systems to be mounted.
11571 * systemd does not support direct live-upgrades (via systemctl
11572 daemon-reexec) from versions older than v44 anymore. As no
11573 distribution we are aware of shipped such old versions in a
11574 stable release this should not be problematic.
11576 * When systemd forks off a new per-connection service instance
11577 it will now set the $REMOTE_ADDR environment variable to the
11578 remote IP address, and $REMOTE_PORT environment variable to
11579 the remote IP port. This behaviour is similar to the
11580 corresponding environment variables defined by CGI.
11582 * systemd-networkd gained support for uplink failure
11583 detection. The BindCarrier= option allows binding interface
11584 configuration dynamically to the link sense of other
11585 interfaces. This is useful to achieve behaviour like in
11588 * systemd-networkd gained support for configuring the DHCP
11589 client identifier to use when requesting leases.
11591 * systemd-networkd now has a per-network UseNTP= option to
11592 configure whether NTP server information acquired via DHCP
11593 is passed on to services like systemd-timesyncd.
11595 * systemd-networkd gained support for vti6 tunnels.
11597 * Note that systemd-networkd manages the sysctl variable
11598 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
11599 it is configured for since v219. The variable controls IP
11600 forwarding, and is a per-interface alternative to the global
11601 /proc/sys/net/ipv[46]/ip_forward. This setting is
11602 configurable in the IPForward= option, which defaults to
11603 "no". This means if networkd is used for an interface it is
11604 no longer sufficient to set the global sysctl option to turn
11605 on IP forwarding! Instead, the .network file option
11606 IPForward= needs to be turned on! Note that the
11607 implementation of this behaviour was broken in v219 and has
11608 been fixed in v220.
11610 * Many bonding and vxlan options are now configurable in
11613 * systemd-nspawn gained a new --property= setting to set unit
11614 properties for the container scope. This is useful for
11615 setting resource parameters (e.g. "CPUShares=500") on
11616 containers started from the command line.
11618 * systemd-nspawn gained a new --private-users= switch to make
11619 use of user namespacing available on recent Linux kernels.
11621 * systemd-nspawn may now be called as part of a shell pipeline
11622 in which case the pipes used for stdin and stdout are passed
11623 directly to the process invoked in the container, without
11624 indirection via a pseudo tty.
11626 * systemd-nspawn gained a new switch to control the UNIX
11627 signal to use when killing the init process of the container
11628 when shutting down.
11630 * systemd-nspawn gained a new --overlay= switch for mounting
11631 overlay file systems into the container using the new kernel
11634 * When a container image is imported via systemd-importd and
11635 the host file system is not btrfs, a loopback block device
11636 file is created in /var/lib/machines.raw with a btrfs file
11637 system inside. It is then mounted to /var/lib/machines to
11638 enable btrfs features for container management. The loopback
11639 file and btrfs file system is grown as needed when container
11640 images are imported via systemd-importd.
11642 * systemd-machined/systemd-importd gained support for btrfs
11643 quota, to enforce container disk space limits on disk. This
11644 is exposed in "machinectl set-limit".
11646 * systemd-importd now can import containers from local .tar,
11647 .raw and .qcow2 images, and export them to .tar and .raw. It
11648 can also import dkr v2 images now from the network (on top
11651 * systemd-importd gained support for verifying downloaded
11652 images with gpg2 (previously only gpg1 was supported).
11654 * systemd-machined, systemd-logind, systemd: most bus calls are
11655 now accessible to unprivileged processes via polkit. Also,
11656 systemd-logind will now allow users to kill their own sessions
11657 without further privileges or authorization.
11659 * systemd-shutdownd has been removed. This service was
11660 previously responsible for implementing scheduled shutdowns
11661 as exposed in /usr/bin/shutdown's time parameter. This
11662 functionality has now been moved into systemd-logind and is
11663 accessible via a bus interface.
11665 * "systemctl reboot" gained a new switch --firmware-setup that
11666 can be used to reboot into the EFI firmware setup, if that
11667 is available. systemd-logind now exposes an API on the bus
11668 to trigger such reboots, in case graphical desktop UIs want
11669 to cover this functionality.
11671 * "systemctl enable", "systemctl disable" and "systemctl mask"
11672 now support a new "--now" switch. If specified the units
11673 that are enabled will also be started, and the ones
11674 disabled/masked also stopped.
11676 * The Gummiboot EFI boot loader tool has been merged into
11677 systemd, and renamed to "systemd-boot". The bootctl tool has been
11678 updated to support systemd-boot.
11680 * An EFI kernel stub has been added that may be used to create
11681 kernel EFI binaries that contain not only the actual kernel,
11682 but also an initrd, boot splash, command line and OS release
11683 information. This combined binary can then be signed as a
11684 single image, so that the firmware can verify it all in one
11685 step. systemd-boot has special support for EFI binaries created
11686 like this and can extract OS release information from them
11687 and show them in the boot menu. This functionality is useful
11688 to implement cryptographically verified boot schemes.
11690 * Optional support has been added to systemd-fsck to pass
11691 fsck's progress report to an AF_UNIX socket in the file
11694 * udev will no longer create device symlinks for all block devices by
11695 default. A deny list for excluding special block devices from this
11696 logic has been turned into an allow list that requires picking block
11697 devices explicitly that require device symlinks.
11699 * A new (currently still internal) API sd-device.h has been
11700 added to libsystemd. This modernized API is supposed to
11701 replace libudev eventually. In fact, already much of libudev
11702 is now just a wrapper around sd-device.h.
11704 * A new hwdb database for storing metadata about pointing
11705 stick devices has been added.
11707 * systemd-tmpfiles gained support for setting file attributes
11708 similar to the "chattr" tool with new 'h' and 'H' lines.
11710 * systemd-journald will no longer unconditionally set the
11711 btrfs NOCOW flag on new journal files. This is instead done
11712 with tmpfiles snippet using the new 'h' line type. This
11713 allows easy disabling of this logic, by masking the
11714 journal-nocow.conf tmpfiles file.
11716 * systemd-journald will now translate audit message types to
11717 human readable identifiers when writing them to the
11718 journal. This should improve readability of audit messages.
11720 * The LUKS logic gained support for the offset= and skip=
11721 options in /etc/crypttab, as previously implemented by
11724 * /usr/lib/os-release gained a new optional field VARIANT= for
11725 distributions that support multiple variants (such as a
11726 desktop edition, a server edition, …)
11728 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
11729 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
11730 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
11731 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
11732 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
11733 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
11734 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
11735 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
11736 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
11737 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
11738 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
11739 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
11740 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
11741 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
11742 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
11743 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
11744 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
11745 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
11746 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
11747 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
11748 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
11749 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
11750 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
11751 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
11752 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
11753 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
11754 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
11756 — Berlin, 2015-05-22
11760 * Introduce a new API "sd-hwdb.h" for querying the hardware
11761 metadata database. With this minimal interface one can query
11762 and enumerate the udev hwdb, decoupled from the old libudev
11763 library. libudev's interface for this is now only a wrapper
11764 around sd-hwdb. A new tool systemd-hwdb has been added to
11765 interface with and update the database.
11767 * When any of systemd's tools copies files (for example due to
11768 tmpfiles' C lines) a btrfs reflink will attempted first,
11769 before bytewise copying is done.
11771 * systemd-nspawn gained a new --ephemeral switch. When
11772 specified a btrfs snapshot is taken of the container's root
11773 directory, and immediately removed when the container
11774 terminates again. Thus, a container can be started whose
11775 changes never alter the container's root directory, and are
11776 lost on container termination. This switch can also be used
11777 for starting a container off the root file system of the
11778 host without affecting the host OS. This switch is only
11779 available on btrfs file systems.
11781 * systemd-nspawn gained a new --template= switch. It takes the
11782 path to a container tree to use as template for the tree
11783 specified via --directory=, should that directory be
11784 missing. This allows instantiating containers dynamically,
11785 on first run. This switch is only available on btrfs file
11788 * When a .mount unit refers to a mount point on which multiple
11789 mounts are stacked, and the .mount unit is stopped all of
11790 the stacked mount points will now be unmounted until no
11791 mount point remains.
11793 * systemd now has an explicit notion of supported and
11794 unsupported unit types. Jobs enqueued for unsupported unit
11795 types will now fail with an "unsupported" error code. More
11796 specifically .swap, .automount and .device units are not
11797 supported in containers, .busname units are not supported on
11798 non-kdbus systems. .swap and .automount are also not
11799 supported if their respective kernel compile time options
11802 * machinectl gained support for two new "copy-from" and
11803 "copy-to" commands for copying files from a running
11804 container to the host or vice versa.
11806 * machinectl gained support for a new "bind" command to bind
11807 mount host directories into local containers. This is
11808 currently only supported for nspawn containers.
11810 * networkd gained support for configuring bridge forwarding
11811 database entries (fdb) from .network files.
11813 * A new tiny daemon "systemd-importd" has been added that can
11814 download container images in tar, raw, qcow2 or dkr formats,
11815 and make them available locally in /var/lib/machines, so
11816 that they can run as nspawn containers. The daemon can GPG
11817 verify the downloads (not supported for dkr, since it has no
11818 provisions for verifying downloads). It will transparently
11819 decompress bz2, xz, gzip compressed downloads if necessary,
11820 and restore sparse files on disk. The daemon uses privilege
11821 separation to ensure the actual download logic runs with
11822 fewer privileges than the daemon itself. machinectl has
11823 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
11824 make the functionality of importd available to the
11825 user. With this in place the Fedora and Ubuntu "Cloud"
11826 images can be downloaded and booted as containers unmodified
11827 (the Fedora images lack the appropriate GPG signature files
11828 currently, so they cannot be verified, but this will change
11829 soon, hopefully). Note that downloading images is currently
11830 only fully supported on btrfs.
11832 * machinectl is now able to list container images found in
11833 /var/lib/machines, along with some metadata about sizes of
11834 disk and similar. If the directory is located on btrfs and
11835 quota is enabled, this includes quota display. A new command
11836 "image-status" has been added that shows additional
11837 information about images.
11839 * machinectl is now able to clone container images
11840 efficiently, if the underlying file system (btrfs) supports
11841 it, with the new "machinectl clone" command. It also
11842 gained commands for renaming and removing images, as well as
11843 marking them read-only or read-write (supported also on
11844 legacy file systems).
11846 * networkd gained support for collecting LLDP network
11847 announcements, from hardware that supports this. This is
11848 shown in networkctl output.
11850 * systemd-run gained support for a new -t (--pty) switch for
11851 invoking a binary on a pty whose input and output is
11852 connected to the invoking terminal. This allows executing
11853 processes as system services while interactively
11854 communicating with them via the terminal. Most interestingly
11855 this is supported across container boundaries. Invoking
11856 "systemd-run -t /bin/bash" is an alternative to running a
11857 full login session, the difference being that the former
11858 will not register a session, nor go through the PAM session
11861 * tmpfiles gained support for a new "v" line type for creating
11862 btrfs subvolumes. If the underlying file system is a legacy
11863 file system, this automatically degrades to creating a
11864 normal directory. Among others /var/lib/machines is now
11865 created like this at boot, should it be missing.
11867 * The directory /var/lib/containers/ has been deprecated and
11868 been replaced by /var/lib/machines. The term "machines" has
11869 been used in the systemd context as generic term for both
11870 VMs and containers, and hence appears more appropriate for
11871 this, as the directory can also contain raw images bootable
11874 * systemd-nspawn when invoked with -M but without --directory=
11875 or --image= is now capable of searching for the container
11876 root directory, subvolume or disk image automatically, in
11877 /var/lib/machines. systemd-nspawn@.service has been updated
11878 to make use of this, thus allowing it to be used for raw
11881 * A new machines.target unit has been introduced that is
11882 supposed to group all containers/VMs invoked as services on
11883 the system. systemd-nspawn@.service has been updated to
11884 integrate with that.
11886 * machinectl gained a new "start" command, for invoking a
11887 container as a service. "machinectl start foo" is mostly
11888 equivalent to "systemctl start systemd-nspawn@foo.service",
11889 but handles escaping in a nicer way.
11891 * systemd-nspawn will now mount most of the cgroupfs tree
11892 read-only into each container, with the exception of the
11893 container's own subtree in the name=systemd hierarchy.
11895 * journald now sets the special FS_NOCOW file flag for its
11896 journal files. This should improve performance on btrfs, by
11897 avoiding heavy fragmentation when journald's write-pattern
11898 is used on COW file systems. It degrades btrfs' data
11899 integrity guarantees for the files to the same levels as for
11900 ext3/ext4 however. This should be OK though as journald does
11901 its own data integrity checks and all its objects are
11902 checksummed on disk. Also, journald should handle btrfs disk
11903 full events a lot more gracefully now, by processing SIGBUS
11904 errors, and not relying on fallocate() anymore.
11906 * When journald detects that journal files it is writing to
11907 have been deleted it will immediately start new journal
11910 * systemd now provides a way to store file descriptors
11911 per-service in PID 1. This is useful for daemons to ensure
11912 that fds they require are not lost during a daemon
11913 restart. The fds are passed to the daemon on the next
11914 invocation in the same way socket activation fds are
11915 passed. This is now used by journald to ensure that the
11916 various sockets connected to all the system's stdout/stderr
11917 are not lost when journald is restarted. File descriptors
11918 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
11919 an extension to sd_notify(). Note that a limit is enforced
11920 on the number of fds a service can store in PID 1, and it
11921 defaults to 0, so that no fds may be stored, unless this is
11922 explicitly turned on.
11924 * The default TERM variable to use for units connected to a
11925 terminal, when no other value is explicitly is set is now
11926 vt220 rather than vt102. This should be fairly safe still,
11927 but allows PgUp/PgDn work.
11929 * The /etc/crypttab option header= as known from Debian is now
11932 * "loginctl user-status" and "loginctl session-status" will
11933 now show the last 10 lines of log messages of the
11934 user/session following the status output. Similar,
11935 "machinectl status" will show the last 10 log lines
11936 associated with a virtual machine or container
11937 service. (Note that this is usually not the log messages
11938 done in the VM/container itself, but simply what the
11939 container manager logs. For nspawn this includes all console
11942 * "loginctl session-status" without further argument will now
11943 show the status of the session of the caller. Similar,
11944 "lock-session", "unlock-session", "activate",
11945 "enable-linger", "disable-linger" may now be called without
11946 session/user parameter in which case they apply to the
11947 caller's session/user.
11949 * An X11 session scriptlet is now shipped that uploads
11950 $DISPLAY and $XAUTHORITY into the environment of the systemd
11951 --user daemon if a session begins. This should improve
11952 compatibility with X11 enabled applications run as systemd
11955 * Generators are now subject to masking via /etc and /run, the
11956 same way as unit files.
11958 * networkd .network files gained support for configuring
11959 per-link IPv4/IPv6 packet forwarding as well as IPv4
11960 masquerading. This is by default turned on for veth links to
11961 containers, as registered by systemd-nspawn. This means that
11962 nspawn containers run with --network-veth will now get
11963 automatic routed access to the host's networks without any
11964 further configuration or setup, as long as networkd runs on
11967 * systemd-nspawn gained the --port= (-p) switch to expose TCP
11968 or UDP posts of a container on the host. With this in place
11969 it is possible to run containers with private veth links
11970 (--network-veth), and have their functionality exposed on
11971 the host as if their services were running directly on the
11974 * systemd-nspawn's --network-veth switch now gained a short
11975 version "-n", since with the changes above it is now truly
11976 useful out-of-the-box. The systemd-nspawn@.service has been
11977 updated to make use of it too by default.
11979 * systemd-nspawn will now maintain a per-image R/W lock, to
11980 ensure that the same image is not started more than once
11981 writable. (It's OK to run an image multiple times
11982 simultaneously in read-only mode.)
11984 * systemd-nspawn's --image= option is now capable of
11985 dissecting and booting MBR and GPT disk images that contain
11986 only a single active Linux partition. Previously it
11987 supported only GPT disk images with proper GPT type
11988 IDs. This allows running cloud images from major
11989 distributions directly with systemd-nspawn, without
11992 * In addition to collecting mouse dpi data in the udev
11993 hardware database, there's now support for collecting angle
11994 information for mouse scroll wheels. The database is
11995 supposed to guarantee similar scrolling behavior on mice
11996 that it knows about. There's also support for collecting
11997 information about Touchpad types.
11999 * udev's input_id built-in will now also collect touch screen
12000 dimension data and attach it to probed devices.
12002 * /etc/os-release gained support for a Distribution Privacy
12005 * networkd gained support for creating "ipvlan", "gretap",
12006 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
12008 * systemd-tmpfiles gained support for "a" lines for setting
12011 * systemd-nspawn will now mount /tmp in the container to
12012 tmpfs, automatically.
12014 * systemd now exposes the memory.usage_in_bytes cgroup
12015 attribute and shows it for each service in the "systemctl
12016 status" output, if available.
12018 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
12019 immediate reboot is triggered. This useful if shutdown is
12020 hung and is unable to complete, to expedite the
12021 operation. Note that this kind of reboot will still unmount
12022 all file systems, and hence should not result in fsck being
12023 run on next reboot.
12025 * A .device unit for an optical block device will now be
12026 considered active only when a medium is in the drive. Also,
12027 mount units are now bound to their backing devices thus
12028 triggering automatic unmounting when devices become
12029 unavailable. With this in place systemd will now
12030 automatically unmount left-over mounts when a CD-ROM is
12031 ejected or a USB stick is yanked from the system.
12033 * networkd-wait-online now has support for waiting for
12034 specific interfaces only (with globbing), and for giving up
12035 after a configurable timeout.
12037 * networkd now exits when idle. It will be automatically
12038 restarted as soon as interfaces show up, are removed or
12039 change state. networkd will stay around as long as there is
12040 at least one DHCP state machine or similar around, that keep
12043 * networkd may now configure IPv6 link-local addressing in
12044 addition to IPv4 link-local addressing.
12046 * The IPv6 "token" for use in SLAAC may now be configured for
12047 each .network interface in networkd.
12049 * Routes configured with networkd may now be assigned a scope
12052 * networkd's [Match] sections now support globbing and lists
12053 of multiple space-separated matches per item.
12055 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
12056 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
12057 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
12058 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
12059 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
12060 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
12061 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
12062 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
12063 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
12064 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
12065 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
12066 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
12067 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
12068 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
12069 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
12070 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
12071 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
12072 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
12073 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
12074 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
12075 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
12076 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
12077 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
12078 Hoffmann, Zbigniew Jędrzejewski-Szmek
12080 — Berlin, 2015-02-16
12084 * When querying unit file enablement status (for example via
12085 "systemctl is-enabled"), a new state "indirect" is now known
12086 which indicates that a unit might not be enabled itself, but
12087 another unit listed in its Also= setting might be.
12089 * Similar to the various existing ConditionXYZ= settings for
12090 units, there are now matching AssertXYZ= settings. While
12091 failing conditions cause a unit to be skipped, but its job
12092 to succeed, failing assertions declared like this will cause
12093 a unit start operation and its job to fail.
12095 * hostnamed now knows a new chassis type "embedded".
12097 * systemctl gained a new "edit" command. When used on a unit
12098 file, this allows extending unit files with .d/ drop-in
12099 configuration snippets or editing the full file (after
12100 copying it from /usr/lib to /etc). This will invoke the
12101 user's editor (as configured with $EDITOR), and reload the
12102 modified configuration after editing.
12104 * "systemctl status" now shows the suggested enablement state
12105 for a unit, as declared in the (usually vendor-supplied)
12106 system preset files.
12108 * nss-myhostname will now resolve the single-label hostname
12109 "gateway" to the locally configured default IP routing
12110 gateways, ordered by their metrics. This assigns a stable
12111 name to the used gateways, regardless which ones are
12112 currently configured. Note that the name will only be
12113 resolved after all other name sources (if nss-myhostname is
12114 configured properly) and should hence not negatively impact
12115 systems that use the single-label hostname "gateway" in
12118 * systemd-inhibit now allows filtering by mode when listing
12121 * Scope and service units gained a new "Delegate" boolean
12122 property, which, when set, allows processes running inside the
12123 unit to further partition resources. This is primarily
12124 useful for systemd user instances as well as container
12127 * journald will now pick up audit messages directly from
12128 the kernel, and log them like any other log message. The
12129 audit fields are split up and fully indexed. This means that
12130 journalctl in many ways is now a (nicer!) alternative to
12131 ausearch, the traditional audit client. Note that this
12132 implements only a minimal audit client. If you want the
12133 special audit modes like reboot-on-log-overflow, please use
12134 the traditional auditd instead, which can be used in
12135 parallel to journald.
12137 * The ConditionSecurity= unit file option now understands the
12138 special string "audit" to check whether auditing is
12141 * journalctl gained two new commands --vacuum-size= and
12142 --vacuum-time= to delete old journal files until the
12143 remaining ones take up no more than the specified size on disk,
12144 or are not older than the specified time.
12146 * A new, native PPPoE library has been added to sd-network,
12147 systemd's library of light-weight networking protocols. This
12148 library will be used in a future version of networkd to
12149 enable PPPoE communication without an external pppd daemon.
12151 * The busctl tool now understands a new "capture" verb that
12152 works similar to "monitor", but writes a packet capture
12153 trace to STDOUT that can be redirected to a file which is
12154 compatible with libcap's capture file format. This can then
12155 be loaded in Wireshark and similar tools to inspect bus
12158 * The busctl tool now understands a new "tree" verb that shows
12159 the object trees of a specific service on the bus, or of all
12162 * The busctl tool now understands a new "introspect" verb that
12163 shows all interfaces and members of objects on the bus,
12164 including their signature and values. This is particularly
12165 useful to get more information about bus objects shown by
12166 the new "busctl tree" command.
