1 systemd System and Service Manager
4 Backwards-incompatible changes:
6 * systemd-logind's background session class, which is by default used
7 for sessions started via cron, is no longer going to start systemd
8 --user instance. Note that because of that the user dbus instance
9 will not start either. Other session types remain unaffected.
13 Backwards-incompatible changes:
15 * The minimum kernel version required has been bumped from 3.13 to 4.15,
16 and CLOCK_BOOTTIME is now assumed to always exist.
18 * C11 with GNU extensions (aka "gnu11") is now used to build our
19 components. Public API headers are still restricted to ISO C89.
21 * In v250, a systemd-networkd feature that automatically configures
22 routes to addresses specified in AllowedIPs= was added and enabled by
23 default. However, this causes network connectivity issues in many
24 existing setups. Hence, it has been disabled by default since
25 systemd-stable 250.3. The feature can still be used by explicitly
26 configuring RouteTable= setting in .netdev files.
28 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
29 when a Condition*= check does not succeed, restoring the JobRemoved
30 signal to the behaviour it had before v250.
32 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
33 GetImageMetadataWithExtensions() have been fixed to provide an extra
34 return parameter, containing the actual extension release metadata.
35 The current implementation was judged to be broken and unusable, and
36 thus the usual procedure of adding a new set of methods was skipped,
37 and backward compatibility broken instead on the assumption that
38 nobody can be affected given the current state of this interface.
40 * All kernels supported by systemd mix bytes returned by RDRAND (or
41 similar) into the entropy pool at early boot. This means that on
42 those systems, even if /dev/urandom is not yet initialized, it still
43 returns bytes that are of at least RDRAND quality. For that reason,
44 we no longer have reason to invoke RDRAND from systemd itself, which
45 has historically been a source of bugs. Furthermore, kernels ≥5.6
46 provide the getrandom(GRND_INSECURE) interface for returning random
47 bytes before the entropy pool is initialized without warning into
48 kmsg, which is what we attempt to use if available. systemd's direct
49 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
50 running an older kernel may experience kmsg warnings that were not
51 seen with 250. For newer kernels, non-x86 systems, or older x86
52 systems, there should be no visible changes.
54 * sd-boot will now measure the kernel command line into TPM PCR 12
55 rather than PCR 8. This improves usefulness of the measurements on
56 systems where sd-boot is chainloaded from Grub. Grub measures all
57 commands its executes into PCR 8, which makes it very hard to use
58 reasonably, hence separate ourselves from that and use PCR 12
59 instead, which is what certain Ubuntu editions already do. To retain
60 compatibility with systems running older systemd systems a new meson
61 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
62 If enabled, the measurement is done twice: into the new-style PCR 12
63 *and* the old-style PCR 8. It's strongly advised to migrate all users
64 to PCR 12 for this purpose in the long run, as we intend to remove
65 this compatibility feature in two years' time.
67 * busctl capture now writes output in the newer pcapng format instead
70 * A udev rule that imported hwdb matches for USB devices with lowercase
71 hexadecimal vendor/product ID digits was added in systemd 250. This
72 has been reverted, since uppercase hexadecimal digits are supposed to
73 be used, and we already had a rule with the appropriate match.
75 Users might need to adjust their local hwdb entries.
77 * arch_prctl(2) has been moved to the @default set in the syscall filters
78 (as exposed via the SystemCallFilter= setting in service unit files).
79 It is apparently used by the linker now.
81 * The tmpfiles entries that create the /run/systemd/netif directory and
82 its subdirectories were moved from tmpfiles.d/systemd.conf to
83 tmpfiles.d/systemd-network.conf.
85 Users might need to adjust their files that override tmpfiles.d/systemd.conf
86 to account for this change.
88 * The requirement for Portable Services images to contain a well-formed
89 os-release file (i.e.: contain at least an ID field) is now enforced.
90 This applies to base images and extensions, and also to systemd-sysext.
92 Changes in the Boot Loader Specification, kernel-install and sd-boot:
94 * kernel-install's and bootctl's Boot Loader Specification Type #1
95 entry generation logic has been reworked. The user may now pick
96 explicitly by which "token" string to name the installation's boot
97 entries, via the new /etc/kernel/entry-token file or the new
98 --entry-token= switch to bootctl. By default — as before — the
99 entries are named after the local machine ID. However, in "golden
100 image" environments, where the machine ID shall be initialized on
101 first boot (as opposed to at installation time before first boot) the
102 machine ID will not be available at build time. In this case the
103 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
104 file) may be used to override the "token" for the entries, for
105 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
106 make the OS images independent of any machine ID, and ensure that the
107 images will not carry any identifiable information before first boot,
108 but on the other hand means that multiple parallel installations of
109 the very same image on the same disk cannot be supported.
111 Summary: if you are building golden images that shall acquire
112 identity information exclusively on first boot, make sure to both
113 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
114 value of the IMAGE_ID= or ID= field of /etc/os-release or another
115 suitable identifier before deploying the image.
117 * The Boot Loader Specification has been extended with
118 /loader/entries.srel file located in the EFI System Partition (ESP)
119 that disambiguates the format of the entries in the /loader/entries/
120 directory (in order to discern them from incompatible uses of this
121 directory by other projects). For entries that follow the
122 Specification, the string "type1" is stored in this file.
124 bootctl will now write this file automatically when installing the
125 systemd-boot boot loader.
127 * kernel-install supports a new initrd_generator= setting in
128 /etc/kernel/install.conf, that is exported as
129 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
130 allows choosing different initrd generators.
132 * kernel-install will now create a "staging area" (an initially-empty
133 directory to gather files for a Boot Loader Specification Type #1
134 entry). The path to this directory is exported as
135 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
136 drop files there instead of writing them directly to the final
137 location. kernel-install will move them when all files have been
138 prepared successfully.
140 * New option sort-key= has been added to the Boot Loader Specification
141 to override the sorting order of the entries in the boot menu. It is
142 read by sd-boot and bootctl, and will be written by kernel-install,
143 with the default value of IMAGE_ID= or ID= fields from
144 os-release. Together, this means that on multiboot installations,
145 entries should be grouped and sorted in a predictable way.
147 * The sort order of boot entries has been updated: entries which have
148 the new field sort-key= are sorted by it first, and all entries
149 without it are ordered later. After that, entries are sorted by
150 version so that newest entries are towards the beginning of the list.
152 * The kernel-install tool gained a new 'inspect' verb which shows the
153 paths and other settings used.
155 * sd-boot can now optionally beep when the menu is shown and menu
156 entries are selected, which can be useful on machines without a
157 working display. (Controllable via a loader.conf setting.)
159 * The --make-machine-id-directory= switch to bootctl has been replaced
160 by --make-entry-directory=, given that the entry directory is not
161 necessarily named after the machine ID, but