12168 * The busctl tool now understands new verbs "call",
12169 "set-property" and "get-property" for invoking bus method
12170 calls, setting and getting bus object properties in a
12173 * busctl gained a new --augment-creds= argument that controls
12174 whether the tool shall augment credential information it
12175 gets from the bus with data from /proc, in a possibly
12178 * nspawn's --link-journal= switch gained two new values
12179 "try-guest" and "try-host" that work like "guest" and
12180 "host", but do not fail if the host has no persistent
12181 journaling enabled. -j is now equivalent to
12182 --link-journal=try-guest.
12184 * macvlan network devices created by nspawn will now have
12185 stable MAC addresses.
12187 * A new SmackProcessLabel= unit setting has been added, which
12188 controls the SMACK security label processes forked off by
12189 the respective unit shall use.
12191 * If compiled with --enable-xkbcommon, systemd-localed will
12192 verify x11 keymap settings by compiling the given keymap. It
12193 will spew out warnings if the compilation fails. This
12194 requires libxkbcommon to be installed.
12196 * When a coredump is collected, a larger number of metadata
12197 fields is now collected and included in the journal records
12198 created for it. More specifically, control group membership,
12199 environment variables, memory maps, working directory,
12200 chroot directory, /proc/$PID/status, and a list of open file
12201 descriptors is now stored in the log entry.
12203 * The udev hwdb now contains DPI information for mice. For
12206 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
12208 * All systemd programs that read standalone configuration
12209 files in /etc now also support a corresponding series of
12210 .conf.d configuration directories in /etc/, /run/,
12211 /usr/local/lib/, /usr/lib/, and (if configured with
12212 --enable-split-usr) /lib/. In particular, the following
12213 configuration files now have corresponding configuration
12214 directories: system.conf user.conf, logind.conf,
12215 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
12216 resolved.conf, timesyncd.conf, journal-remote.conf, and
12217 journal-upload.conf. Note that distributions should use the
12218 configuration directories in /usr/lib/; the directories in
12219 /etc/ are reserved for the system administrator.
12221 * systemd-rfkill will no longer take the rfkill device name
12222 into account when storing rfkill state on disk, as the name
12223 might be dynamically assigned and not stable. Instead, the
12224 ID_PATH udev variable combined with the rfkill type (wlan,
12225 bluetooth, …) is used.
12227 * A new service systemd-machine-id-commit.service has been
12228 added. When used on systems where /etc is read-only during
12229 boot, and /etc/machine-id is not initialized (but an empty
12230 file), this service will copy the temporary machine ID
12231 created as replacement into /etc after the system is fully
12232 booted up. This is useful for systems that are freshly
12233 installed with a non-initialized machine ID, but should get
12234 a fixed machine ID for subsequent boots.
12236 * networkd's .netdev files now provide a large set of
12237 configuration parameters for VXLAN devices. Similarly, the
12238 bridge port cost parameter is now configurable in .network
12239 files. There's also new support for configuring IP source
12240 routing. networkd .link files gained support for a new
12241 OriginalName= match that is useful to match against the
12242 original interface name the kernel assigned. .network files
12243 may include MTU= and MACAddress= fields for altering the MTU
12244 and MAC address while being connected to a specific network
12247 * The LUKS logic gained supported for configuring
12248 UUID-specific key files. There's also new support for naming
12249 LUKS device from the kernel command line, using the new
12250 luks.name= argument.
12252 * Timer units may now be transiently created via the bus API
12253 (this was previously already available for scope and service
12254 units). In addition it is now possible to create multiple
12255 transient units at the same time with a single bus call. The
12256 "systemd-run" tool has been updated to make use of this for
12257 running commands on a specified time, in at(1)-style.
12259 * tmpfiles gained support for "t" lines, for assigning
12260 extended attributes to files. Among other uses this may be
12261 used to assign SMACK labels to files.
12263 Contributions from: Alin Rauta, Alison Chaiken, Andrej
12264 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
12265 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
12266 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
12267 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
12268 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
12269 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
12270 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
12271 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
12272 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
12273 Michael Biebl, Michael Chapman, Michael Marineau, Michal
12274 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
12275 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
12276 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
12277 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
12278 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
12279 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
12280 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
12282 — Berlin, 2014-12-10
12286 * journalctl gained the new options -t/--identifier= to match
12287 on the syslog identifier (aka "tag"), as well as --utc to
12288 show log timestamps in the UTC timezone. journalctl now also
12289 accepts -n/--lines=all to disable line capping in a pager.
12291 * journalctl gained a new switch, --flush, that synchronously
12292 flushes logs from /run/log/journal to /var/log/journal if
12293 persistent storage is enabled. systemd-journal-flush.service
12294 now waits until the operation is complete.
12296 * Services can notify the manager before they start a reload
12297 (by sending RELOADING=1) or shutdown (by sending
12298 STOPPING=1). This allows the manager to track and show the
12299 internal state of daemons and closes a race condition when
12300 the process is still running but has closed its D-Bus
12303 * Services with Type=oneshot do not have to have any ExecStart
12306 * User units are now loaded also from
12307 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
12308 /run/systemd/user directory that was already previously
12309 supported, but is under the control of the user.
12311 * Job timeouts (i.e. timeouts on the time a job that is
12312 queued stays in the run queue) can now optionally result in
12313 immediate reboot or power-off actions (JobTimeoutAction= and
12314 JobTimeoutRebootArgument=). This is useful on ".target"
12315 units, to limit the maximum time a target remains
12316 undispatched in the run queue, and to trigger an emergency
12317 operation in such a case. This is now used by default to
12318 turn off the system if boot-up (as defined by everything in
12319 basic.target) hangs and does not complete for at least
12320 15min. Also, if power-off or reboot hang for at least 30min
12321 an immediate power-off/reboot operation is triggered. This
12322 functionality is particularly useful to increase reliability
12323 on embedded devices, but also on laptops which might
12324 accidentally get powered on when carried in a backpack and
12325 whose boot stays stuck in a hard disk encryption passphrase
12328 * systemd-logind can be configured to also handle lid switch
12329 events even when the machine is docked or multiple displays
12330 are attached (HandleLidSwitchDocked= option).
12332 * A helper binary and a service have been added which can be
12333 used to resume from hibernation in the initramfs. A
12334 generator will parse the resume= option on the kernel
12335 command line to trigger resume.
12337 * A user console daemon systemd-consoled has been
12338 added. Currently, it is a preview, and will so far open a
12339 single terminal on each session of the user marked as
12340 Desktop=systemd-console.
12342 * Route metrics can be specified for DHCP routes added by
12345 * The SELinux context of socket-activated services can be set
12346 from the information provided by the networking stack
12347 (SELinuxContextFromNet= option).
12349 * Userspace firmware loading support has been removed and
12350 the minimum supported kernel version is thus bumped to 3.7.
12352 * Timeout for udev workers has been increased from 1 to 3
12353 minutes, but a warning will be printed after 1 minute to
12354 help diagnose kernel modules that take a long time to load.
12356 * Udev rules can now remove tags on devices with TAG-="foobar".
12358 * systemd's readahead implementation has been removed. In many
12359 circumstances it didn't give expected benefits even for
12360 rotational disk drives and was becoming less relevant in the
12361 age of SSDs. As none of the developers has been using
12362 rotating media anymore, and nobody stepped up to actively
12363 maintain this component of systemd it has now been removed.
12365 * Swap units can use Options= to specify discard options.
12366 Discard options specified for swaps in /etc/fstab are now
12369 * Docker containers are now detected as a separate type of
12372 * The Password Agent protocol gained support for queries where
12373 the user input is shown, useful e.g. for user names.
12374 systemd-ask-password gained a new --echo option to turn that
12377 * The default sysctl.d/ snippets will now set:
12379 net.core.default_qdisc = fq_codel
12381 This selects Fair Queuing Controlled Delay as the default
12382 queuing discipline for network interfaces. fq_codel helps
12383 fight the network bufferbloat problem. It is believed to be
12384 a good default with no tuning required for most workloads.
12385 Downstream distributions may override this choice. On 10Gbit
12386 servers that do not do forwarding, "fq" may perform better.
12387 Systems without a good clocksource should use "pfifo_fast".
12389 * If kdbus is enabled during build a new option BusPolicy= is
12390 available for service units, that allows locking all service
12391 processes into a stricter bus policy, in order to limit
12392 access to various bus services, or even hide most of them
12393 from the service's view entirely.
12395 * networkctl will now show the .network and .link file
12396 networkd has applied to a specific interface.
12398 * sd-login gained a new API call sd_session_get_desktop() to
12399 query which desktop environment has been selected for a
12402 * UNIX utmp support is now compile-time optional to support
12403 legacy-free systems.
12405 * systemctl gained two new commands "add-wants" and
12406 "add-requires" for pulling in units from specific targets
12409 * If the word "rescue" is specified on the kernel command line
12410 the system will now boot into rescue mode (aka
12411 rescue.target), which was previously available only by
12412 specifying "1" or "systemd.unit=rescue.target" on the kernel
12413 command line. This new kernel command line option nicely
12414 mirrors the already existing "emergency" kernel command line
12417 * New kernel command line options mount.usr=, mount.usrflags=,
12418 mount.usrfstype= have been added that match root=, rootflags=,
12419 rootfstype= but allow mounting a specific file system to
12422 * The $NOTIFY_SOCKET is now also passed to control processes of
12423 services, not only the main process.
12425 * This version reenables support for fsck's -l switch. This
12426 means at least version v2.25 of util-linux is required for
12427 operation, otherwise dead-locks on device nodes may
12428 occur. Again: you need to update util-linux to at least
12429 v2.25 when updating systemd to v217.
12431 * The "multi-seat-x" tool has been removed from systemd, as
12432 its functionality has been integrated into X servers 1.16,
12433 and the tool is hence redundant. It is recommended to update
12434 display managers invoking this tool to simply invoke X
12435 directly from now on, again.
12437 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
12438 message flag has been added for all of systemd's polkit
12439 authenticated method calls has been added. In particular this
12440 now allows optional interactive authorization via polkit for
12441 many of PID1's privileged operations such as unit file
12442 enabling and disabling.
12444 * "udevadm hwdb --update" learnt a new switch "--usr" for
12445 placing the rebuilt hardware database in /usr instead of
12446 /etc. When used only hardware database entries stored in
12447 /usr will be used, and any user database entries in /etc are
12448 ignored. This functionality is useful for vendors to ship a
12449 pre-built database on systems where local configuration is
12450 unnecessary or unlikely.
12452 * Calendar time specifications in .timer units now also
12453 understand the strings "semi-annually", "quarterly" and
12454 "minutely" as shortcuts (in addition to the preexisting
12455 "annually", "hourly", …).
12457 * systemd-tmpfiles will now correctly create files in /dev
12458 at boot which are marked for creation only at boot. It is
12459 recommended to always create static device nodes with 'c!'
12460 and 'b!', so that they are created only at boot and not
12461 overwritten at runtime.
12463 * When the watchdog logic is used for a service (WatchdogSec=)
12464 and the watchdog timeout is hit the service will now be
12465 terminated with SIGABRT (instead of just SIGTERM), in order
12466 to make sure a proper coredump and backtrace is
12467 generated. This ensures that hanging services will result in
12468 similar coredump/backtrace behaviour as services that hit a
12469 segmentation fault.
12471 Contributions from: Andreas Henriksson, Andrei Borzenkov,
12472 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
12473 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
12474 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
12475 Herrmann, David Sommerseth, David Strauss, Emil Renner
12476 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
12477 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
12478 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
12479 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
12480 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
12481 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
12482 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
12483 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
12484 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
12485 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
12486 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
12487 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
12488 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
12489 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
12490 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
12491 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
12494 — Berlin, 2014-10-28
12498 * timedated no longer reads NTP implementation unit names from
12499 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
12500 implementations should add a
12502 Conflicts=systemd-timesyncd.service
12504 to their unit files to take over and replace systemd's NTP
12505 default functionality.
12507 * systemd-sysusers gained a new line type "r" for configuring
12508 which UID/GID ranges to allocate system users/groups
12509 from. Lines of type "u" may now add an additional column
12510 that specifies the home directory for the system user to be
12511 created. Also, systemd-sysusers may now optionally read user
12512 information from STDIN instead of a file. This is useful for
12513 invoking it from RPM preinst scriptlets that need to create
12514 users before the first RPM file is installed since these
12515 files might need to be owned by them. A new
12516 %sysusers_create_inline RPM macro has been introduced to do
12517 just that. systemd-sysusers now updates the shadow files as
12518 well as the user/group databases, which should enhance
12519 compatibility with certain tools like grpck.
12521 * A number of bus APIs of PID 1 now optionally consult polkit to
12522 permit access for otherwise unprivileged clients under certain
12523 conditions. Note that this currently doesn't support
12524 interactive authentication yet, but this is expected to be
12525 added eventually, too.
12527 * /etc/machine-info now has new fields for configuring the
12528 deployment environment of the machine, as well as the
12529 location of the machine. hostnamectl has been updated with
12530 new command to update these fields.
12532 * systemd-timesyncd has been updated to automatically acquire
12533 NTP server information from systemd-networkd, which might
12534 have been discovered via DHCP.
12536 * systemd-resolved now includes a caching DNS stub resolver
12537 and a complete LLMNR name resolution implementation. A new
12538 NSS module "nss-resolve" has been added which can be used
12539 instead of glibc's own "nss-dns" to resolve hostnames via
12540 systemd-resolved. Hostnames, addresses and arbitrary RRs may
12541 be resolved via systemd-resolved D-Bus APIs. In contrast to
12542 the glibc internal resolver systemd-resolved is aware of
12543 multi-homed system, and keeps DNS server and caches separate
12544 and per-interface. Queries are sent simultaneously on all
12545 interfaces that have DNS servers configured, in order to
12546 properly handle VPNs and local LANs which might resolve
12547 separate sets of domain names. systemd-resolved may acquire
12548 DNS server information from systemd-networkd automatically,
12549 which in turn might have discovered them via DHCP. A tool
12550 "systemd-resolve-host" has been added that may be used to
12551 query the DNS logic in resolved. systemd-resolved implements
12552 IDNA and automatically uses IDNA or UTF-8 encoding depending
12553 on whether classic DNS or LLMNR is used as transport. In the
12554 next releases we intend to add a DNSSEC and mDNS/DNS-SD
12555 implementation to systemd-resolved.
12557 * A new NSS module nss-mymachines has been added, that
12558 automatically resolves the names of all local registered
12559 containers to their respective IP addresses.
12561 * A new client tool "networkctl" for systemd-networkd has been
12562 added. It currently is entirely passive and will query
12563 networking configuration from udev, rtnetlink and networkd,
12564 and present it to the user in a very friendly
12565 way. Eventually, we hope to extend it to become a full
12566 control utility for networkd.
12568 * .socket units gained a new DeferAcceptSec= setting that
12569 controls the kernels' TCP_DEFER_ACCEPT sockopt for
12570 TCP. Similarly, support for controlling TCP keep-alive
12571 settings has been added (KeepAliveTimeSec=,
12572 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
12573 turning off Nagle's algorithm on TCP has been added
12576 * logind learned a new session type "web", for use in projects
12577 like Cockpit which register web clients as PAM sessions.
12579 * timer units with at least one OnCalendar= setting will now
12580 be started only after time-sync.target has been
12581 reached. This way they will not elapse before the system
12582 clock has been corrected by a local NTP client or
12583 similar. This is particular useful on RTC-less embedded
12584 machines, that come up with an invalid system clock.
12586 * systemd-nspawn's --network-veth= switch should now result in
12587 stable MAC addresses for both the outer and the inner side
12590 * systemd-nspawn gained a new --volatile= switch for running
12591 container instances with /etc or /var unpopulated.
12593 * The kdbus client code has been updated to use the new Linux
12594 3.17 memfd subsystem instead of the old kdbus-specific one.
12596 * systemd-networkd's DHCP client and server now support
12597 FORCERENEW. There are also new configuration options to
12598 configure the vendor client identifier and broadcast mode
12601 * systemd will no longer inform the kernel about the current
12602 timezone, as this is necessarily incorrect and racy as the
12603 kernel has no understanding of DST and similar
12604 concepts. This hence means FAT timestamps will be always
12605 considered UTC, similar to what Android is already
12606 doing. Also, when the RTC is configured to the local time
12607 (rather than UTC) systemd will never synchronize back to it,
12608 as this might confuse Windows at a later boot.
12610 * systemd-analyze gained a new command "verify" for offline
12611 validation of unit files.
12613 * systemd-networkd gained support for a couple of additional
12614 settings for bonding networking setups. Also, the metric for
12615 statically configured routes may now be configured. For
12616 network interfaces where this is appropriate the peer IP
12617 address may now be configured.
12619 * systemd-networkd's DHCP client will no longer request
12620 broadcasting by default, as this tripped up some networks.
12621 For hardware where broadcast is required the feature should
12622 be switched back on using RequestBroadcast=yes.
12624 * systemd-networkd will now set up IPv4LL addresses (when
12625 enabled) even if DHCP is configured successfully.
12627 * udev will now default to respect network device names given
12628 by the kernel when the kernel indicates that these are
12629 predictable. This behavior can be tweaked by changing
12630 NamePolicy= in the relevant .link file.
12632 * A new library systemd-terminal has been added that
12633 implements full TTY stream parsing and rendering. This
12634 library is supposed to be used later on for implementing a
12635 full userspace VT subsystem, replacing the current kernel
12638 * A new tool systemd-journal-upload has been added to push
12639 journal data to a remote system running
12640 systemd-journal-remote.
12642 * journald will no longer forward all local data to another
12643 running syslog daemon. This change has been made because
12644 rsyslog (which appears to be the most commonly used syslog
12645 implementation these days) no longer makes use of this, and
12646 instead pulls the data out of the journal on its own. Since
12647 forwarding the messages to a non-existent syslog server is
12648 more expensive than we assumed we have now turned this
12649 off. If you run a syslog server that is not a recent rsyslog
12650 version, you have to turn this option on again
12651 (ForwardToSyslog= in journald.conf).
12653 * journald now optionally supports the LZ4 compressor for
12654 larger journal fields. This compressor should perform much
12655 better than XZ which was the previous default.
12657 * machinectl now shows the IP addresses of local containers,
12658 if it knows them, plus the interface name of the container.
12660 * A new tool "systemd-escape" has been added that makes it
12661 easy to escape strings to build unit names and similar.
12663 * sd_notify() messages may now include a new ERRNO= field
12664 which is parsed and collected by systemd and shown among the
12665 "systemctl status" output for a service.
12667 * A new component "systemd-firstboot" has been added that
12668 queries the most basic systemd information (timezone,
12669 hostname, root password) interactively on first
12670 boot. Alternatively it may also be used to provision these
12671 things offline on OS images installed into directories.
12673 * The default sysctl.d/ snippets will now set
12675 net.ipv4.conf.default.promote_secondaries=1
12677 This has the benefit of no flushing secondary IP addresses
12678 when primary addresses are removed.
12680 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
12681 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
12682 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
12683 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
12684 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
12685 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
12686 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
12687 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
12688 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
12689 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
12690 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
12691 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
12692 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
12693 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
12694 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
12696 — Berlin, 2014-08-19
12700 * A new tool systemd-sysusers has been added. This tool
12701 creates system users and groups in /etc/passwd and
12702 /etc/group, based on static declarative system user/group
12703 definitions in /usr/lib/sysusers.d/. This is useful to
12704 enable factory resets and volatile systems that boot up with
12705 an empty /etc directory, and thus need system users and
12706 groups created during early boot. systemd now also ships
12707 with two default sysusers.d/ files for the most basic
12708 users and groups systemd and the core operating system
12711 * A new tmpfiles snippet has been added that rebuilds the
12712 essential files in /etc on boot, should they be missing.
12714 * A directive for ensuring automatic clean-up of
12715 /var/cache/man/ has been removed from the default
12716 configuration. This line should now be shipped by the man
12717 implementation. The necessary change has been made to the
12718 man-db implementation. Note that you need to update your man
12719 implementation to one that ships this line, otherwise no
12720 automatic clean-up of /var/cache/man will take place.
12722 * A new condition ConditionNeedsUpdate= has been added that
12723 may conditionalize services to only run when /etc or /var
12724 are "older" than the vendor operating system resources in
12725 /usr. This is useful for reconstructing or updating /etc
12726 after an offline update of /usr or a factory reset, on the
12727 next reboot. Services that want to run once after such an
12728 update or reset should use this condition and order
12729 themselves before the new systemd-update-done.service, which
12730 will mark the two directories as fully updated. A number of
12731 service files have been added making use of this, to rebuild
12732 the udev hardware database, the journald message catalog and
12733 dynamic loader cache (ldconfig). The systemd-sysusers tool
12734 described above also makes use of this now. With this in
12735 place it is now possible to start up a minimal operating
12736 system with /etc empty cleanly. For more information on the
12737 concepts involved see this recent blog story:
12739 https://0pointer.de/blog/projects/stateless.html
12741 * A new system group "input" has been introduced, and all
12742 input device nodes get this group assigned. This is useful
12743 for system-level software to get access to input devices. It
12744 complements what is already done for "audio" and "video".
12746 * systemd-networkd learnt minimal DHCPv4 server support in
12747 addition to the existing DHCPv4 client support. It also
12748 learnt DHCPv6 client and IPv6 Router Solicitation client
12749 support. The DHCPv4 client gained support for static routes
12750 passed in from the server. Note that the [DHCPv4] section
12751 known in older systemd-networkd versions has been renamed to
12752 [DHCP] and is now also used by the DHCPv6 client. Existing
12753 .network files using settings of this section should be
12754 updated, though compatibility is maintained. Optionally, the
12755 client hostname may now be sent to the DHCP server.
12757 * networkd gained support for vxlan virtual networks as well
12758 as tun/tap and dummy devices.
12760 * networkd gained support for automatic allocation of address
12761 ranges for interfaces from a system-wide pool of
12762 addresses. This is useful for dynamically managing a large
12763 number of interfaces with a single network configuration
12764 file. In particular this is useful to easily assign
12765 appropriate IP addresses to the veth links of a large number
12766 of nspawn instances.
12768 * RPM macros for processing sysusers, sysctl and binfmt
12769 drop-in snippets at package installation time have been
12772 * The /etc/os-release file should now be placed in
12773 /usr/lib/os-release. The old location is automatically
12774 created as symlink. /usr/lib is the more appropriate
12775 location of this file, since it shall actually describe the
12776 vendor operating system shipped in /usr, and not the
12777 configuration stored in /etc.
12779 * .mount units gained a new boolean SloppyOptions= setting
12780 that maps to mount(8)'s -s option which enables permissive
12781 parsing of unknown mount options.
12783 * tmpfiles learnt a new "L+" directive which creates a symlink
12784 but (unlike "L") deletes a pre-existing file first, should
12785 it already exist and not already be the correct
12786 symlink. Similarly, "b+", "c+" and "p+" directives have been
12787 added as well, which create block and character devices, as
12788 well as fifos in the filesystem, possibly removing any
12789 pre-existing files of different types.
12791 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
12792 'argument' field (which so far specified the source to
12793 symlink/copy the files from) is now optional. If omitted the
12794 same file os copied from /usr/share/factory/ suffixed by the
12795 full destination path. This is useful for populating /etc
12796 with essential files, by copying them from vendor defaults
12797 shipped in /usr/share/factory/etc.
12799 * A new command "systemctl preset-all" has been added that
12800 applies the service preset settings to all installed unit
12801 files. A new switch --preset-mode= has been added that
12802 controls whether only enable or only disable operations
12805 * A new command "systemctl is-system-running" has been added
12806 that allows checking the overall state of the system, for
12807 example whether it is fully up and running.
12809 * When the system boots up with an empty /etc, the equivalent
12810 to "systemctl preset-all" is executed during early boot, to
12811 make sure all default services are enabled after a factory
12814 * systemd now contains a minimal preset file that enables the
12815 most basic services systemd ships by default.
12817 * Unit files' [Install] section gained a new DefaultInstance=
12818 field for defining the default instance to create if a
12819 template unit is enabled with no instance specified.
12821 * A new passive target cryptsetup-pre.target has been added
12822 that may be used by services that need to make they run and
12823 finish before the first LUKS cryptographic device is set up.
12825 * The /dev/loop-control and /dev/btrfs-control device nodes
12826 are now owned by the "disk" group by default, opening up
12827 access to this group.
12829 * systemd-coredump will now automatically generate a
12830 stack trace of all core dumps taking place on the system,
12831 based on elfutils' libdw library. This stack trace is logged
12834 * systemd-coredump may now optionally store coredumps directly
12835 on disk (in /var/lib/systemd/coredump, possibly compressed),
12836 instead of storing them unconditionally in the journal. This
12837 mode is the new default. A new configuration file
12838 /etc/systemd/coredump.conf has been added to configure this
12839 and other parameters of systemd-coredump.
12841 * coredumpctl gained a new "info" verb to show details about a
12842 specific coredump. A new switch "-1" has also been added
12843 that makes sure to only show information about the most
12844 recent entry instead of all entries. Also, as the tool is
12845 generally useful now the "systemd-" prefix of the binary
12846 name has been removed. Distributions that want to maintain
12847 compatibility with the old name should add a symlink from
12848 the old name to the new name.
12850 * journald's SplitMode= now defaults to "uid". This makes sure
12851 that unprivileged users can access their own coredumps with
12852 coredumpctl without restrictions.
12854 * New kernel command line options "systemd.wants=" (for
12855 pulling an additional unit during boot), "systemd.mask="
12856 (for masking a specific unit for the boot), and
12857 "systemd.debug-shell" (for enabling the debug shell on tty9)
12858 have been added. This is implemented in the new generator
12859 "systemd-debug-generator".
12861 * systemd-nspawn will now by default filter a couple of
12862 syscalls for containers, among them those required for
12863 kernel module loading, direct x86 IO port access, swap
12864 management, and kexec. Most importantly though
12865 open_by_handle_at() is now prohibited for containers,
12866 closing a hole similar to a recently discussed vulnerability
12867 in docker regarding access to files on file hierarchies the
12868 container should normally not have access to. Note that, for
12869 nspawn, we generally make no security claims anyway (and
12870 this is explicitly documented in the man page), so this is
12871 just a fix for one of the most obvious problems.
12873 * A new man page file-hierarchy(7) has been added that
12874 contains a minimized, modernized version of the file system
12875 layout systemd expects, similar in style to the FHS
12876 specification or hier(5). A new tool systemd-path(1) has
12877 been added to query many of these paths for the local
12880 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
12881 longer done. Since the directory now has a per-user size
12882 limit, and is cleaned on logout this appears unnecessary,
12883 in particular since this now brings the lifecycle of this
12884 directory closer in line with how IPC objects are handled.
12886 * systemd.pc now exports a number of additional directories,
12887 including $libdir (which is useful to identify the library
12888 path for the primary architecture of the system), and a
12889 couple of drop-in directories.
12891 * udev's predictable network interface names now use the dev_port
12892 sysfs attribute, introduced in linux 3.15 instead of dev_id to
12893 distinguish between ports of the same PCI function. dev_id should
12894 only be used for ports using the same HW address, hence the need
12897 * machined has been updated to export the OS version of a
12898 container (read from /etc/os-release and
12899 /usr/lib/os-release) on the bus. This is now shown in
12900 "machinectl status" for a machine.
12902 * A new service setting RestartForceExitStatus= has been
12903 added. If configured to a set of exit signals or process
12904 return values, the service will be restarted when the main
12905 daemon process exits with any of them, regardless of the
12908 * systemctl's -H switch for connecting to remote systemd
12909 machines has been extended so that it may be used to
12910 directly connect to a specific container on the
12911 host. "systemctl -H root@foobar:waldi" will now connect as
12912 user "root" to host "foobar", and then proceed directly to
12913 the container named "waldi". Note that currently you have to
12914 authenticate as user "root" for this to work, as entering
12915 containers is a privileged operation.
12917 Contributions from: Andreas Henriksson, Benjamin Steinwender,
12918 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
12919 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
12920 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
12921 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
12922 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
12923 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
12924 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
12925 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
12926 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
12927 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
12928 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
12930 — Berlin, 2014-07-03
12934 * As an experimental feature, udev now tries to lock the
12935 disk device node (flock(LOCK_SH|LOCK_NB)) while it
12936 executes events for the disk or any of its partitions.
12937 Applications like partitioning programs can lock the
12938 disk device node (flock(LOCK_EX)) and claim temporary
12939 device ownership that way; udev will entirely skip all event
12940 handling for this disk and its partitions. If the disk
12941 was opened for writing, the close will trigger a partition
12942 table rescan in udev's "watch" facility, and if needed
12943 synthesize "change" events for the disk and all its partitions.
12944 This is now unconditionally enabled, and if it turns out to
12945 cause major problems, we might turn it on only for specific
12946 devices, or might need to disable it entirely. Device Mapper
12947 devices are excluded from this logic.
12949 * We temporarily dropped the "-l" switch for fsck invocations,
12950 since they collide with the flock() logic above. util-linux
12951 upstream has been changed already to avoid this conflict,
12952 and we will re-add "-l" as soon as util-linux with this
12953 change has been released.
12955 * The dependency on libattr has been removed. Since a long
12956 time, the extended attribute calls have moved to glibc, and
12957 libattr is thus unnecessary.
12959 * Virtualization detection works without privileges now. This
12960 means the systemd-detect-virt binary no longer requires
12961 CAP_SYS_PTRACE file capabilities, and our daemons can run
12962 with fewer privileges.
12964 * systemd-networkd now runs under its own "systemd-network"
12965 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
12966 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
12967 loses the ability to write to files owned by root this way.
12969 * Similarly, systemd-resolved now runs under its own
12970 "systemd-resolve" user with no capabilities remaining.
12972 * Similarly, systemd-bus-proxyd now runs under its own
12973 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
12975 * systemd-networkd gained support for setting up "veth"
12976 virtual Ethernet devices for container connectivity, as well
12977 as GRE and VTI tunnels.
12979 * systemd-networkd will no longer automatically attempt to
12980 manually load kernel modules necessary for certain tunnel
12981 transports. Instead, it is assumed the kernel loads them
12982 automatically when required. This only works correctly on
12983 very new kernels. On older kernels, please consider adding
12984 the kernel modules to /etc/modules-load.d/ as a work-around.
12986 * The resolv.conf file systemd-resolved generates has been
12987 moved to /run/systemd/resolve/. If you have a symlink from
12988 /etc/resolv.conf, it might be necessary to correct it.
12990 * Two new service settings, ProtectHome= and ProtectSystem=,
12991 have been added. When enabled, they will make the user data
12992 (such as /home) inaccessible or read-only and the system
12993 (such as /usr) read-only, for specific services. This allows
12994 very light-weight per-service sandboxing to avoid
12995 modifications of user data or system files from
12996 services. These two new switches have been enabled for all
12997 of systemd's long-running services, where appropriate.
12999 * Socket units gained new SocketUser= and SocketGroup=
13000 settings to set the owner user and group of AF_UNIX sockets
13001 and FIFOs in the file system.
13003 * Socket units gained a new RemoveOnStop= setting. If enabled,
13004 all FIFOS and sockets in the file system will be removed
13005 when the specific socket unit is stopped.
13007 * Socket units gained a new Symlinks= setting. It takes a list
13008 of symlinks to create to file system sockets or FIFOs
13009 created by the specific Unix sockets. This is useful to
13010 manage symlinks to socket nodes with the same lifecycle as
13013 * The /dev/log socket and /dev/initctl FIFO have been moved to
13014 /run, and have been replaced by symlinks. This allows
13015 connecting to these facilities even if PrivateDevices=yes is
13016 used for a service (which makes /dev/log itself unavailable,
13017 but /run is left). This also has the benefit of ensuring
13018 that /dev only contains device nodes, directories and
13019 symlinks, and nothing else.
13021 * sd-daemon gained two new calls sd_pid_notify() and
13022 sd_pid_notifyf(). They are similar to sd_notify() and
13023 sd_notifyf(), but allow overriding of the source PID of
13024 notification messages if permissions permit this. This is
13025 useful to send notify messages on behalf of a different
13026 process (for example, the parent process). The
13027 systemd-notify tool has been updated to make use of this
13028 when sending messages (so that notification messages now
13029 originate from the shell script invoking systemd-notify and
13030 not the systemd-notify process itself. This should minimize
13031 a race where systemd fails to associate notification
13032 messages to services when the originating process already
13035 * A new "on-abnormal" setting for Restart= has been added. If
13036 set, it will result in automatic restarts on all "abnormal"
13037 reasons for a process to exit, which includes unclean
13038 signals, core dumps, timeouts and watchdog timeouts, but
13039 does not include clean and unclean exit codes or clean
13040 signals. Restart=on-abnormal is an alternative for
13041 Restart=on-failure for services that shall be able to
13042 terminate and avoid restarts on certain errors, by
13043 indicating so with an unclean exit code. Restart=on-failure
13044 or Restart=on-abnormal is now the recommended setting for
13045 all long-running services.
13047 * If the InaccessibleDirectories= service setting points to a
13048 mount point (or if there are any submounts contained within
13049 it), it is now attempted to completely unmount it, to make
13050 the file systems truly unavailable for the respective
13053 * The ReadOnlyDirectories= service setting and
13054 systemd-nspawn's --read-only parameter are now recursively
13055 applied to all submounts, too.
13057 * Mount units may now be created transiently via the bus APIs.
13059 * The support for SysV and LSB init scripts has been removed
13060 from the systemd daemon itself. Instead, it is now
13061 implemented as a generator that creates native systemd units
13062 from these scripts when needed. This enables us to remove a
13063 substantial amount of legacy code from PID 1, following the
13064 fact that many distributions only ship a very small number
13065 of LSB/SysV init scripts nowadays.
13067 * Privileged Xen (dom0) domains are not considered
13068 virtualization anymore by the virtualization detection
13069 logic. After all, they generally have unrestricted access to
13070 the hardware and usually are used to manage the unprivileged
13073 * systemd-tmpfiles gained a new "C" line type, for copying
13074 files or entire directories.
13076 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
13077 lines. So far, they have been non-globbing versions of the
13078 latter, and have thus been redundant. In future, it is
13079 recommended to only use "z". "m" has hence been removed
13080 from the documentation, even though it stays supported.
13082 * A tmpfiles snippet to recreate the most basic structure in
13083 /var has been added. This is enough to create the /var/run →
13084 /run symlink and create a couple of structural
13085 directories. This allows systems to boot up with an empty or
13086 volatile /var. Of course, while with this change, the core OS
13087 now is capable with dealing with a volatile /var, not all
13088 user services are ready for it. However, we hope that sooner
13089 or later, many service daemons will be changed upstream so
13090 that they are able to automatically create their necessary
13091 directories in /var at boot, should they be missing. This is
13092 the first step to allow state-less systems that only require
13093 the vendor image for /usr to boot.
13095 * systemd-nspawn has gained a new --tmpfs= switch to mount an
13096 empty tmpfs instance to a specific directory. This is
13097 particularly useful for making use of the automatic
13098 reconstruction of /var (see above), by passing --tmpfs=/var.
13100 * Access modes specified in tmpfiles snippets may now be
13101 prefixed with "~", which indicates that they shall be masked
13102 by whether the existing file or directory is currently
13103 writable, readable or executable at all. Also, if specified,
13104 the sgid/suid/sticky bits will be masked for all
13107 * A new passive target unit "network-pre.target" has been
13108 added which is useful for services that shall run before any
13109 network is configured, for example firewall scripts.
13111 * The "floppy" group that previously owned the /dev/fd*
13112 devices is no longer used. The "disk" group is now used
13113 instead. Distributions should probably deprecate usage of
13116 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
13117 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
13118 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
13119 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
13120 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
13121 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
13124 — Berlin, 2014-06-11
13128 * A new "systemd-timesyncd" daemon has been added for
13129 synchronizing the system clock across the network. It
13130 implements an SNTP client. In contrast to NTP
13131 implementations such as chrony or the NTP reference server,
13132 this only implements a client side, and does not bother with
13133 the full NTP complexity, focusing only on querying time from
13134 one remote server and synchronizing the local clock to
13135 it. Unless you intend to serve NTP to networked clients or
13136 want to connect to local hardware clocks, this simple NTP
13137 client should be more than appropriate for most
13138 installations. The daemon runs with minimal privileges, and
13139 has been hooked up with networkd to only operate when
13140 network connectivity is available. The daemon saves the
13141 current clock to disk every time a new NTP sync has been
13142 acquired, and uses this to possibly correct the system clock
13143 early at bootup, in order to accommodate for systems that
13144 lack an RTC such as the Raspberry Pi and embedded devices,
13145 and to make sure that time monotonically progresses on these
13146 systems, even if it is not always correct. To make use of
13147 this daemon, a new system user and group "systemd-timesync"
13148 needs to be created on installation of systemd.
13150 * The queue "seqnum" interface of libudev has been disabled, as
13151 it was generally incompatible with device namespacing as
13152 sequence numbers of devices go "missing" if the devices are
13153 part of a different namespace.
13155 * "systemctl list-timers" and "systemctl list-sockets" gained
13156 a --recursive switch for showing units of these types also
13157 for all local containers, similar in style to the already
13158 supported --recursive switch for "systemctl list-units".
13160 * A new RebootArgument= setting has been added for service
13161 units, which may be used to specify a kernel reboot argument
13162 to use when triggering reboots with StartLimitAction=.
13164 * A new FailureAction= setting has been added for service
13165 units which may be used to specify an operation to trigger
13166 when a service fails. This works similarly to
13167 StartLimitAction=, but unlike it, controls what is done
13168 immediately rather than only after several attempts to
13169 restart the service in question.
13171 * hostnamed got updated to also expose the kernel name,
13172 release, and version on the bus. This is useful for
13173 executing commands like hostnamectl with the -H switch.
13174 systemd-analyze makes use of this to properly display
13175 details when running non-locally.
13177 * The bootchart tool can now show cgroup information in the
13178 graphs it generates.
13180 * The CFS CPU quota cgroup attribute is now exposed for
13181 services. The new CPUQuota= switch has been added for this
13182 which takes a percentage value. Setting this will have the
13183 result that a service may never get more CPU time than the
13184 specified percentage, even if the machine is otherwise idle.
13186 * systemd-networkd learned IPIP and SIT tunnel support.
13188 * LSB init scripts exposing a dependency on $network will now
13189 get a dependency on network-online.target rather than simply
13190 network.target. This should bring LSB handling closer to
13191 what it was on SysV systems.
13193 * A new fsck.repair= kernel option has been added to control
13194 how fsck shall deal with unclean file systems at boot.
13196 * The (.ini) configuration file parser will now silently ignore
13197 sections whose names begin with "X-". This may be used to maintain
13198 application-specific extension sections in unit files.
13200 * machined gained a new API to query the IP addresses of
13201 registered containers. "machinectl status" has been updated
13202 to show these addresses in its output.
13204 * A new call sd_uid_get_display() has been added to the
13205 sd-login APIs for querying the "primary" session of a
13206 user. The "primary" session of the user is elected from the
13207 user's sessions and generally a graphical session is
13208 preferred over a text one.
13210 * A minimal systemd-resolved daemon has been added. It
13211 currently simply acts as a companion to systemd-networkd and
13212 manages resolv.conf based on per-interface DNS
13213 configuration, possibly supplied via DHCP. In the long run
13214 we hope to extend this into a local DNSSEC enabled DNS and
13217 * The systemd-networkd-wait-online tool is now enabled by
13218 default. It will delay network-online.target until a network
13219 connection has been configured. The tool primarily integrates
13220 with networkd, but will also make a best effort to make sense
13221 of network configuration performed in some other way.
13223 * Two new service options StartupCPUShares= and
13224 StartupBlockIOWeight= have been added that work similarly to
13225 CPUShares= and BlockIOWeight= however only apply during
13226 system startup. This is useful to prioritize certain services
13227 differently during bootup than during normal runtime.
13229 * hostnamed has been changed to prefer the statically
13230 configured hostname in /etc/hostname (unless set to
13231 'localhost' or empty) over any dynamic one supplied by
13232 dhcp. With this change, the rules for picking the hostname
13233 match more closely the rules of other configuration settings
13234 where the local administrator's configuration in /etc always
13235 overrides any other settings.
13237 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
13238 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
13239 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
13240 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
13241 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
13242 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
13243 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
13244 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
13245 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
13246 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
13247 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
13248 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
13249 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
13250 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
13251 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
13252 Lindskog, WaLyong Cho, Will Woods, Zbigniew
13255 — Beijing, 2014-05-28
13259 * When restoring the screen brightness at boot, stay away from
13260 the darkest setting or from the lowest 5% of the available
13261 range, depending on which is the larger value of both. This
13262 should effectively protect the user from rebooting into a
13263 black screen, should the brightness have been set to minimum
13266 * sd-login gained a new sd_machine_get_class() call to
13267 determine the class ("vm" or "container") of a machine
13268 registered with machined.
13270 * sd-login gained new calls
13271 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
13272 to query the identity of the peer of a local AF_UNIX
13273 connection. They operate similarly to their sd_pid_get_xyz()
13276 * PID 1 will now maintain a system-wide system state engine
13277 with the states "starting", "running", "degraded",
13278 "maintenance", "stopping". These states are bound to system
13279 startup, normal runtime, runtime with at least one failed
13280 service, rescue/emergency mode and system shutdown. This
13281 state is shown in the "systemctl status" output when no unit
13282 name is passed. It is useful to determine system state, in
13283 particularly when doing so for many systems or containers at
13286 * A new command "list-machines" has been added to "systemctl"
13287 that lists all local OS containers and shows their system
13288 state (see above), if systemd runs inside of them.
13290 * systemctl gained a new "-r" switch to recursively enumerate
13291 units on all local containers, when used with the
13292 "list-unit" command (which is the default one that is
13293 executed when no parameters are specified).
13295 * The GPT automatic partition discovery logic will now honour
13296 two GPT partition flags: one may be set on a partition to
13297 cause it to be mounted read-only, and the other may be set
13298 on a partition to ignore it during automatic discovery.
13300 * Two new GPT type UUIDs have been added for automatic root
13301 partition discovery, for 32-bit and 64-bit ARM. This is not
13302 particularly useful for discovering the root directory on
13303 these architectures during bare-metal boots (since UEFI is
13304 not common there), but still very useful to allow booting of
13305 ARM disk images in nspawn with the -i option.
13307 * MAC addresses of interfaces created with nspawn's
13308 --network-interface= switch will now be generated from the
13309 machine name, and thus be stable between multiple invocations
13312 * logind will now automatically remove all IPC objects owned
13313 by a user if she or he fully logs out. This makes sure that
13314 users who are logged out cannot continue to consume IPC
13315 resources. This covers SysV memory, semaphores and message
13316 queues as well as POSIX shared memory and message
13317 queues. Traditionally, SysV and POSIX IPC had no lifecycle
13318 limits. With this functionality, that is corrected. This may
13319 be turned off by using the RemoveIPC= switch of logind.conf.
13321 * The systemd-machine-id-setup and tmpfiles tools gained a
13322 --root= switch to operate on a specific root directory,
13325 * journald can now forward logged messages to the TTYs of all
13326 logged in users ("wall"). This is the default for all
13327 emergency messages now.
13329 * A new tool systemd-journal-remote has been added to stream
13330 journal log messages across the network.
13332 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
13333 controller trees are mounted into it. Note that the
13334 directories mounted beneath it are not read-only. This is a
13335 security measure and is particularly useful because glibc
13336 actually includes a search logic to pick any tmpfs it can
13337 find to implement shm_open() if /dev/shm is not available
13338 (which it might very well be in namespaced setups).
13340 * machinectl gained a new "poweroff" command to cleanly power
13341 down a local OS container.
13343 * The PrivateDevices= unit file setting will now also drop the
13344 CAP_MKNOD capability from the capability bound set, and
13345 imply DevicePolicy=closed.
13347 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
13348 comprehensively on all long-running systemd services where
13349 this is appropriate.
13351 * systemd-udevd will now run in a disassociated mount
13352 namespace. To mount directories from udev rules, make sure to
13353 pull in mount units via SYSTEMD_WANTS properties.
13355 * The kdbus support gained support for uploading policy into
13356 the kernel. sd-bus gained support for creating "monitoring"
13357 connections that can eavesdrop into all bus communication
13358 for debugging purposes.
13360 * Timestamps may now be specified in seconds since the UNIX
13361 epoch Jan 1st, 1970 by specifying "@" followed by the value
13364 * Native tcpwrap support in systemd has been removed. tcpwrap
13365 is old code, not really maintained anymore and has serious
13366 shortcomings, and better options such as firewalls
13367 exist. For setups that require tcpwrap usage, please
13368 consider invoking your socket-activated service via tcpd,
13369 like on traditional inetd.
13371 * A new system.conf configuration option
13372 DefaultTimerAccuracySec= has been added that controls the
13373 default AccuracySec= setting of .timer units.
13375 * Timer units gained a new WakeSystem= switch. If enabled,
13376 timers configured this way will cause the system to resume
13377 from system suspend (if the system supports that, which most
13380 * Timer units gained a new Persistent= switch. If enabled,
13381 timers configured this way will save to disk when they have
13382 been last triggered. This information is then used on next
13383 reboot to possible execute overdue timer events, that
13384 could not take place because the system was powered off.
13385 This enables simple anacron-like behaviour for timer units.
13387 * systemctl's "list-timers" will now also list the time a
13388 timer unit was last triggered in addition to the next time
13389 it will be triggered.
13391 * systemd-networkd will now assign predictable IPv4LL
13392 addresses to its local interfaces.
13394 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
13395 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
13396 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
13397 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
13398 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
13399 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
13400 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
13401 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
13404 — Berlin, 2014-03-25
13408 * A new unit file setting RestrictAddressFamilies= has been
13409 added to restrict which socket address families unit
13410 processes gain access to. This takes address family names
13411 like "AF_INET" or "AF_UNIX", and is useful to minimize the
13412 attack surface of services via exotic protocol stacks. This
13413 is built on seccomp system call filters.
13415 * Two new unit file settings RuntimeDirectory= and
13416 RuntimeDirectoryMode= have been added that may be used to
13417 manage a per-daemon runtime directories below /run. This is
13418 an alternative for setting up directory permissions with
13419 tmpfiles snippets, and has the advantage that the runtime
13420 directory's lifetime is bound to the daemon runtime and that
13421 the daemon starts up with an empty directory each time. This
13422 is particularly useful when writing services that drop
13423 privileges using the User= or Group= setting.
13425 * The DeviceAllow= unit setting now supports globbing for
13426 matching against device group names.
13428 * The systemd configuration file system.conf gained new
13429 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
13430 DefaultMemoryAccounting= to globally turn on/off accounting
13431 for specific resources (cgroups) for all units. These
13432 settings may still be overridden individually in each unit
13435 * systemd-gpt-auto-generator is now able to discover /srv and
13436 root partitions in addition to /home and swap partitions. It
13437 also supports LUKS-encrypted partitions now. With this in
13438 place, automatic discovery of partitions to mount following
13439 the Discoverable Partitions Specification
13440 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
13441 is now a lot more complete. This allows booting without
13442 /etc/fstab and without root= on the kernel command line on
13443 systems prepared appropriately.
13445 * systemd-nspawn gained a new --image= switch which allows
13446 booting up disk images and Linux installations on any block
13447 device that follow the Discoverable Partitions Specification
13448 (see above). This means that installations made with
13449 appropriately updated installers may now be started and
13450 deployed using container managers, completely
13451 unmodified. (We hope that libvirt-lxc will add support for
13452 this feature soon, too.)
13454 * systemd-nspawn gained a new --network-macvlan= setting to
13455 set up a private macvlan interface for the
13456 container. Similarly, systemd-networkd gained a new
13457 Kind=macvlan setting in .netdev files.
13459 * systemd-networkd now supports configuring local addresses
13462 * A new tool systemd-network-wait-online has been added to
13463 synchronously wait for network connectivity using
13466 * The sd-bus.h bus API gained a new sd_bus_track object for
13467 tracking the lifecycle of bus peers. Note that sd-bus.h is
13468 still not a public API though (unless you specify
13469 --enable-kdbus on the configure command line, which however
13470 voids your warranty and you get no API stability guarantee).
13472 * The $XDG_RUNTIME_DIR runtime directories for each user are
13473 now individual tmpfs instances, which has the benefit of
13474 introducing separate pools for each user, with individual
13475 size limits, and thus making sure that unprivileged clients
13476 can no longer negatively impact the system or other users by
13477 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
13478 RuntimeDirectorySize= has been introduced that allows
13479 controlling the default size limit for all users. It
13480 defaults to 10% of the available physical memory. This is no
13481 replacement for quotas on tmpfs though (which the kernel
13482 still does not support), as /dev/shm and /tmp are still
13483 shared resources used by both the system and unprivileged
13486 * logind will now automatically turn off automatic suspending
13487 on laptop lid close when more than one display is
13488 connected. This was previously expected to be implemented
13489 individually in desktop environments (such as GNOME),
13490 however has been added to logind now, in order to fix a
13491 boot-time race where a desktop environment might not have
13492 been started yet and thus not been able to take an inhibitor
13493 lock at the time where logind already suspends the system
13494 due to a closed lid.
13496 * logind will now wait at least 30s after each system
13497 suspend/resume cycle, and 3min after system boot before
13498 suspending the system due to a closed laptop lid. This
13499 should give USB docking stations and similar enough time to
13500 be probed and configured after system resume and boot in
13501 order to then act as suspend blocker.
13503 * systemd-run gained a new --property= setting which allows
13504 initialization of resource control properties (and others)
13505 for the created scope or service unit. Example: "systemd-run
13506 --property=BlockIOWeight=10 updatedb" may be used to run
13507 updatedb at a low block IO scheduling weight.
13509 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
13510 now also work in --scope mode.
13512 * When systemd is compiled with kdbus support, basic support
13513 for enforced policies is now in place. (Note that enabling
13514 kdbus still voids your warranty and no API compatibility
13515 promises are made.)
13517 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
13518 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
13519 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
13520 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
13521 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
13522 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
13523 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
13524 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
13525 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
13526 Zbigniew Jędrzejewski-Szmek
13528 — Berlin, 2014-03-12
13532 * systemd will now relabel /dev after loading the SMACK policy
13533 according to SMACK rules.
13535 * A new unit file option AppArmorProfile= has been added to
13536 set the AppArmor profile for the processes of a unit.
13538 * A new condition check ConditionArchitecture= has been added
13539 to conditionalize units based on the system architecture, as
13540 reported by uname()'s "machine" field.
13542 * systemd-networkd now supports matching on the system
13543 virtualization, architecture, kernel command line, hostname
13546 * logind is now a lot more aggressive when suspending the
13547 machine due to a closed laptop lid. Instead of acting only
13548 on the lid close action, it will continuously watch the lid
13549 status and act on it. This is useful for laptops where the
13550 power button is on the outside of the chassis so that it can
13551 be reached without opening the lid (such as the Lenovo
13552 Yoga). On those machines, logind will now immediately
13553 re-suspend the machine if the power button has been
13554 accidentally pressed while the laptop was suspended and in a
13555 backpack or similar.
13557 * logind will now watch SW_DOCK switches and inhibit reaction
13558 to the lid switch if it is pressed. This means that logind
13559 will not suspend the machine anymore if the lid is closed
13560 and the system is docked, if the laptop supports SW_DOCK
13561 notifications via the input layer. Note that ACPI docking
13562 stations do not generate this currently. Also note that this
13563 logic is usually not fully sufficient and Desktop
13564 Environments should take a lid switch inhibitor lock when an
13565 external display is connected, as systemd will not watch
13568 * nspawn will now make use of the devices cgroup controller by
13569 default, and only permit creation of and access to the usual
13570 API device nodes like /dev/null or /dev/random, as well as
13571 access to (but not creation of) the pty devices.
13573 * We will now ship a default .network file for
13574 systemd-networkd that automatically configures DHCP for
13575 network interfaces created by nspawn's --network-veth or
13576 --network-bridge= switches.
13578 * systemd will now understand the usual M, K, G, T suffixes
13579 according to SI conventions (i.e. to the base 1000) when
13580 referring to throughput and hardware metrics. It will stay
13581 with IEC conventions (i.e. to the base 1024) for software
13582 metrics, according to what is customary according to
13583 Wikipedia. We explicitly document which base applies for
13584 each configuration option.
13586 * The DeviceAllow= setting in unit files now supports a syntax to
13587 allow-list an entire group of devices node majors at once, based on
13588 the /proc/devices listing. For example, with the string "char-pts",
13589 it is now possible to allow-list all current and future pseudo-TTYs
13592 * sd-event learned a new "post" event source. Event sources of
13593 this type are triggered by the dispatching of any event
13594 source of a type that is not "post". This is useful for
13595 implementing clean-up and check event sources that are
13596 triggered by other work being done in the program.
13598 * systemd-networkd is no longer statically enabled, but uses
13599 the usual [Install] sections so that it can be
13600 enabled/disabled using systemctl. It still is enabled by
13603 * When creating a veth interface pair with systemd-nspawn, the
13604 host side will now be prefixed with "vb-" if
13605 --network-bridge= is used, and with "ve-" if --network-veth
13606 is used. This way, it is easy to distinguish these cases on
13607 the host, for example to apply different configuration to
13608 them with systemd-networkd.
13610 * The compatibility libraries for libsystemd-journal.so,
13611 libsystem-id128.so, libsystemd-login.so and
13612 libsystemd-daemon.so do not make use of IFUNC
13613 anymore. Instead, we now build libsystemd.so multiple times
13614 under these alternative names. This means that the footprint
13615 is drastically increased, but given that these are
13616 transitional compatibility libraries, this should not matter
13617 much. This change has been made necessary to support the ARM
13618 platform for these compatibility libraries, as the ARM
13619 toolchain is not really at the same level as the toolchain
13620 for other architectures like x86 and does not support
13621 IFUNC. Please make sure to use --enable-compat-libs only
13622 during a transitional period!
13624 * The .include syntax has been deprecated and is not documented
13625 anymore. Drop-in files in .d directories should be used instead.
13627 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
13628 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
13629 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
13630 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
13631 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
13632 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
13633 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
13634 Zbigniew Jędrzejewski-Szmek
13636 — Berlin, 2014-02-24
13640 * A new component "systemd-networkd" has been added that can
13641 be used to configure local network interfaces statically or
13642 via DHCP. It is capable of bringing up bridges, VLANs, and
13643 bonding. Currently, no hook-ups for interactive network
13644 configuration are provided. Use this for your initrd,
13645 container, embedded, or server setup if you need a simple,
13646 yet powerful, network configuration solution. This
13647 configuration subsystem is quite nifty, as it allows wildcard
13648 hotplug matching in interfaces. For example, with a single
13649 configuration snippet, you can configure that all Ethernet
13650 interfaces showing up are automatically added to a bridge,
13651 or similar. It supports link-sensing and more.
13653 * A new tool "systemd-socket-proxyd" has been added which can
13654 act as a bidirectional proxy for TCP sockets. This is
13655 useful for adding socket activation support to services that
13656 do not actually support socket activation, including virtual
13657 machines and the like.
13659 * Add a new tool to save/restore rfkill state on
13662 * Save/restore state of keyboard backlights in addition to
13663 display backlights on shutdown/boot.
13665 * udev learned a new SECLABEL{} construct to label device
13666 nodes with a specific security label when they appear. For
13667 now, only SECLABEL{selinux} is supported, but the syntax is
13668 prepared for additional security frameworks.
13670 * udev gained a new scheme to configure link-level attributes
13671 from files in /etc/systemd/network/*.link. These files can
13672 match against MAC address, device path, driver name and type,
13673 and will apply attributes like the naming policy, link speed,
13674 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
13675 address assignment policy (randomized, …).
13677 * The configuration of network interface naming rules for
13678 "permanent interface names" has changed: a new NamePolicy=
13679 setting in the [Link] section of .link files determines the
13680 priority of possible naming schemes (onboard, slot, MAC,
13681 path). The default value of this setting is determined by
13682 /usr/lib/net/links/99-default.link. Old
13683 80-net-name-slot.rules udev configuration file has been
13684 removed, so local configuration overriding this file should
13685 be adapted to override 99-default.link instead.
13687 * When the User= switch is used in a unit file, also
13688 initialize $SHELL= based on the user database entry.
13690 * systemd no longer depends on libdbus. All communication is
13691 now done with sd-bus, systemd's low-level bus library
13694 * kdbus support has been added to PID 1 itself. When kdbus is
13695 enabled, this causes PID 1 to set up the system bus and
13696 enable support for a new ".busname" unit type that
13697 encapsulates bus name activation on kdbus. It works a little
13698 bit like ".socket" units, except for bus names. A new
13699 generator has been added that converts classic dbus1 service
13700 activation files automatically into native systemd .busname
13701 and .service units.
13703 * sd-bus: add a light-weight vtable implementation that allows
13704 defining objects on the bus with a simple static const
13705 vtable array of its methods, signals and properties.
13707 * systemd will not generate or install static dbus
13708 introspection data anymore to /usr/share/dbus-1/interfaces,
13709 as the precise format of these files is unclear, and
13710 nothing makes use of it.
13712 * A proxy daemon is now provided to proxy clients connecting
13713 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
13714 compatibility with classic D-Bus.
13716 * A bus driver implementation has been added that supports the
13717 classic D-Bus bus driver calls on kdbus, also for
13718 compatibility purposes.
13720 * A new API "sd-event.h" has been added that implements a
13721 minimal event loop API built around epoll. It provides a
13722 couple of features that direct epoll usage is lacking:
13723 prioritization of events, scales to large numbers of timer
13724 events, per-event timer slack (accuracy), system-wide
13725 coalescing of timer events, exit handlers, watchdog
13726 supervision support using systemd's sd_notify() API, child
13729 * A new API "sd-rntl.h" has been added that provides an API
13730 around the route netlink interface of the kernel, similar in
13731 style to "sd-bus.h".
13733 * A new API "sd-dhcp-client.h" has been added that provides a
13734 small DHCPv4 client-side implementation. This is used by
13735 "systemd-networkd".
13737 * There is a new kernel command line option
13738 "systemd.restore_state=0|1". When set to "0", none of the
13739 systemd tools will restore saved runtime state to hardware
13740 devices. More specifically, the rfkill and backlight states
13743 * The FsckPassNo= compatibility option in mount/service units
13744 has been removed. The fstab generator will now add the
13745 necessary dependencies automatically, and does not require
13746 PID1's support for that anymore.
13748 * journalctl gained a new switch, --list-boots, that lists
13749 recent boots with their times and boot IDs.
13751 * The various tools like systemctl, loginctl, timedatectl,
13752 busctl, systemd-run, … have gained a new switch "-M" to
13753 connect to a specific, local OS container (as direct
13754 connection, without requiring SSH). This works on any
13755 container that is registered with machined, such as those
13756 created by libvirt-lxc or nspawn.
13758 * systemd-run and systemd-analyze also gained support for "-H"
13759 to connect to remote hosts via SSH. This is particularly
13760 useful for systemd-run because it enables queuing of jobs
13761 onto remote systems.
13763 * machinectl gained a new command "login" to open a getty
13764 login in any local container. This works with any container
13765 that is registered with machined (such as those created by
13766 libvirt-lxc or nspawn), and which runs systemd inside.
13768 * machinectl gained a new "reboot" command that may be used to
13769 trigger a reboot on a specific container that is registered
13770 with machined. This works on any container that runs an init
13771 system of some kind.
13773 * systemctl gained a new "list-timers" command to print a nice
13774 listing of installed timer units with the times they elapse
13777 * Alternative reboot() parameters may now be specified on the
13778 "systemctl reboot" command line and are passed to the
13779 reboot() system call.
13781 * systemctl gained a new --job-mode= switch to configure the
13782 mode to queue a job with. This is a more generic version of
13783 --fail, --irreversible, and --ignore-dependencies, which are
13784 still available but not advertised anymore.
13786 * /etc/systemd/system.conf gained new settings to configure
13787 various default timeouts of units, as well as the default
13788 start limit interval and burst. These may still be overridden
13791 * PID1 will now export on the bus profile data of the security
13792 policy upload process (such as the SELinux policy upload to
13795 * journald: when forwarding logs to the console, include
13796 timestamps (following the setting in
13797 /sys/module/printk/parameters/time).
13799 * OnCalendar= in timer units now understands the special
13800 strings "yearly" and "annually". (Both are equivalent)
13802 * The accuracy of timer units is now configurable with the new
13803 AccuracySec= setting. It defaults to 1min.
13805 * A new dependency type JoinsNamespaceOf= has been added that
13806 allows running two services within the same /tmp and network
13807 namespace, if PrivateNetwork= or PrivateTmp= are used.
13809 * A new command "cat" has been added to systemctl. It outputs
13810 the original unit file of a unit, and concatenates the
13811 contents of additional "drop-in" unit file snippets, so that
13812 the full configuration is shown.
13814 * systemctl now supports globbing on the various "list-xyz"
13815 commands, like "list-units" or "list-sockets", as well as on
13816 those commands which take multiple unit names.
13818 * journalctl's --unit= switch gained support for globbing.
13820 * All systemd daemons now make use of the watchdog logic so
13821 that systemd automatically notices when they hang.
13823 * If the $container_ttys environment variable is set,
13824 getty-generator will automatically spawn a getty for each
13825 listed tty. This is useful for container managers to request
13826 login gettys to be spawned on as many ttys as needed.
13828 * %h, %s, %U specifier support is not available anymore when
13829 used in unit files for PID 1. This is because NSS calls are
13830 not safe from PID 1. They stay available for --user
13831 instances of systemd, and as special case for the root user.
13833 * loginctl gained a new "--no-legend" switch to turn off output
13834 of the legend text.
13836 * The "sd-login.h" API gained three new calls:
13837 sd_session_is_remote(), sd_session_get_remote_user(),
13838 sd_session_get_remote_host() to query information about
13841 * The udev hardware database now also carries vendor/product
13842 information of SDIO devices.
13844 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
13845 determine whether watchdog notifications are requested by
13846 the system manager.
13848 * Socket-activated per-connection services now include a
13849 short description of the connection parameters in the
13852 * tmpfiles gained a new "--boot" option. When this is not used,
13853 only lines where the command character is not suffixed with
13854 "!" are executed. When this option is specified, those
13855 options are executed too. This partitions tmpfiles
13856 directives into those that can be safely executed at any
13857 time, and those which should be run only at boot (for
13858 example, a line that creates /run/nologin).
13860 * A new API "sd-resolve.h" has been added which provides a simple
13861 asynchronous wrapper around glibc NSS hostname resolution
13862 calls, such as getaddrinfo(). In contrast to glibc's
13863 getaddrinfo_a(), it does not use signals. In contrast to most
13864 other asynchronous name resolution libraries, this one does
13865 not reimplement DNS, but reuses NSS, so that alternate
13866 hostname resolution systems continue to work, such as mDNS,
13867 LDAP, etc. This API is based on libasyncns, but it has been
13868 cleaned up for inclusion in systemd.
13870 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
13871 "sd-daemon.h" are no longer found in individual libraries
13872 libsystemd-journal.so, libsystemd-login.so,
13873 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
13874 merged them into a single library, libsystemd.so, which
13875 provides all symbols. The reason for this is cyclic
13876 dependencies, as these libraries tend to use each other's
13877 symbols. So far, we have managed to workaround that by linking
13878 a copy of a good part of our code into each of these
13879 libraries again and again, which, however, makes certain
13880 things hard to do, like sharing static variables. Also, it
13881 substantially increases footprint. With this change, there
13882 is only one library for the basic APIs systemd
13883 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
13884 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
13885 library as well, however are subject to the --enable-kdbus
13886 switch (see below). Note that "sd-dhcp-client.h" is not part
13887 of this library (this is because it only consumes, never
13888 provides, services of/to other APIs). To make the transition
13889 easy from the separate libraries to the unified one, we
13890 provide the --enable-compat-libs compile-time switch which
13891 will generate stub libraries that are compatible with the
13892 old ones but redirect all calls to the new one.
13894 * All of the kdbus logic and the new APIs "sd-bus.h",
13895 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
13896 and "sd-utf8.h" are compile-time optional via the
13897 "--enable-kdbus" switch, and they are not compiled in by
13898 default. To make use of kdbus, you have to explicitly enable
13899 the switch. Note however, that neither the kernel nor the
13900 userspace API for all of this is considered stable yet. We
13901 want to maintain the freedom to still change the APIs for
13902 now. By specifying this build-time switch, you acknowledge
13903 that you are aware of the instability of the current
13906 * Also, note that while kdbus is pretty much complete,
13907 it lacks one thing: proper policy support. This means you
13908 can build a fully working system with all features; however,
13909 it will be highly insecure. Policy support will be added in
13910 one of the next releases, at the same time that we will
13911 declare the APIs stable.
13913 * When the kernel command line argument "kdbus" is specified,
13914 systemd will automatically load the kdbus.ko kernel module. At
13915 this stage of development, it is only useful for testing kdbus
13916 and should not be used in production. Note: if "--enable-kdbus"
13917 is specified, and the kdbus.ko kernel module is available, and
13918 "kdbus" is added to the kernel command line, the entire system
13919 runs with kdbus instead of dbus-daemon, with the above mentioned
13920 problem of missing the system policy enforcement. Also a future
13921 version of kdbus.ko or a newer systemd will not be compatible with
13922 each other, and will unlikely be able to boot the machine if only
13923 one of them is updated.
13925 * systemctl gained a new "import-environment" command which
13926 uploads the caller's environment (or parts thereof) into the
13927 service manager so that it is inherited by services started
13928 by the manager. This is useful to upload variables like
13929 $DISPLAY into the user service manager.
13931 * A new PrivateDevices= switch has been added to service units
13932 which allows running a service with a namespaced /dev
13933 directory that does not contain any device nodes for
13934 physical devices. More specifically, it only includes devices
13935 such as /dev/null, /dev/urandom, and /dev/zero which are API
13938 * logind has been extended to support behaviour like VT
13939 switching on seats that do not support a VT. This makes
13940 multi-session available on seats that are not the first seat
13941 (seat0), and on systems where kernel support for VTs has
13942 been disabled at compile-time.
13944 * If a process holds a delay lock for system sleep or shutdown
13945 and fails to release it in time, we will now log its
13946 identity. This makes it easier to identify processes that
13947 cause slow suspends or power-offs.
13949 * When parsing /etc/crypttab, support for a new key-slot=
13950 option as supported by Debian is added. It allows indicating
13951 which LUKS slot to use on disk, speeding up key loading.
13953 * The sd_journal_sendv() API call has been checked and
13954 officially declared to be async-signal-safe so that it may
13955 be invoked from signal handlers for logging purposes.
13957 * Boot-time status output is now enabled automatically after a
13958 short timeout if boot does not progress, in order to give
13959 the user an indication what she or he is waiting for.
13961 * The boot-time output has been improved to show how much time
13962 remains until jobs expire.
13964 * The KillMode= switch in service units gained a new possible
13965 value "mixed". If set, and the unit is shut down, then the
13966 initial SIGTERM signal is sent only to the main daemon
13967 process, while the following SIGKILL signal is sent to
13968 all remaining processes of the service.
13970 * When a scope unit is registered, a new property "Controller"
13971 may be set. If set to a valid bus name, systemd will send a
13972 RequestStop() signal to this name when it would like to shut
13973 down the scope. This may be used to hook manager logic into
13974 the shutdown logic of scope units. Also, scope units may now
13975 be put in a special "abandoned" state, in which case the
13976 manager process which created them takes no further
13977 responsibilities for it.
13979 * When reading unit files, systemd will now verify
13980 the access mode of these files, and warn about certain
13981 suspicious combinations. This has been added to make it
13982 easier to track down packaging bugs where unit files are
13983 marked executable or world-writable.
13985 * systemd-nspawn gained a new "--setenv=" switch to set
13986 container-wide environment variables. The similar option in
13987 systemd-activate was renamed from "--environment=" to
13988 "--setenv=" for consistency.
13990 * systemd-nspawn has been updated to create a new kdbus domain
13991 for each container that is invoked, thus allowing each
13992 container to have its own set of system and user buses,
13993 independent of the host.
13995 * systemd-nspawn gained a new --drop-capability= switch to run
13996 the container with less capabilities than the default. Both
13997 --drop-capability= and --capability= now take the special
13998 string "all" for dropping or keeping all capabilities.
14000 * systemd-nspawn gained new switches for executing containers
14001 with specific SELinux labels set.
14003 * systemd-nspawn gained a new --quiet switch to not generate
14004 any additional output but the container's own console
14007 * systemd-nspawn gained a new --share-system switch to run a
14008 container without PID namespacing enabled.
14010 * systemd-nspawn gained a new --register= switch to control
14011 whether the container is registered with systemd-machined or
14012 not. This is useful for containers that do not run full
14013 OS images, but only specific apps.
14015 * systemd-nspawn gained a new --keep-unit which may be used
14016 when invoked as the only program from a service unit, and
14017 results in registration of the unit service itself in
14018 systemd-machined, instead of a newly opened scope unit.
14020 * systemd-nspawn gained a new --network-interface= switch for
14021 moving arbitrary interfaces to the container. The new
14022 --network-veth switch creates a virtual Ethernet connection
14023 between host and container. The new --network-bridge=
14024 switch then allows assigning the host side of this virtual
14025 Ethernet connection to a bridge device.
14027 * systemd-nspawn gained a new --personality= switch for
14028 setting the kernel personality for the container. This is
14029 useful when running a 32-bit container on a 64-bit host. A
14030 similar option Personality= is now also available for service
14033 * logind will now also track a "Desktop" identifier for each
14034 session which encodes the desktop environment of it. This is
14035 useful for desktop environments that want to identify
14036 multiple running sessions of itself easily.
14038 * A new SELinuxContext= setting for service units has been
14039 added that allows setting a specific SELinux execution
14040 context for a service.
14042 * Most systemd client tools will now honour $SYSTEMD_LESS for
14043 settings of the "less" pager. By default, these tools will
14044 override $LESS to allow certain operations to work, such as
14045 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
14046 influence this logic.
14048 * systemd's "seccomp" hook-up has been changed to make use of
14049 the libseccomp library instead of using its own
14050 implementation. This has benefits for portability among
14053 * For usage together with SystemCallFilter=, a new
14054 SystemCallErrorNumber= setting has been introduced that
14055 allows configuration of a system error number to be returned
14056 on filtered system calls, instead of immediately killing the
14057 process. Also, SystemCallArchitectures= has been added to
14058 limit access to system calls of a particular architecture
14059 (in order to turn off support for unused secondary
14060 architectures). There is also a global
14061 SystemCallArchitectures= setting in system.conf now to turn
14062 off support for non-native system calls system-wide.
14064 * systemd requires a kernel with a working name_to_handle_at(),
14065 please see the kernel config requirements in the README file.
14067 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
14068 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
14069 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
14070 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
14071 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
14072 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
14073 Elia Pinto, Florian Weimer, George McCollister, Goffredo
14074 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
14075 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
14076 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
14077 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
14078 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
14079 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
14080 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
14081 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
14082 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
14083 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
14084 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
14085 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
14086 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
14087 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
14088 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
14089 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
14090 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
14092 — Berlin, 2014-02-20
14096 * logind has gained support for facilitating privileged input
14097 and drm device access for unprivileged clients. This work is
14098 useful to allow Wayland display servers (and similar
14099 programs, such as kmscon) to run under the user's ID and
14100 access input and drm devices which are normally
14101 protected. When this is used (and the kernel is new enough)
14102 logind will "mute" IO on the file descriptors passed to
14103 Wayland as long as it is in the background and "unmute" it
14104 if it returns into the foreground. This allows secure
14105 session switching without allowing background sessions to
14106 eavesdrop on input and display data. This also introduces
14107 session switching support if VT support is turned off in the
14108 kernel, and on seats that are not seat0.
14110 * A new kernel command line option luks.options= is understood
14111 now which allows specifying LUKS options for usage for LUKS
14112 encrypted partitions specified with luks.uuid=.
14114 * tmpfiles.d(5) snippets may now use specifier expansion in
14115 path names. More specifically %m, %b, %H, %v, are now
14116 replaced by the local machine id, boot id, hostname, and
14117 kernel version number.
14119 * A new tmpfiles.d(5) command "m" has been introduced which
14120 may be used to change the owner/group/access mode of a file
14121 or directory if it exists, but do nothing if it does not.
14123 * This release removes high-level support for the
14124 MemorySoftLimit= cgroup setting. The underlying kernel
14125 cgroup attribute memory.soft_limit= is currently badly
14126 designed and likely to be removed from the kernel API in its
14127 current form, hence we should not expose it for now.
14129 * The memory.use_hierarchy cgroup attribute is now enabled for
14130 all cgroups systemd creates in the memory cgroup
14131 hierarchy. This option is likely to be come the built-in
14132 default in the kernel anyway, and the non-hierarchical mode
14133 never made much sense in the intrinsically hierarchical
14136 * A new field _SYSTEMD_SLICE= is logged along with all journal
14137 messages containing the slice a message was generated
14138 from. This is useful to allow easy per-customer filtering of
14139 logs among other things.
14141 * systemd-journald will no longer adjust the group of journal
14142 files it creates to the "systemd-journal" group. Instead we
14143 rely on the journal directory to be owned by the
14144 "systemd-journal" group, and its setgid bit set, so that the
14145 kernel file system layer will automatically enforce that
14146 journal files inherit this group assignment. The reason for
14147 this change is that we cannot allow NSS look-ups from
14148 journald which would be necessary to resolve
14149 "systemd-journal" to a numeric GID, because this might
14150 create deadlocks if NSS involves synchronous queries to
14151 other daemons (such as nscd, or sssd) which in turn are
14152 logging clients of journald and might block on it, which
14153 would then dead lock. A tmpfiles.d(5) snippet included in
14154 systemd will make sure the setgid bit and group are
14155 properly set on the journal directory if it exists on every
14156 boot. However, we recommend adjusting it manually after
14157 upgrades too (or from RPM scriptlets), so that the change is
14158 not delayed until next reboot.
14160 * Backlight and random seed files in /var/lib/ have moved into
14161 the /var/lib/systemd/ directory, in order to centralize all
14162 systemd generated files in one directory.
14164 * Boot time performance measurements (as displayed by
14165 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
14166 performance information if that's available to determine how
14167 much time BIOS and boot loader initialization required. With
14168 a sufficiently new BIOS you hence no longer need to boot
14169 with Gummiboot to get access to such information.
14171 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
14172 Cristian Rodríguez, Dave Reisner, David Herrmann, David
14173 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
14174 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
14175 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
14176 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
14177 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
14179 — Berlin, 2013-10-02
14183 * The Restart= option for services now understands a new
14184 on-watchdog setting, which will restart the service
14185 automatically if the service stops sending out watchdog keep
14186 alive messages (as configured with WatchdogSec=).
14188 * The getty generator (which is responsible for bringing up a
14189 getty on configured serial consoles) will no longer only
14190 start a getty on the primary kernel console but on all
14191 others, too. This makes the order in which console= is
14192 specified on the kernel command line less important.
14194 * libsystemd-logind gained a new sd_session_get_vt() call to
14195 retrieve the VT number of a session.
14197 * If the option "tries=0" is set for an entry of /etc/crypttab
14198 its passphrase is queried indefinitely instead of any
14199 maximum number of tries.
14201 * If a service with a configure PID file terminates its PID
14202 file will now be removed automatically if it still exists
14203 afterwards. This should put an end to stale PID files.
14205 * systemd-run will now also take relative binary path names
14206 for execution and no longer insists on absolute paths.
14208 * InaccessibleDirectories= and ReadOnlyDirectories= now take
14209 paths that are optionally prefixed with "-" to indicate that
14210 it should not be considered a failure if they do not exist.
14212 * journalctl -o (and similar commands) now understands a new
14213 output mode "short-precise", it is similar to "short" but
14214 shows timestamps with usec accuracy.
14216 * The option "discard" (as known from Debian) is now
14217 synonymous to "allow-discards" in /etc/crypttab. In fact,
14218 "discard" is preferred now (since it is easier to remember
14221 * Some licensing clean-ups were made, so that more code is now
14222 LGPL-2.1 licensed than before.
14224 * A minimal tool to save/restore the display backlight
14225 brightness across reboots has been added. It will store the
14226 backlight setting as late as possible at shutdown, and
14227 restore it as early as possible during reboot.
14229 * A logic to automatically discover and enable home and swap
14230 partitions on GPT disks has been added. With this in place
14231 /etc/fstab becomes optional for many setups as systemd can
14232 discover certain partitions located on the root disk
14233 automatically. Home partitions are recognized under their
14234 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
14235 partitions are recognized under their GPT type ID
14236 0657fd6da4ab43c484e50933c84b4f4f.
14238 * systemd will no longer pass any environment from the kernel
14239 or initrd to system services. If you want to set an
14240 environment for all services, do so via the kernel command
14241 line systemd.setenv= assignment.
14243 * The systemd-sysctl tool no longer natively reads the file
14244 /etc/sysctl.conf. If desired, the file should be symlinked
14245 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
14246 legacy support by a symlink rather than built-in code, it
14247 also makes the otherwise hidden order of application of the
14248 different files visible. (Note that this partly reverts to a
14249 pre-198 application order of sysctl knobs!)
14251 * The "systemctl set-log-level" and "systemctl dump" commands
14252 have been moved to systemd-analyze.
14254 * systemd-run learned the new --remain-after-exit switch,
14255 which causes the scope unit not to be cleaned up
14256 automatically after the process terminated.
14258 * tmpfiles learned a new --exclude-prefix= switch to exclude
14259 certain paths from operation.
14261 * journald will now automatically flush all messages to disk
14262 as soon as a message at the log level CRIT, ALERT or EMERG
14265 Contributions from: Andrew Cook, Brandon Philips, Christian
14266 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
14267 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
14268 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
14269 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
14270 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
14271 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
14272 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
14273 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
14274 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
14275 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
14276 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
14277 William Giokas, Zbigniew Jędrzejewski-Szmek
14279 — Berlin, 2013-09-13
14283 * The documentation has been updated to cover the various new
14284 concepts introduced with 205.
14286 * Unit files now understand the new %v specifier which
14287 resolves to the kernel version string as returned by "uname
14290 * systemctl now supports filtering the unit list output by
14291 load state, active state and sub state, using the new
14292 --state= parameter.
14294 * "systemctl status" will now show the results of the
14295 condition checks (like ConditionPathExists= and similar) of
14296 the last start attempts of the unit. They are also logged to
14299 * "journalctl -b" may now be used to look for boot output of a
14300 specific boot. Try "journalctl -b -1" for the previous boot,
14301 but the syntax is substantially more powerful.
14303 * "journalctl --show-cursor" has been added which prints the
14304 cursor string the last shown log line. This may then be used
14305 with the new "journalctl --after-cursor=" switch to continue
14306 browsing logs from that point on.
14308 * "journalctl --force" may now be used to force regeneration
14311 * Creation of "dead" device nodes has been moved from udev
14312 into kmod and tmpfiles. Previously, udev would read the kmod
14313 databases to pre-generate dead device nodes based on meta
14314 information contained in kernel modules, so that these would
14315 be auto-loaded on access rather then at boot. As this
14316 does not really have much to do with the exposing actual
14317 kernel devices to userspace this has always been slightly
14318 alien in the udev codebase. Following the new scheme kmod
14319 will now generate a runtime snippet for tmpfiles from the
14320 module meta information and it now is tmpfiles' job to the
14321 create the nodes. This also allows overriding access and
14322 other parameters for the nodes using the usual tmpfiles
14323 facilities. As side effect this allows us to remove the
14324 CAP_SYS_MKNOD capability bit from udevd entirely.
14326 * logind's device ACLs may now be applied to these "dead"
14327 devices nodes too, thus finally allowing managed access to
14328 devices such as /dev/snd/sequencer without loading the
14329 backing module right-away.
14331 * A new RPM macro has been added that may be used to apply
14332 tmpfiles configuration during package installation.
14334 * systemd-detect-virt and ConditionVirtualization= now can
14335 detect User-Mode-Linux machines (UML).
14337 * journald will now implicitly log the effective capabilities
14338 set of processes in the message metadata.
14340 * systemd-cryptsetup has gained support for TrueCrypt volumes.
14342 * The initrd interface has been simplified (more specifically,
14343 support for passing performance data via environment
14344 variables and fsck results via files in /run has been
14345 removed). These features were non-essential, and are
14346 nowadays available in a much nicer way by having systemd in
14347 the initrd serialize its state and have the hosts systemd
14348 deserialize it again.
14350 * The udev "keymap" data files and tools to apply keyboard
14351 specific mappings of scan to key codes, and force-release
14352 scan code lists have been entirely replaced by a udev
14353 "keyboard" builtin and a hwdb data file.
14355 * systemd will now honour the kernel's "quiet" command line
14356 argument also during late shutdown, resulting in a
14357 completely silent shutdown when used.
14359 * There's now an option to control the SO_REUSEPORT socket
14360 option in .socket units.
14362 * Instance units will now automatically get a per-template
14363 subslice of system.slice unless something else is explicitly
14364 configured. For example, instances of sshd@.service will now
14365 implicitly be placed in system-sshd.slice rather than
14366 system.slice as before.
14368 * Test coverage support may now be enabled at build time.
14370 Contributions from: Dave Reisner, Frederic Crozat, Harald
14371 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
14372 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
14373 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
14374 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
14375 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
14376 Giokas, Zbigniew Jędrzejewski-Szmek
14378 — Berlin, 2013-07-23
14382 * Two new unit types have been introduced:
14384 Scope units are very similar to service units, however, are
14385 created out of pre-existing processes — instead of PID 1
14386 forking off the processes. By using scope units it is
14387 possible for system services and applications to group their
14388 own child processes (worker processes) in a powerful way
14389 which then maybe used to organize them, or kill them
14390 together, or apply resource limits on them.
14392 Slice units may be used to partition system resources in an
14393 hierarchical fashion and then assign other units to them. By
14394 default there are now three slices: system.slice (for all
14395 system services), user.slice (for all user sessions),
14396 machine.slice (for VMs and containers).
14398 Slices and scopes have been introduced primarily in
14399 context of the work to move cgroup handling to a
14400 single-writer scheme, where only PID 1
14401 creates/removes/manages cgroups.
14403 * There's a new concept of "transient" units. In contrast to
14404 normal units these units are created via an API at runtime,
14405 not from configuration from disk. More specifically this
14406 means it is now possible to run arbitrary programs as
14407 independent services, with all execution parameters passed
14408 in via bus APIs rather than read from disk. Transient units
14409 make systemd substantially more dynamic then it ever was,
14410 and useful as a general batch manager.
14412 * logind has been updated to make use of scope and slice units
14413 for managing user sessions. As a user logs in he will get
14414 his own private slice unit, to which all sessions are added
14415 as scope units. We also added support for automatically
14416 adding an instance of user@.service for the user into the
14417 slice. Effectively logind will no longer create cgroup
14418 hierarchies on its own now, it will defer entirely to PID 1
14419 for this by means of scope, service and slice units. Since
14420 user sessions this way become entities managed by PID 1
14421 the output of "systemctl" is now a lot more comprehensive.
14423 * A new mini-daemon "systemd-machined" has been added which
14424 may be used by virtualization managers to register local
14425 VMs/containers. nspawn has been updated accordingly, and
14426 libvirt will be updated shortly. machined will collect a bit
14427 of meta information about the VMs/containers, and assign
14428 them their own scope unit (see above). The collected
14429 meta-data is then made available via the "machinectl" tool,
14430 and exposed in "ps" and similar tools. machined/machinectl
14431 is compile-time optional.
14433 * As discussed earlier, the low-level cgroup configuration
14434 options ControlGroup=, ControlGroupModify=,
14435 ControlGroupPersistent=, ControlGroupAttribute= have been
14436 removed. Please use high-level attribute settings instead as
14437 well as slice units.
14439 * A new bus call SetUnitProperties() has been added to alter
14440 various runtime parameters of a unit. This is primarily
14441 useful to alter cgroup parameters dynamically in a nice way,
14442 but will be extended later on to make more properties
14443 modifiable at runtime. systemctl gained a new set-properties
14444 command that wraps this call.
14446 * A new tool "systemd-run" has been added which can be used to
14447 run arbitrary command lines as transient services or scopes,
14448 while configuring a number of settings via the command
14449 line. This tool is currently very basic, however already
14450 very useful. We plan to extend this tool to even allow
14451 queuing of execution jobs with time triggers from the
14452 command line, similar in fashion to "at".
14454 * nspawn will now inform the user explicitly that kernels with
14455 audit enabled break containers, and suggest the user to turn
14458 * Support for detecting the IMA and AppArmor security
14459 frameworks with ConditionSecurity= has been added.
14461 * journalctl gained a new "-k" switch for showing only kernel
14462 messages, mimicking dmesg output; in addition to "--user"
14463 and "--system" switches for showing only user's own logs
14466 * systemd-delta can now show information about drop-in
14467 snippets extending unit files.
14469 * libsystemd-bus has been substantially updated but is still
14470 not available as public API.
14472 * systemd will now look for the "debug" argument on the kernel
14473 command line and enable debug logging, similar to what
14474 "systemd.log_level=debug" already did before.
14476 * "systemctl set-default", "systemctl get-default" has been
14477 added to configure the default.target symlink, which
14478 controls what to boot into by default.
14480 * "systemctl set-log-level" has been added as a convenient
14481 way to raise and lower systemd logging threshold.
14483 * "systemd-analyze plot" will now show the time the various
14484 generators needed for execution, as well as information
14485 about the unit file loading.
14487 * libsystemd-journal gained a new sd_journal_open_files() call
14488 for opening specific journal files. journactl also gained a
14489 new switch to expose this new functionality. Previously we
14490 only supported opening all files from a directory, or all
14491 files from the system, as opening individual files only is
14492 racy due to journal file rotation.
14494 * systemd gained the new DefaultEnvironment= setting in
14495 /etc/systemd/system.conf to set environment variables for
14498 * If a privileged process logs a journal message with the
14499 OBJECT_PID= field set, then journald will automatically
14500 augment this with additional OBJECT_UID=, OBJECT_GID=,
14501 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
14502 system services want to log events about specific client
14503 processes. journactl/systemctl has been updated to make use
14504 of this information if all log messages regarding a specific
14507 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
14508 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
14509 Reisner, David Coppa, David King, David Strauss, Eelco
14510 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
14511 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
14512 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
14513 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
14514 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
14515 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
14516 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
14517 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
14518 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
14519 Łukasz Stelmach, 장동준
14523 * The Python bindings gained some minimal support for the APIs
14524 exposed by libsystemd-logind.
14526 * ConditionSecurity= gained support for detecting SMACK. Since
14527 this condition already supports SELinux and AppArmor we only
14528 miss IMA for this. Patches welcome!
14530 Contributions from: Karol Lewandowski, Lennart Poettering,
14531 Zbigniew Jędrzejewski-Szmek
14535 * systemd-nspawn will now create /etc/resolv.conf if
14536 necessary, before bind-mounting the host's file onto it.
14538 * systemd-nspawn will now store meta information about a
14539 container on the container's cgroup as extended attribute
14540 fields, including the root directory.
14542 * The cgroup hierarchy has been reworked in many ways. All
14543 objects any of the components systemd creates in the cgroup
14544 tree are now suffixed. More specifically, user sessions are
14545 now placed in cgroups suffixed with ".session", users in
14546 cgroups suffixed with ".user", and nspawn containers in
14547 cgroups suffixed with ".nspawn". Furthermore, all cgroup
14548 names are now escaped in a simple scheme to avoid collision
14549 of userspace object names with kernel filenames. This work
14550 is preparation for making these objects relocatable in the
14551 cgroup tree, in order to allow easy resource partitioning of
14552 these objects without causing naming conflicts.
14554 * systemctl list-dependencies gained the new switches
14555 --plain, --reverse, --after and --before.
14557 * systemd-inhibit now shows the process name of processes that
14558 have taken an inhibitor lock.
14560 * nss-myhostname will now also resolve "localhost"
14561 implicitly. This makes /etc/hosts an optional file and
14562 nicely handles that on IPv6 ::1 maps to both "localhost" and
14563 the local hostname.
14565 * libsystemd-logind.so gained a new call
14566 sd_get_machine_names() to enumerate running containers and
14567 VMs (currently only supported by very new libvirt and
14568 nspawn). sd_login_monitor can now be used to watch
14569 VMs/containers coming and going.
14571 * .include is not allowed recursively anymore, and only in
14572 unit files. Usually it is better to use drop-in snippets in
14573 .d/*.conf anyway, as introduced with systemd 198.
14575 * systemd-analyze gained a new "critical-chain" command that
14576 determines the slowest chain of units run during system
14577 boot-up. It is very useful for tracking down where
14578 optimizing boot time is the most beneficial.
14580 * systemd will no longer allow manipulating service paths in
14581 the name=systemd:/system cgroup tree using ControlGroup= in
14582 units. (But is still fine with it in all other dirs.)
14584 * There's a new systemd-nspawn@.service service file that may
14585 be used to easily run nspawn containers as system
14586 services. With the container's root directory in
14587 /var/lib/container/foobar it is now sufficient to run
14588 "systemctl start systemd-nspawn@foobar.service" to boot it.
14590 * systemd-cgls gained a new parameter "--machine" to list only
14591 the processes within a certain container.
14593 * ConditionSecurity= now can check for "apparmor". We still
14594 are lacking checks for SMACK and IMA for this condition
14595 check though. Patches welcome!
14597 * A new configuration file /etc/systemd/sleep.conf has been
14598 added that may be used to configure which kernel operation
14599 systemd is supposed to execute when "suspend", "hibernate"
14600 or "hybrid-sleep" is requested. This makes the new kernel
14601 "freeze" state accessible to the user.
14603 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
14604 the passed argument if applicable.
14606 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
14607 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
14608 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
14609 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
14610 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
14611 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
14612 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
14617 * The output of 'systemctl list-jobs' got some polishing. The
14618 '--type=' argument may now be passed more than once. A new
14619 command 'systemctl list-sockets' has been added which shows
14620 a list of kernel sockets systemd is listening on with the
14621 socket units they belong to, plus the units these socket
14624 * The experimental libsystemd-bus library got substantial
14625 updates to work in conjunction with the (also experimental)
14626 kdbus kernel project. It works well enough to exchange
14627 messages with some sophistication. Note that kdbus is not
14628 ready yet, and the library is mostly an elaborate test case
14629 for now, and not installable.
14631 * systemd gained a new unit 'systemd-static-nodes.service'
14632 that generates static device nodes earlier during boot, and
14633 can run in conjunction with udev.
14635 * libsystemd-login gained a new call sd_pid_get_user_unit()
14636 to retrieve the user systemd unit a process is running
14637 in. This is useful for systems where systemd is used as
14640 * systemd-nspawn now places all containers in the new /machine
14641 top-level cgroup directory in the name=systemd
14642 hierarchy. libvirt will soon do the same, so that we get a
14643 uniform separation of /system, /user and /machine for system
14644 services, user processes and containers/virtual
14645 machines. This new cgroup hierarchy is also useful to stick
14646 stable names to specific container instances, which can be
14647 recognized later this way (this name may be controlled
14648 via systemd-nspawn's new -M switch). libsystemd-login also
14649 gained a new call sd_pid_get_machine_name() to retrieve the
14650 name of the container/VM a specific process belongs to.
14652 * bootchart can now store its data in the journal.
14654 * libsystemd-journal gained a new call
14655 sd_journal_add_conjunction() for AND expressions to the
14656 matching logic. This can be used to express more complex
14657 logical expressions.
14659 * journactl can now take multiple --unit= and --user-unit=
14662 * The cryptsetup logic now understands the "luks.key=" kernel
14663 command line switch for specifying a file to read the
14664 decryption key from. Also, if a configured key file is not
14665 found the tool will now automatically fall back to prompting
14668 * Python systemd.journal module was updated to wrap recently
14669 added functions from libsystemd-journal. The interface was
14670 changed to bring the low level interface in s.j._Reader
14671 closer to the C API, and the high level interface in
14672 s.j.Reader was updated to wrap and convert all data about
14675 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
14676 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
14677 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
14678 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
14679 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
14680 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
14684 * journalctl --update-catalog now understands a new --root=
14685 option to operate on catalogs found in a different root
14688 * During shutdown after systemd has terminated all running
14689 services a final killing loop kills all remaining left-over
14690 processes. We will now print the name of these processes
14691 when we send SIGKILL to them, since this usually indicates a
14694 * If /etc/crypttab refers to password files stored on
14695 configured mount points automatic dependencies will now be
14696 generated to ensure the specific mount is established first
14697 before the key file is attempted to be read.
14699 * 'systemctl status' will now show information about the
14700 network sockets a socket unit is listening on.
14702 * 'systemctl status' will also shown information about any
14703 drop-in configuration file for units. (Drop-In configuration
14704 files in this context are files such as
14705 /etc/systemd/system/foobar.service.d/*.conf)
14707 * systemd-cgtop now optionally shows summed up CPU times of
14708 cgroups. Press '%' while running cgtop to switch between
14709 percentage and absolute mode. This is useful to determine
14710 which cgroups use up the most CPU time over the entire
14711 runtime of the system. systemd-cgtop has also been updated
14712 to be 'pipeable' for processing with further shell tools.
14714 * 'hostnamectl set-hostname' will now allow setting of FQDN
14717 * The formatting and parsing of time span values has been
14718 changed. The parser now understands fractional expressions
14719 such as "5.5h". The formatter will now output fractional
14720 expressions for all time spans under 1min, i.e. "5.123456s"
14721 rather than "5s 123ms 456us". For time spans under 1s
14722 millisecond values are shown, for those under 1ms
14723 microsecond values are shown. This should greatly improve
14724 all time-related output of systemd.
14726 * libsystemd-login and libsystemd-journal gained new
14727 functions for querying the poll() events mask and poll()
14728 timeout value for integration into arbitrary event
14731 * localectl gained the ability to list available X11 keymaps
14732 (models, layouts, variants, options).
14734 * 'systemd-analyze dot' gained the ability to filter for
14735 specific units via shell-style globs, to create smaller,
14736 more useful graphs. I.e. it is now possible to create simple
14737 graphs of all the dependencies between only target units, or
14738 of all units that Avahi has dependencies with.
14740 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
14741 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
14742 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
14743 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
14744 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
14745 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
14746 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
14750 * The boot-time readahead implementation for rotating media
14751 will now read the read-ahead data in multiple passes which
14752 consist of all read requests made in equidistant time
14753 intervals. This means instead of strictly reading read-ahead
14754 data in its physical order on disk we now try to find a
14755 middle ground between physical and access time order.
14757 * /etc/os-release files gained a new BUILD_ID= field for usage
14758 on operating systems that provide continuous builds of OS
14761 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
14762 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
14763 William Douglas, Zbigniew Jędrzejewski-Szmek
14767 * systemd-python gained an API exposing libsystemd-daemon.
14769 * The SMACK setup logic gained support for uploading CIPSO
14772 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
14773 ReadOnlyDirectories= and InaccessibleDirectories= has
14774 changed. The private /tmp and /var/tmp directories are now
14775 shared by all processes of a service (which means
14776 ExecStartPre= may now leave data in /tmp that ExecStart= of
14777 the same service can still access). When a service is
14778 stopped its temporary directories are immediately deleted
14779 (normal clean-up with tmpfiles is still done in addition to
14782 * By default, systemd will now set a couple of sysctl
14783 variables in the kernel: the safe sysrq options are turned
14784 on, IP route verification is turned on, and source routing
14785 disabled. The recently added hardlink and softlink
14786 protection of the kernel is turned on. These settings should
14787 be reasonably safe, and good defaults for all new systems.
14789 * The predictable network naming logic may now be turned off
14790 with a new kernel command line switch: net.ifnames=0.
14792 * A new libsystemd-bus module has been added that implements a
14793 pretty complete D-Bus client library. For details see:
14795 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
14797 * journald will now explicitly flush the journal files to disk
14798 at the latest 5min after each write. The file will then also
14799 be marked offline until the next write. This should increase
14800 reliability in case of a crash. The synchronization delay
14801 can be configured via SyncIntervalSec= in journald.conf.
14803 * There's a new remote-fs-setup.target unit that can be used
14804 to pull in specific services when at least one remote file
14805 system is to be mounted.
14807 * There are new targets timers.target and paths.target as
14808 canonical targets to pull user timer and path units in
14809 from. This complements sockets.target with a similar
14810 purpose for socket units.
14812 * libudev gained a new call udev_device_set_attribute_value()
14813 to set sysfs attributes of a device.
14815 * The udev daemon now sets the default number of worker
14816 processes executed in parallel based on the number of available
14817 CPUs instead of the amount of available RAM. This is supposed
14818 to provide a more reliable default and limit a too aggressive
14819 parallelism for setups with 1000s of devices connected.
14821 Contributions from: Auke Kok, Colin Walters, Cristian
14822 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
14823 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
14824 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
14825 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
14826 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
14827 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
14828 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
14829 Zbigniew Jędrzejewski-Szmek
14833 * Configuration of unit files may now be extended via drop-in
14834 files without having to edit/override the unit files
14835 themselves. More specifically, if the administrator wants to
14836 change one value for a service file foobar.service he can
14837 now do so by dropping in a configuration snippet into
14838 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
14839 will load all these snippets and apply them on top of the
14840 main unit configuration file, possibly extending or
14841 overriding its settings. Using these drop-in snippets is
14842 generally nicer than the two earlier options for changing
14843 unit files locally: copying the files from
14844 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
14845 them there; or creating a new file in /etc/systemd/system/
14846 that incorporates the original one via ".include". Drop-in
14847 snippets into these .d/ directories can be placed in any
14848 directory systemd looks for units in, and the usual
14849 overriding semantics between /usr/lib, /etc and /run apply
14852 * Most unit file settings which take lists of items can now be
14853 reset by assigning the empty string to them. For example,
14854 normally, settings such as Environment=FOO=BAR append a new
14855 environment variable assignment to the environment block,
14856 each time they are used. By assigning Environment= the empty
14857 string the environment block can be reset to empty. This is
14858 particularly useful with the .d/*.conf drop-in snippets
14859 mentioned above, since this adds the ability to reset list
14860 settings from vendor unit files via these drop-ins.
14862 * systemctl gained a new "list-dependencies" command for
14863 listing the dependencies of a unit recursively.
14865 * Inhibitors are now honored and listed by "systemctl
14866 suspend", "systemctl poweroff" (and similar) too, not only
14867 GNOME. These commands will also list active sessions by
14870 * Resource limits (as exposed by the various control group
14871 controllers) can now be controlled dynamically at runtime
14872 for all units. More specifically, you can now use a command
14873 like "systemctl set-cgroup-attr foobar.service cpu.shares
14874 2000" to alter the CPU shares a specific service gets. These
14875 settings are stored persistently on disk, and thus allow the
14876 administrator to easily adjust the resource usage of
14877 services with a few simple commands. This dynamic resource
14878 management logic is also available to other programs via the
14879 bus. Almost any kernel cgroup attribute and controller is
14882 * systemd-vconsole-setup will now copy all font settings to
14883 all allocated VTs, where it previously applied them only to
14886 * libsystemd-login gained the new sd_session_get_tty() API
14889 * This release drops support for a few legacy or
14890 distribution-specific LSB facility names when parsing init
14891 scripts: $x-display-manager, $mail-transfer-agent,
14892 $mail-transport-agent, $mail-transfer-agent, $smtp,
14893 $null. Also, the mail-transfer-agent.target unit backing
14894 this has been removed. Distributions which want to retain
14895 compatibility with this should carry the burden for
14896 supporting this themselves and patch support for these back
14897 in, if they really need to. Also, the facilities $syslog and
14898 $local_fs are now ignored, since systemd does not support
14899 early-boot LSB init scripts anymore, and these facilities
14900 are implied anyway for normal services. syslog.target has
14903 * There are new bus calls on PID1's Manager object for
14904 cancelling jobs, and removing snapshot units. Previously,
14905 both calls were only available on the Job and Snapshot
14906 objects themselves.
14908 * systemd-journal-gatewayd gained SSL support.
14910 * The various "environment" files, such as /etc/locale.conf
14911 now support continuation lines with a backslash ("\") as
14912 last character in the line, similarly in style (but different)
14913 to how this is supported in shells.
14915 * For normal user processes the _SYSTEMD_USER_UNIT= field is
14916 now implicitly appended to every log entry logged. systemctl
14917 has been updated to filter by this field when operating on a
14918 user systemd instance.
14920 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
14921 CAP_AUDIT_CONTROL capabilities to the capabilities set for
14922 the container. This makes it easier to boot unmodified
14923 Fedora systems in a container, which however still requires
14924 audit=0 to be passed on the kernel command line. Auditing in
14925 kernel and userspace is unfortunately still too broken in
14926 context of containers, hence we recommend compiling it out
14927 of the kernel or using audit=0. Hopefully this will be fixed
14928 one day for good in the kernel.
14930 * nspawn gained the new --bind= and --bind-ro= parameters to
14931 bind mount specific directories from the host into the
14934 * nspawn will now mount its own devpts file system instance
14935 into the container, in order not to leak pty devices from
14936 the host into the container.
14938 * systemd will now read the firmware boot time performance
14939 information from the EFI variables, if the used boot loader
14940 supports this, and takes it into account for boot performance
14941 analysis via "systemd-analyze". This is currently supported
14942 only in conjunction with Gummiboot, but could be supported
14943 by other boot loaders too. For details see:
14945 https://systemd.io/BOOT_LOADER_INTERFACE
14947 * A new generator has been added that automatically mounts the
14948 EFI System Partition (ESP) to /boot, if that directory
14949 exists, is empty, and no other file system has been
14950 configured to be mounted there.
14952 * logind will now send out PrepareForSleep(false) out
14953 unconditionally, after coming back from suspend. This may be
14954 used by applications as asynchronous notification for
14955 system resume events.
14957 * "systemctl unlock-sessions" has been added, that allows
14958 unlocking the screens of all user sessions at once, similar
14959 to how "systemctl lock-sessions" already locked all users
14960 sessions. This is backed by a new D-Bus call UnlockSessions().
14962 * "loginctl seat-status" will now show the master device of a
14963 seat. (i.e. the device of a seat that needs to be around for
14964 the seat to be considered available, usually the graphics
14967 * tmpfiles gained a new "X" line type, that allows
14968 configuration of files and directories (with wildcards) that
14969 shall be excluded from automatic cleanup ("aging").
14971 * udev default rules set the device node permissions now only
14972 at "add" events, and do not change them any longer with a
14973 later "change" event.
14975 * The log messages for lid events and power/sleep keypresses
14976 now carry a message ID.
14978 * We now have a substantially larger unit test suite, but this
14979 continues to be work in progress.
14981 * udevadm hwdb gained a new --root= parameter to change the
14982 root directory to operate relative to.
14984 * logind will now issue a background sync() request to the kernel
14985 early at shutdown, so that dirty buffers are flushed to disk early
14986 instead of at the last moment, in order to optimize shutdown
14989 * A new bootctl tool has been added that is an interface for
14990 certain boot loader operations. This is currently a preview
14991 and is likely to be extended into a small mechanism daemon
14992 like timedated, localed, hostnamed, and can be used by
14993 graphical UIs to enumerate available boot options, and
14994 request boot into firmware operations.
14996 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
14997 the rest of the package. It also has been updated to work
14998 correctly in initrds.
15000 * polkit previously has been runtime optional, and is now also
15001 compile time optional via a configure switch.
15003 * systemd-analyze has been reimplemented in C. Also "systemctl
15004 dot" has moved into systemd-analyze.
15006 * "systemctl status" with no further parameters will now print
15007 the status of all active or failed units.
15009 * Operations such as "systemctl start" can now be executed
15010 with a new mode "--irreversible" which may be used to queue
15011 operations that cannot accidentally be reversed by a later
15012 job queuing. This is by default used to make shutdown
15013 requests more robust.
15015 * The Python API of systemd now gained a new module for
15016 reading journal files.
15018 * A new tool kernel-install has been added that can install
15019 kernel images according to the Boot Loader Specification:
15021 https://systemd.io/BOOT_LOADER_SPECIFICATION
15023 * Boot time console output has been improved to provide
15024 animated boot time output for hanging jobs.
15026 * A new tool systemd-activate has been added which can be used
15027 to test socket activation with, directly from the command
15028 line. This should make it much easier to test and debug
15029 socket activation in daemons.
15031 * journalctl gained a new "--reverse" (or -r) option to show
15032 journal output in reverse order (i.e. newest line first).
15034 * journalctl gained a new "--pager-end" (or -e) option to jump
15035 to immediately jump to the end of the journal in the
15036 pager. This is only supported in conjunction with "less".
15038 * journalctl gained a new "--user-unit=" option, that works
15039 similarly to "--unit=" but filters for user units rather than
15042 * A number of unit files to ease adoption of systemd in
15043 initrds has been added. This moves some minimal logic from
15044 the various initrd implementations into systemd proper.
15046 * The journal files are now owned by a new group
15047 "systemd-journal", which exists specifically to allow access
15048 to the journal, and nothing else. Previously, we used the
15049 "adm" group for that, which however possibly covers more
15050 than just journal/log file access. This new group is now
15051 already used by systemd-journal-gatewayd to ensure this
15052 daemon gets access to the journal files and as little else
15053 as possible. Note that "make install" will also set FS ACLs
15054 up for /var/log/journal to give "adm" and "wheel" read
15055 access to it, in addition to "systemd-journal" which owns
15056 the journal files. We recommend that packaging scripts also
15057 add read access to "adm" + "wheel" to /var/log/journal, and
15058 all existing/future journal files. To normal users and
15059 administrators little changes, however packagers need to
15060 ensure to create the "systemd-journal" system group at
15061 package installation time.
15063 * The systemd-journal-gatewayd now runs as unprivileged user
15064 systemd-journal-gateway:systemd-journal-gateway. Packaging
15065 scripts need to create these system user/group at
15068 * timedated now exposes a new boolean property CanNTP that
15069 indicates whether a local NTP service is available or not.
15071 * systemd-detect-virt will now also detect xen PVs
15073 * The pstore file system is now mounted by default, if it is
15076 * In addition to the SELinux and IMA policies we will now also
15077 load SMACK policies at early boot.
15079 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
15080 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
15081 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
15082 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
15083 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
15084 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
15085 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
15086 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
15087 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
15088 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
15089 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
15090 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
15091 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
15092 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
15096 * Timer units now support calendar time events in addition to
15097 monotonic time events. That means you can now trigger a unit
15098 based on a calendar time specification such as "Thu,Fri
15099 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
15100 or fifth day of any month of the year 2013, given that it is
15101 a Thursday or a Friday. This brings timer event support
15102 considerably closer to cron's capabilities. For details on
15103 the supported calendar time specification language see
15106 * udev now supports a number of different naming policies for
15107 network interfaces for predictable names, and a combination
15108 of these policies is now the default. Please see this wiki
15109 document for details:
15111 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
15113 * Auke Kok's bootchart implementation has been added to the
15114 systemd tree. It is an optional component that can graph the
15115 boot in quite some detail. It is one of the best bootchart
15116 implementations around and minimal in its code and
15119 * nss-myhostname has been integrated into the systemd source
15120 tree. nss-myhostname guarantees that the local hostname
15121 always stays resolvable via NSS. It has been a weak
15122 requirement of systemd-hostnamed since a long time, and
15123 since its code is actually trivial we decided to just
15124 include it in systemd's source tree. It can be turned off
15125 with a configure switch.
15127 * The read-ahead logic is now capable of properly detecting
15128 whether a btrfs file system is on SSD or rotating media, in
15129 order to optimize the read-ahead scheme. Previously, it was
15130 only capable of detecting this on traditional file systems
15133 * In udev, additional device properties are now read from the
15134 IAB in addition to the OUI database. Also, Bluetooth company
15135 identities are attached to the devices as well.
15137 * In service files %U may be used as specifier that is
15138 replaced by the configured user name of the service.
15140 * nspawn may now be invoked without a controlling TTY. This
15141 makes it suitable for invocation as its own service. This
15142 may be used to set up a simple containerized server system
15143 using only core OS tools.
15145 * systemd and nspawn can now accept socket file descriptors
15146 when they are started for socket activation. This enables
15147 implementation of socket activated nspawn
15148 containers. i.e. think about autospawning an entire OS image
15149 when the first SSH or HTTP connection is received. We expect
15150 that similar functionality will also be added to libvirt-lxc
15153 * journalctl will now suppress ANSI color codes when
15154 presenting log data.
15156 * systemctl will no longer show control group information for
15157 a unit if the control group is empty anyway.
15159 * logind can now automatically suspend/hibernate/shutdown the
15162 * /etc/machine-info and hostnamed now also expose the chassis
15163 type of the system. This can be used to determine whether
15164 the local system is a laptop, desktop, handset or
15165 tablet. This information may either be configured by the
15166 user/vendor or is automatically determined from ACPI and DMI
15167 information if possible.
15169 * A number of polkit actions are now bound together with "imply"
15170 rules. This should simplify creating UIs because many actions
15171 will now authenticate similar ones as well.
15173 * Unit files learnt a new condition ConditionACPower= which
15174 may be used to conditionalize a unit depending on whether an
15175 AC power source is connected or not, of whether the system
15176 is running on battery power.
15178 * systemctl gained a new "is-failed" verb that may be used in
15179 shell scripts and suchlike to check whether a specific unit
15180 is in the "failed" state.
15182 * The EnvironmentFile= setting in unit files now supports file
15183 globbing, and can hence be used to easily read a number of
15184 environment files at once.
15186 * systemd will no longer detect and recognize specific
15187 distributions. All distribution-specific #ifdeffery has been
15188 removed, systemd is now fully generic and
15189 distribution-agnostic. Effectively, not too much is lost as
15190 a lot of the code is still accessible via explicit configure
15191 switches. However, support for some distribution specific
15192 legacy configuration file formats has been dropped. We
15193 recommend distributions to simply adopt the configuration
15194 files everybody else uses now and convert the old
15195 configuration from packaging scripts. Most distributions
15196 already did that. If that's not possible or desirable,
15197 distributions are welcome to forward port the specific
15198 pieces of code locally from the git history.
15200 * When logging a message about a unit systemd will now always
15201 log the unit name in the message meta data.
15203 * localectl will now also discover system locale data that is
15204 not stored in locale archives, but directly unpacked.
15206 * logind will no longer unconditionally use framebuffer
15207 devices as seat masters, i.e. as devices that are required
15208 to be existing before a seat is considered preset. Instead,
15209 it will now look for all devices that are tagged as
15210 "seat-master" in udev. By default, framebuffer devices will
15211 be marked as such, but depending on local systems, other
15212 devices might be marked as well. This may be used to
15213 integrate graphics cards using closed source drivers (such
15214 as NVidia ones) more nicely into logind. Note however, that
15215 we recommend using the open source NVidia drivers instead,
15216 and no udev rules for the closed-source drivers will be
15217 shipped from us upstream.
15219 Contributions from: Adam Williamson, Alessandro Crismani, Auke
15220 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
15221 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
15222 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
15223 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
15224 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
15225 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
15226 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
15227 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
15228 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
15229 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
15230 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
15235 * udev gained support for loading additional device properties
15236 from an indexed database that is keyed by vendor/product IDs
15237 and similar device identifiers. For the beginning this
15238 "hwdb" is populated with data from the well-known PCI and
15239 USB database, but also includes PNP, ACPI and OID data. In
15240 the longer run this indexed database shall grow into
15241 becoming the one central database for non-essential
15242 userspace device metadata. Previously, data from the PCI/USB
15243 database was only attached to select devices, since the
15244 lookup was a relatively expensive operation due to O(n) time
15245 complexity (with n being the number of entries in the
15246 database). Since this is now O(1), we decided to add in this
15247 data for all devices where this is available, by
15248 default. Note that the indexed database needs to be rebuilt
15249 when new data files are installed. To achieve this you need
15250 to update your packaging scripts to invoke "udevadm hwdb
15251 --update" after installation of hwdb data files. For
15252 RPM-based distributions we introduced the new
15253 %udev_hwdb_update macro for this purpose.
15255 * The Journal gained support for the "Message Catalog", an
15256 indexed database to link up additional information with
15257 journal entries. For further details please check:
15259 https://www.freedesktop.org/wiki/Software/systemd/catalog
15261 The indexed message catalog database also needs to be
15262 rebuilt after installation of message catalog files. Use
15263 "journalctl --update-catalog" for this. For RPM-based
15264 distributions we introduced the %journal_catalog_update
15265 macro for this purpose.
15267 * The Python Journal bindings gained support for the standard
15268 Python logging framework.
15270 * The Journal API gained new functions for checking whether
15271 the underlying file system of a journal file is capable of
15272 properly reporting file change notifications, or whether
15273 applications that want to reflect journal changes "live"
15274 need to recheck journal files continuously in appropriate
15277 * It is now possible to set the "age" field for tmpfiles
15278 entries to 0, indicating that files matching this entry
15279 shall always be removed when the directories are cleaned up.
15281 * coredumpctl gained a new "gdb" verb which invokes gdb
15282 right-away on the selected coredump.
15284 * There's now support for "hybrid sleep" on kernels that
15285 support this, in addition to "suspend" and "hibernate". Use
15286 "systemctl hybrid-sleep" to make use of this.
15288 * logind's HandleSuspendKey= setting (and related settings)
15289 now gained support for a new "lock" setting to simply
15290 request the screen lock on all local sessions, instead of
15291 actually executing a suspend or hibernation.
15293 * systemd will now mount the EFI variables file system by
15296 * Socket units now gained support for configuration of the
15297 SMACK security label.
15299 * timedatectl will now output the time of the last and next
15300 daylight saving change.
15302 * We dropped support for various legacy and distro-specific
15303 concepts, such as insserv, early-boot SysV services
15304 (i.e. those for non-standard runlevels such as 'b' or 'S')
15305 or ArchLinux /etc/rc.conf support. We recommend the
15306 distributions who still need support this to either continue
15307 to maintain the necessary patches downstream, or find a
15308 different solution. (Talk to us if you have questions!)
15310 * Various systemd components will now bypass polkit checks for
15311 root and otherwise handle properly if polkit is not found to
15312 be around. This should fix most issues for polkit-less
15313 systems. Quite frankly this should have been this way since
15314 day one. It is absolutely our intention to make systemd work
15315 fine on polkit-less systems, and we consider it a bug if
15316 something does not work as it should if polkit is not around.
15318 * For embedded systems it is now possible to build udev and
15319 systemd without blkid and/or kmod support.
15321 * "systemctl switch-root" is now capable of switching root
15322 more than once. I.e. in addition to transitions from the
15323 initrd to the host OS it is now possible to transition to
15324 further OS images from the host. This is useful to implement
15325 offline updating tools.
15327 * Various other additions have been made to the RPM macros
15328 shipped with systemd. Use %udev_rules_update() after
15329 installing new udev rules files. %_udevhwdbdir,
15330 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
15331 %_sysctldir are now available which resolve to the right
15332 directories for packages to place various data files in.
15334 * journalctl gained the new --full switch (in addition to
15335 --all, to disable ellipsation for long messages.
15337 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
15338 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
15339 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
15340 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
15341 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
15342 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
15343 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
15344 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
15345 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
15349 * journalctl gained new --since= and --until= switches to
15350 filter by time. It also now supports nice filtering for
15351 units via --unit=/-u.
15353 * Type=oneshot services may use ExecReload= and do the
15356 * The journal daemon now supports time-based rotation and
15357 vacuuming, in addition to the usual disk-space based
15360 * The journal will now index the available field values for
15361 each field name. This enables clients to show pretty drop
15362 downs of available match values when filtering. The bash
15363 completion of journalctl has been updated
15364 accordingly. journalctl gained a new switch -F to list all
15365 values a certain field takes in the journal database.
15367 * More service events are now written as structured messages
15368 to the journal, and made recognizable via message IDs.
15370 * The timedated, localed and hostnamed mini-services which
15371 previously only provided support for changing time, locale
15372 and hostname settings from graphical DEs such as GNOME now
15373 also have a minimal (but very useful) text-based client
15374 utility each. This is probably the nicest way to changing
15375 these settings from the command line now, especially since
15376 it lists available options and is fully integrated with bash
15379 * There's now a new tool "systemd-coredumpctl" to list and
15380 extract coredumps from the journal.
15382 * We now install a README each in /var/log/ and
15383 /etc/rc.d/init.d explaining where the system logs and init
15384 scripts went. This hopefully should help folks who go to
15385 that dirs and look into the otherwise now empty void and
15386 scratch their heads.
15388 * When user-services are invoked (by systemd --user) the
15389 $MANAGERPID env var is set to the PID of systemd.
15391 * SIGRTMIN+24 when sent to a --user instance will now result
15392 in immediate termination of systemd.
15394 * gatewayd received numerous feature additions such as a
15395 "follow" mode, for live syncing and filtering.
15397 * browse.html now allows filtering and showing detailed
15398 information on specific entries. Keyboard navigation and
15399 mouse screen support has been added.
15401 * gatewayd/journalctl now supports HTML5/JSON
15402 Server-Sent-Events as output.
15404 * The SysV init script compatibility logic will now
15405 heuristically determine whether a script supports the
15406 "reload" verb, and only then make this available as
15407 "systemctl reload".
15409 * "systemctl status --follow" has been removed, use "journalctl
15412 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
15413 have been removed since they are hardly useful to be
15416 * And I'd like to take the opportunity to specifically mention
15417 Zbigniew for his great contributions. Zbigniew, you rock!
15419 Contributions from: Andrew Eikum, Christian Hesse, Colin
15420 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
15421 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
15422 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
15423 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
15424 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
15425 Jędrzejewski-Szmek, Сковорода Никита Андреевич
15429 * If /etc/vconsole.conf is non-existent or empty we will no
15430 longer load any console font or key map at boot by
15431 default. Instead the kernel defaults will be left
15432 intact. This is definitely the right thing to do, as no
15433 configuration should mean no configuration, and hard-coding
15434 font names that are different on all archs is probably a bad
15435 idea. Also, the kernel default key map and font should be
15436 good enough for most cases anyway, and mostly identical to
15437 the userspace fonts/key maps we previously overloaded them
15438 with. If distributions want to continue to default to a
15439 non-kernel font or key map they should ship a default
15440 /etc/vconsole.conf with the appropriate contents.
15442 Contributions from: Colin Walters, Daniel J Walsh, Dave
15443 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
15444 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15448 * journalctl gained a new --cursor= switch to show entries
15449 starting from the specified location in the journal.
15451 * We now enforce a size limit on journal entry fields exported
15452 with "-o json" in journalctl. Fields larger than 4K will be
15453 assigned null. This can be turned off with --all.
15455 * An (optional) journal gateway daemon is now available as
15456 "systemd-journal-gatewayd.service". This service provides
15457 access to the journal via HTTP and JSON. This functionality
15458 will be used to implement live log synchronization in both
15459 pull and push modes, but has various other users too, such
15460 as easy log access for debugging of embedded devices. Right
15461 now it is already useful to retrieve the journal via HTTP:
15463 # systemctl start systemd-journal-gatewayd.service
15464 # wget http://localhost:19531/entries
15466 This will download the journal contents in a
15467 /var/log/messages compatible format. The same as JSON:
15469 # curl -H"Accept: application/json" http://localhost:19531/entries
15471 This service is also accessible via a web browser where a
15472 single static HTML5 app is served that uses the JSON logic
15473 to enable the user to do some basic browsing of the
15474 journal. This will be extended later on. Here's an example
15475 screenshot of this app in its current state:
15477 https://0pointer.de/public/journal-gatewayd
15479 Contributions from: Kay Sievers, Lennart Poettering, Robert
15480 Milasan, Tom Gundersen
15484 * The bash completion logic is now available for journalctl
15487 * We do not mount the "cpuset" controller anymore together with
15488 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
15489 started if no parameters are assigned to it. "cpuset" hence
15490 broke code that assumed it could create "cpu" groups and
15493 * journalctl -f will now subscribe to terminal size changes,
15494 and line break accordingly.
15496 Contributions from: Dave Reisner, Kay Sievers, Lennart
15497 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
15501 * nspawn will now create a symlink /etc/localtime in the
15502 container environment, copying the host's timezone
15503 setting. Previously this has been done via a bind mount, but
15504 since symlinks cannot be bind mounted this has now been
15505 changed to create/update the appropriate symlink.
15507 * journalctl -n's line number argument is now optional, and
15508 will default to 10 if omitted.
15510 * journald will now log the maximum size the journal files may
15511 take up on disk. This is particularly useful if the default
15512 built-in logic of determining this parameter from the file
15513 system size is used. Use "systemctl status
15514 systemd-journald.service" to see this information.
15516 * The multi-seat X wrapper tool has been stripped down. As X
15517 is now capable of enumerating graphics devices via udev in a
15518 seat-aware way the wrapper is not strictly necessary
15519 anymore. A stripped down temporary stop-gap is still shipped
15520 until the upstream display managers have been updated to
15521 fully support the new X logic. Expect this wrapper to be
15522 removed entirely in one of the next releases.
15524 * HandleSleepKey= in logind.conf has been split up into
15525 HandleSuspendKey= and HandleHibernateKey=. The old setting
15526 is not available anymore. X11 and the kernel are
15527 distinguishing between these keys and we should too. This
15528 also means the inhibition lock for these keys has been split
15531 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
15532 Poettering, Lukas Nykryn, Václav Pavlín
15536 * Whenever a unit changes state we will now log this to the
15537 journal and show along the unit's own log output in
15538 "systemctl status".
15540 * ConditionPathIsMountPoint= can now properly detect bind
15541 mount points too. (Previously, a bind mount of one file
15542 system to another place in the same file system could not be
15543 detected as mount, since they shared struct stat's st_dev
15546 * We will now mount the cgroup controllers cpu, cpuacct,
15547 cpuset and the controllers net_cls, net_prio together by
15550 * nspawn containers will now have a virtualized boot
15551 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
15552 over with a randomized ID at container initialization). This
15553 has the effect of making "journalctl -b" do the right thing
15556 * The JSON output journal serialization has been updated not
15557 to generate "endless" list objects anymore, but rather one
15558 JSON object per line. This is more in line how most JSON
15559 parsers expect JSON objects. The new output mode
15560 "json-pretty" has been added to provide similar output, but
15561 neatly aligned for readability by humans.
15563 * We dropped all explicit sync() invocations in the shutdown
15564 code. The kernel does this implicitly anyway in the kernel
15565 reboot() syscall. halt(8)'s -n option is now a compatibility
15568 * We now support virtualized reboot() in containers, as
15569 supported by newer kernels. We will fall back to exit() if
15570 CAP_SYS_REBOOT is not available to the container. Also,
15571 nspawn makes use of this now and will actually reboot the
15572 container if the containerized OS asks for that.
15574 * journalctl will only show local log output by default
15575 now. Use --merge (-m) to show remote log output, too.
15577 * libsystemd-journal gained the new sd_journal_get_usage()
15578 call to determine the current disk usage of all journal
15579 files. This is exposed in the new "journalctl --disk-usage"
15582 * journald gained a new configuration setting SplitMode= in
15583 journald.conf which may be used to control how user journals
15584 are split off. See journald.conf(5) for details.
15586 * A new condition type ConditionFileNotEmpty= has been added.
15588 * tmpfiles' "w" lines now support file globbing, to write
15589 multiple files at once.
15591 * We added Python bindings for the journal submission
15592 APIs. More Python APIs for a number of selected APIs will
15593 likely follow. Note that we intend to add native bindings
15594 only for the Python language, as we consider it common
15595 enough to deserve bindings shipped within systemd. There are
15596 various projects outside of systemd that provide bindings
15597 for languages such as PHP or Lua.
15599 * Many conditions will now resolve specifiers such as %i. In
15600 addition, PathChanged= and related directives of .path units
15601 now support specifiers as well.
15603 * There's now a new RPM macro definition for the system preset
15606 * journald will now warn if it ca not forward a message to the
15607 syslog daemon because its socket is full.
15609 * timedated will no longer write or process /etc/timezone,
15610 except on Debian. As we do not support late mounted /usr
15611 anymore /etc/localtime always being a symlink is now safe,
15612 and hence the information in /etc/timezone is not necessary
15615 * logind will now always reserve one VT for a text getty (VT6
15616 by default). Previously if more than 6 X sessions where
15617 started they took up all the VTs with auto-spawned gettys,
15618 so that no text gettys were available anymore.
15620 * udev will now automatically inform the btrfs kernel logic
15621 about btrfs RAID components showing up. This should make
15622 simple hotplug based btrfs RAID assembly work.
15624 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
15625 (but not for its children which will stay at the kernel
15626 default). This should allow setups with a lot more listening
15629 * systemd will now always pass the configured timezone to the
15630 kernel at boot. timedated will do the same when the timezone
15633 * logind's inhibition logic has been updated. By default,
15634 logind will now handle the lid switch, the power and sleep
15635 keys all the time, even in graphical sessions. If DEs want
15636 to handle these events on their own they should take the new
15637 handle-power-key, handle-sleep-key and handle-lid-switch
15638 inhibitors during their runtime. A simple way to achieve
15639 that is to invoke the DE wrapped in an invocation of:
15641 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
15643 * Access to unit operations is now checked via SELinux taking
15644 the unit file label and client process label into account.
15646 * systemd will now notify the administrator in the journal
15647 when he over-mounts a non-empty directory.
15649 * There are new specifiers that are resolved in unit files,
15650 for the hostname (%H), the machine ID (%m) and the boot ID
15653 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
15654 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
15655 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
15656 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
15657 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
15658 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
15659 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
15663 * Support for reading structured kernel messages from
15664 /dev/kmsg has now been added and is enabled by default.
15666 * Support for reading kernel messages from /proc/kmsg has now
15667 been removed. If you want kernel messages in the journal
15668 make sure to run a recent kernel (>= 3.5) that supports
15669 reading structured messages from /dev/kmsg (see
15670 above). /proc/kmsg is now exclusive property of classic
15671 syslog daemons again.
15673 * The libudev API gained the new
15674 udev_device_new_from_device_id() call.
15676 * The logic for file system namespace (ReadOnlyDirectory=,
15677 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
15678 require pivot_root() anymore. This means fewer temporary
15679 directories are created below /tmp for this feature.
15681 * nspawn containers will now see and receive all submounts
15682 made on the host OS below the root file system of the
15685 * Forward Secure Sealing is now supported for Journal files,
15686 which provide cryptographical sealing of journal files so
15687 that attackers cannot alter log history anymore without this
15688 being detectable. Lennart will soon post a blog story about
15689 this explaining it in more detail.
15691 * There are two new service settings RestartPreventExitStatus=
15692 and SuccessExitStatus= which allow configuration of exit
15693 status (exit code or signal) which will be excepted from the
15694 restart logic, resp. consider successful.
15696 * journalctl gained the new --verify switch that can be used
15697 to check the integrity of the structure of journal files and
15698 (if Forward Secure Sealing is enabled) the contents of
15701 * nspawn containers will now be run with /dev/stdin, /dev/fd/
15702 and similar symlinks pre-created. This makes running shells
15703 as container init process a lot more fun.
15705 * The fstab support can now handle PARTUUID= and PARTLABEL=
15708 * A new ConditionHost= condition has been added to match
15709 against the hostname (with globs) and machine ID. This is
15710 useful for clusters where a single OS image is used to
15711 provision a large number of hosts which shall run slightly
15712 different sets of services.
15714 * Services which hit the restart limit will now be placed in a
15717 Contributions from: Bertram Poettering, Dave Reisner, Huang
15718 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
15719 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
15723 * When running in --user mode systemd will now become a
15724 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
15725 tree a lot more organized.
15727 * A new PartOf= unit dependency type has been introduced that
15728 may be used to group services in a natural way.
15730 * "systemctl enable" may now be used to enable instances of
15733 * journalctl now prints error log levels in red, and
15734 warning/notice log levels in bright white. It also supports
15735 filtering by log level now.
15737 * cgtop gained a new -n switch (similar to top), to configure
15738 the maximum number of iterations to run for. It also gained
15739 -b, to run in batch mode (accepting no input).
15741 * The suffix ".service" may now be omitted on most systemctl
15742 command lines involving service unit names.
15744 * There's a new bus call in logind to lock all sessions, as
15745 well as a loginctl verb for it "lock-sessions".
15747 * libsystemd-logind.so gained a new call sd_journal_perror()
15748 that works similar to libc perror() but logs to the journal
15749 and encodes structured information about the error number.
15751 * /etc/crypttab entries now understand the new keyfile-size=
15754 * shutdown(8) now can send a (configurable) wall message when
15755 a shutdown is cancelled.
15757 * The mount propagation mode for the root file system will now
15758 default to "shared", which is useful to make containers work
15759 nicely out-of-the-box so that they receive new mounts from
15760 the host. This can be undone locally by running "mount
15761 --make-rprivate /" if needed.
15763 * The prefdm.service file has been removed. Distributions
15764 should maintain this unit downstream if they intend to keep
15765 it around. However, we recommend writing normal unit files
15766 for display managers instead.
15768 * Since systemd is a crucial part of the OS we will now
15769 default to a number of compiler switches that improve
15770 security (hardening) such as read-only relocations, stack
15771 protection, and suchlike.
15773 * The TimeoutSec= setting for services is now split into
15774 TimeoutStartSec= and TimeoutStopSec= to allow configuration
15775 of individual time outs for the start and the stop phase of
15778 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
15779 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
15780 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
15781 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
15782 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
15783 Gundersen, Zbigniew Jędrzejewski-Szmek
15787 * The journal and id128 C APIs are now fully documented as man
15790 * Extra safety checks have been added when transitioning from
15791 the initial RAM disk to the main system to avoid accidental
15794 * /etc/crypttab entries now understand the new keyfile-offset=
15797 * systemctl -t can now be used to filter by unit load state.
15799 * The journal C API gained the new sd_journal_wait() call to
15800 make writing synchronous journal clients easier.
15802 * journalctl gained the new -D switch to show journals from a
15803 specific directory.
15805 * journalctl now displays a special marker between log
15806 messages of two different boots.
15808 * The journal is now explicitly flushed to /var via a service
15809 systemd-journal-flush.service, rather than implicitly simply
15810 by seeing /var/log/journal to be writable.
15812 * journalctl (and the journal C APIs) can now match for much
15813 more complex expressions, with alternatives and
15816 * When transitioning from the initial RAM disk to the main
15817 system we will now kill all processes in a killing spree to
15818 ensure no processes stay around by accident.
15820 * Three new specifiers may be used in unit files: %u, %h, %s
15821 resolve to the user name, user home directory resp. user
15822 shell. This is useful for running systemd user instances.
15824 * We now automatically rotate journal files if their data
15825 object hash table gets a fill level > 75%. We also size the
15826 hash table based on the configured maximum file size. This
15827 together should lower hash collisions drastically and thus
15828 speed things up a bit.
15830 * journalctl gained the new "--header" switch to introspect
15831 header data of journal files.
15833 * A new setting SystemCallFilters= has been added to services which may
15834 be used to apply deny lists or allow lists to system calls. This is
15835 based on SECCOMP Mode 2 of Linux 3.5.
15837 * nspawn gained a new --link-journal= switch (and quicker: -j)
15838 to link the container journal with the host. This makes it
15839 very easy to centralize log viewing on the host for all
15840 guests while still keeping the journal files separated.
15842 * Many bugfixes and optimizations
15844 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
15845 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
15846 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
15851 * Several tools now understand kernel command line arguments,
15852 which are only read when run in an initial RAM disk. They
15853 usually follow closely their normal counterparts, but are
15856 * There's a new tool to analyze the readahead files that are
15857 automatically generated at boot. Use:
15859 /usr/lib/systemd/systemd-readahead analyze /.readahead
15861 * We now provide an early debug shell on tty9 if this enabled. Use:
15863 systemctl enable debug-shell.service
15865 * All plymouth related units have been moved into the Plymouth
15866 package. Please make sure to upgrade your Plymouth version
15869 * systemd-tmpfiles now supports getting passed the basename of
15870 a configuration file only, in which case it will look for it
15871 in all appropriate directories automatically.
15873 * udevadm info now takes a /dev or /sys path as argument, and
15874 does the right thing. Example:
15876 udevadm info /dev/sda
15877 udevadm info /sys/class/block/sda
15879 * systemctl now prints a warning if a unit is stopped but a
15880 unit that might trigger it continues to run. Example: a
15881 service is stopped but the socket that activates it is left
15884 * "systemctl status" will now mention if the log output was
15885 shortened due to rotation since a service has been started.
15887 * The journal API now exposes functions to determine the
15888 "cutoff" times due to rotation.
15890 * journald now understands SIGUSR1 and SIGUSR2 for triggering
15891 immediately flushing of runtime logs to /var if possible,
15892 resp. for triggering immediate rotation of the journal
15895 * It is now considered an error if a service is attempted to
15896 be stopped that is not loaded.
15898 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
15900 * systemd-analyze now supports Python 3
15902 * tmpfiles now supports cleaning up directories via aging
15903 where the first level dirs are always kept around but
15904 directories beneath it automatically aged. This is enabled
15905 by prefixing the age field with '~'.
15907 * Seat objects now expose CanGraphical, CanTTY properties
15908 which is required to deal with very fast bootups where the
15909 display manager might be running before the graphics drivers
15910 completed initialization.
15912 * Seat objects now expose a State property.
15914 * We now include RPM macros for service enabling/disabling
15915 based on the preset logic. We recommend RPM based
15916 distributions to make use of these macros if possible. This
15917 makes it simpler to reuse RPM spec files across
15920 * We now make sure that the collected systemd unit name is
15921 always valid when services log to the journal via
15924 * There's a new man page kernel-command-line(7) detailing all
15925 command line options we understand.
15927 * The fstab generator may now be disabled at boot by passing
15928 fstab=0 on the kernel command line.
15930 * A new kernel command line option modules-load= is now understood
15931 to load a specific kernel module statically, early at boot.
15933 * Unit names specified on the systemctl command line are now
15934 automatically escaped as needed. Also, if file system or
15935 device paths are specified they are automatically turned
15936 into the appropriate mount or device unit names. Example:
15938 systemctl status /home
15939 systemctl status /dev/sda
15941 * The SysVConsole= configuration option has been removed from
15942 system.conf parsing.
15944 * The SysV search path is no longer exported on the D-Bus
15947 * The Names= option has been removed from unit file parsing.
15949 * There's a new man page bootup(7) detailing the boot process.
15951 * Every unit and every generator we ship with systemd now
15952 comes with full documentation. The self-explanatory boot is
15955 * A couple of services gained "systemd-" prefixes in their
15956 name if they wrap systemd code, rather than only external
15957 code. Among them fsck@.service which is now
15958 systemd-fsck@.service.
15960 * The HaveWatchdog property has been removed from the D-Bus
15963 * systemd.confirm_spawn= on the kernel command line should now
15966 * There's a new man page crypttab(5) which details all options
15967 we actually understand.
15969 * systemd-nspawn gained a new --capability= switch to pass
15970 additional capabilities to the container.
15972 * timedated will now read known NTP implementation unit names
15973 from /usr/lib/systemd/ntp-units.d/*.list,
15974 systemd-timedated-ntp.target has been removed.
15976 * journalctl gained a new switch "-b" that lists log data of
15977 the current boot only.
15979 * The notify socket is in the abstract namespace again, in
15980 order to support daemons which chroot() at start-up.
15982 * There is a new Storage= configuration option for journald
15983 which allows configuration of where log data should go. This
15984 also provides a way to disable journal logging entirely, so
15985 that data collected is only forwarded to the console, the
15986 kernel log buffer or another syslog implementation.
15988 * Many bugfixes and optimizations
15990 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
15991 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
15992 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
15993 Shawn Landden, Tom Gundersen
15997 * "systemctl help <unit>" now shows the man page if one is
16000 * Several new man pages have been added.
16002 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
16003 MaxLevelConsole= can now be specified in
16004 journald.conf. These options allow reducing the amount of
16005 data stored on disk or forwarded by the log level.
16007 * TimerSlackNSec= can now be specified in system.conf for
16008 PID1. This allows system-wide power savings.
16010 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
16011 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
16016 * logind is now capable of (optionally) handling power and
16017 sleep keys as well as the lid switch.
16019 * journalctl now understands the syntax "journalctl
16020 /usr/bin/avahi-daemon" to get all log output of a specific
16023 * CapabilityBoundingSet= in system.conf now also influences
16024 the capability bound set of usermode helpers of the kernel.
16026 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
16027 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
16028 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
16029 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
16033 * Note that we skipped 139 releases here in order to set the
16034 new version to something that is greater than both udev's
16035 and systemd's most recent version number.
16037 * udev: all udev sources are merged into the systemd source tree now.
16038 All future udev development will happen in the systemd tree. It
16039 is still fully supported to use the udev daemon and tools without
16040 systemd running, like in initramfs or other init systems. Building
16041 udev though, will require the *build* of the systemd tree, but
16042 udev can be properly *run* without systemd.
16044 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
16045 should be used to create dead device nodes as workarounds for broken
16048 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
16049 no longer supported. udev_monitor_new_from_netlink() needs to be
16050 used to subscribe to events.
16052 * udev: when udevd is started by systemd, processes which are left
16053 behind by forking them off of udev rules, are unconditionally cleaned
16054 up and killed now after the event handling has finished. Services or
16055 daemons must be started as systemd services. Services can be
16056 pulled-in by udev to get started, but they can no longer be directly
16057 forked by udev rules.
16059 * udev: the daemon binary is called systemd-udevd now and installed
16060 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
16061 to adapt to that, create symlink, or rename the binary after building
16064 * libudev no longer provides these symbols:
16065 udev_monitor_from_socket()
16066 udev_queue_get_failed_list_entry()
16067 udev_get_{dev,sys,run}_path()
16068 The versions number was bumped and symbol versioning introduced.
16070 * systemd-loginctl and systemd-journalctl have been renamed
16071 to loginctl and journalctl to match systemctl.
16073 * The config files: /etc/systemd/systemd-logind.conf and
16074 /etc/systemd/systemd-journald.conf have been renamed to
16075 logind.conf and journald.conf. Package updates should rename
16076 the files to the new names on upgrade.
16078 * For almost all files the license is now LGPL2.1+, changed
16079 from the previous GPL2.0+. Exceptions are some minor stuff
16080 of udev (which will be changed to LGPL2.1 eventually, too),
16081 and the MIT licensed sd-daemon.[ch] library that is suitable
16082 to be used as drop-in files.
16084 * systemd and logind now handle system sleep states, in
16085 particular suspending and hibernating.
16087 * logind now implements a sleep/shutdown/idle inhibiting logic
16088 suitable for a variety of uses. Soonishly Lennart will blog
16089 about this in more detail.
16091 * var-run.mount and var-lock.mount are no longer provided
16092 (which previously bind mounted these directories to their new
16093 places). Distributions which have not converted these
16094 directories to symlinks should consider stealing these files
16095 from git history and add them downstream.
16097 * We introduced the Documentation= field for units and added
16098 this to all our shipped units. This is useful to make it
16099 easier to explore the boot and the purpose of the various
16102 * All smaller setup units (such as
16103 systemd-vconsole-setup.service) now detect properly if they
16104 are run in a container and are skipped when
16105 appropriate. This guarantees an entirely noise-free boot in
16106 Linux container environments such as systemd-nspawn.
16108 * A framework for implementing offline system updates is now
16109 integrated, for details see:
16110 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
16112 * A new service type Type=idle is available now which helps us
16113 avoiding ugly interleaving of getty output and boot status
16116 * There's now a system-wide CapabilityBoundingSet= option to
16117 globally reduce the set of capabilities for the
16118 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
16119 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
16120 even CAP_NET_ADMIN system-wide for secure systems.
16122 * There are now system-wide DefaultLimitXXX= options to
16123 globally change the defaults of the various resource limits
16124 for all units started by PID 1.
16126 * Harald Hoyer's systemd test suite has been integrated into
16127 systemd which allows easy testing of systemd builds in qemu
16128 and nspawn. (This is really awesome! Ask us for details!)
16130 * The fstab parser is now implemented as generator, not inside
16133 * systemctl will now warn you if .mount units generated from
16134 /etc/fstab are out of date due to changes in fstab that
16135 have not been read by systemd yet.
16137 * systemd is now suitable for usage in initrds. Dracut has
16138 already been updated to make use of this. With this in place
16139 initrds get a slight bit faster but primarily are much
16140 easier to introspect and debug since "systemctl status" in
16141 the host system can be used to introspect initrd services,
16142 and the journal from the initrd is kept around too.
16144 * systemd-delta has been added, a tool to explore differences
16145 between user/admin configuration and vendor defaults.
16147 * PrivateTmp= now affects both /tmp and /var/tmp.
16149 * Boot time status messages are now much prettier and feature
16150 proper english language. Booting up systemd has never been
16153 * Read-ahead pack files now include the inode number of all
16154 files to pre-cache. When the inode changes the pre-caching
16155 is not attempted. This should be nicer to deal with updated
16156 packages which might result in changes of read-ahead
16159 * We now temporaritly lower the kernel's read_ahead_kb variable
16160 when collecting read-ahead data to ensure the kernel's
16161 built-in read-ahead does not add noise to our measurements
16162 of necessary blocks to pre-cache.
16164 * There's now RequiresMountsFor= to add automatic dependencies
16165 for all mounts necessary for a specific file system path.
16167 * MountAuto= and SwapAuto= have been removed from
16168 system.conf. Mounting file systems at boot has to take place
16171 * nspawn now learned a new switch --uuid= to set the machine
16172 ID on the command line.
16174 * nspawn now learned the -b switch to automatically search
16175 for an init system.
16177 * vt102 is now the default TERM for serial TTYs, upgraded from
16180 * systemd-logind now works on VT-less systems.
16182 * The build tree has been reorganized. The individual
16183 components now have directories of their own.
16185 * A new condition type ConditionPathIsReadWrite= is now available.
16187 * nspawn learned the new -C switch to create cgroups for the
16188 container in other hierarchies.
16190 * We now have support for hardware watchdogs, configurable in
16193 * The scheduled shutdown logic now has a public API.
16195 * We now mount /tmp as tmpfs by default, but this can be
16196 masked and /etc/fstab can override it.
16198 * Since udisks does not make use of /media anymore we are not
16199 mounting a tmpfs on it anymore.
16201 * journalctl gained a new --local switch to only interleave
16202 locally generated journal files.
16204 * We can now load the IMA policy at boot automatically.
16206 * The GTK tools have been split off into a systemd-ui.
16208 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
16209 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
16210 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
16211 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
16212 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
16213 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
16214 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
16215 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
16216 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
16221 * This is mostly a bugfix release
16223 * Support optional initialization of the machine ID from the
16224 KVM or container configured UUID.
16226 * Support immediate reboots with "systemctl reboot -ff"
16228 * Show /etc/os-release data in systemd-analyze output
16230 * Many bugfixes for the journal, including endianness fixes and
16231 ensuring that disk space enforcement works
16233 * sd-login.h is C++ compatible again
16235 * Extend the /etc/os-release format on request of the Debian
16238 * We now refuse non-UTF8 strings used in various configuration
16239 and unit files. This is done to ensure we do not pass invalid
16240 data over D-Bus or expose it elsewhere.
16242 * Register Mimo USB Screens as suitable for automatic seat
16245 * Read SELinux client context from journal clients in a race
16248 * Reorder configuration file lookup order. /etc now always
16249 overrides /run in order to allow the administrator to always
16250 and unconditionally override vendor-supplied or
16251 automatically generated data.
16253 * The various user visible bits of the journal now have man
16254 pages. We still lack man pages for the journal API calls
16257 * We now ship all man pages in HTML format again in the
16260 Contributions from: Dave Reisner, Dirk Eibach, Frederic
16261 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
16262 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
16267 * This is mostly a bugfix release
16269 * systems lacking /etc/os-release are no longer supported.
16271 * Various functionality updates to libsystemd-login.so
16273 * Track class of PAM logins to distinguish greeters from
16274 normal user logins.
16276 Contributions from: Kay Sievers, Lennart Poettering, Michael
16281 * This is an important bugfix release for v41.
16283 * Building man pages is now optional which should be useful
16284 for those building systemd from git but unwilling to install
16287 * Watchdog support for supervising services is now usable. In
16288 a future release support for hardware watchdogs
16289 (i.e. /dev/watchdog) will be added building on this.
16291 * Service start rate limiting is now configurable and can be
16292 turned off per service. When a start rate limit is hit a
16293 reboot can automatically be triggered.
16295 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
16297 Contributions from: Benjamin Franzke, Bill Nottingham,
16298 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
16299 Schmidt, Michał Górny, Piotr Drąg
16303 * The systemd binary is installed /usr/lib/systemd/systemd now;
16304 An existing /sbin/init symlink needs to be adapted with the
16307 * The code that loads kernel modules has been ported to invoke
16308 libkmod directly, instead of modprobe. This means we do not
16309 support systems with module-init-tools anymore.
16311 * Watchdog support is now already useful, but still not
16314 * A new kernel command line option systemd.setenv= is
16315 understood to set system wide environment variables
16316 dynamically at boot.
16318 * We now limit the set of capabilities of systemd-journald.
16320 * We now set SIGPIPE to ignore by default, since it only is
16321 useful in shell pipelines, and has little use in general
16322 code. This can be disabled with IgnoreSIPIPE=no in unit
16325 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
16326 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
16331 * This is mostly a bugfix release
16333 * We now expose the reason why a service failed in the
16334 "Result" D-Bus property.
16336 * Rudimentary service watchdog support (will be completed over
16337 the next few releases.)
16339 * When systemd forks off in order execute some service we will
16340 now immediately changes its argv[0] to reflect which process
16341 it will execute. This is useful to minimize the time window
16342 with a generic argv[0], which makes bootcharts more useful
16344 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
16345 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
16346 Mike Kazantsev, Ray Strode
16350 * This is mostly a test release, but incorporates many
16353 * New systemd-cgtop tool to show control groups by their
16356 * Linking against libacl for ACLs is optional again. If
16357 disabled, support tracking device access for active logins
16358 goes becomes unavailable, and so does access to the user
16359 journals by the respective users.
16361 * If a group "adm" exists, journal files are automatically
16362 owned by them, thus allow members of this group full access
16363 to the system journal as well as all user journals.
16365 * The journal now stores the SELinux context of the logging
16366 client for all entries.
16368 * Add C++ inclusion guards to all public headers
16370 * New output mode "cat" in the journal to print only text
16371 messages, without any meta data like date or time.
16373 * Include tiny X server wrapper as a temporary stop-gap to
16374 teach XOrg udev display enumeration. This is used by display
16375 managers such as gdm, and will go away as soon as XOrg
16376 learned native udev hotplugging for display devices.
16378 * Add new systemd-cat tool for executing arbitrary programs
16379 with STDERR/STDOUT connected to the journal. Can also act as
16380 BSD logger replacement, and does so by default.
16382 * Optionally store all locally generated coredumps in the
16383 journal along with meta data.
16385 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
16386 writing short strings to files (for usage for /sys), and for
16387 creating symlinks, character and block device nodes.
16389 * New unit file option ControlGroupPersistent= to make cgroups
16390 persistent, following the mechanisms outlined in
16391 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
16393 * Support multiple local RTCs in a sane way
16395 * No longer monopolize IO when replaying readahead data on
16396 rotating disks, since we might starve non-file-system IO to
16397 death, since fanotify() will not see accesses done by blkid,
16400 * Do not show kernel threads in systemd-cgls anymore, unless
16401 requested with new -k switch.
16403 Contributions from: Dan Horák, Kay Sievers, Lennart
16404 Poettering, Michal Schmidt
16408 * This is mostly a test release, but incorporates many
16411 * The git repository moved to:
16412 git://anongit.freedesktop.org/systemd/systemd
16413 ssh://git.freedesktop.org/git/systemd/systemd
16415 * First release with the journal
16416 https://0pointer.de/blog/projects/the-journal.html
16418 * The journal replaces both systemd-kmsg-syslogd and
16419 systemd-stdout-bridge.
16421 * New sd_pid_get_unit() API call in libsystemd-logind
16423 * Many systemadm clean-ups
16425 * Introduce remote-fs-pre.target which is ordered before all
16426 remote mounts and may be used to start services before all
16429 * Added Mageia support
16431 * Add bash completion for systemd-loginctl
16433 * Actively monitor PID file creation for daemons which exit in
16434 the parent process before having finished writing the PID
16435 file in the daemon process. Daemons which do this need to be
16436 fixed (i.e. PID file creation must have finished before the
16437 parent exits), but we now react a bit more gracefully to them.
16439 * Add colourful boot output, mimicking the well-known output
16440 of existing distributions.
16442 * New option PassCredentials= for socket units, for
16443 compatibility with a recent kernel ABI breakage.
16445 * /etc/rc.local is now hooked in via a generator binary, and
16446 thus will no longer act as synchronization point during
16449 * systemctl list-unit-files now supports --root=.
16451 * systemd-tmpfiles now understands two new commands: z, Z for
16452 relabelling files according to the SELinux database. This is
16453 useful to apply SELinux labels to specific files in /sys,
16454 among other things.
16456 * Output of SysV services is now forwarded to both the console
16457 and the journal by default, not only just the console.
16459 * New man pages for all APIs from libsystemd-login.
16461 * The build tree got reorganized and the build system is a
16462 lot more modular allowing embedded setups to specifically
16463 select the components of systemd they are interested in.
16465 * Support for Linux systems lacking the kernel VT subsystem is
16468 * configure's --with-rootdir= got renamed to
16469 --with-rootprefix= to follow the naming used by udev and
16472 * Unless specified otherwise we will now install to /usr instead
16473 of /usr/local by default.
16475 * Processes with '@' in argv[0][0] are now excluded from the
16476 final shut-down killing spree, following the logic explained
16478 https://systemd.io/ROOT_STORAGE_DAEMONS/
16480 * All processes remaining in a service cgroup when we enter
16481 the START or START_PRE states are now killed with
16482 SIGKILL. That means it is no longer possible to spawn
16483 background processes from ExecStart= lines (which was never
16484 supported anyway, and bad style).
16486 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
16487 reloading of units together.
16489 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
16490 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
16491 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
16492 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
16493 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek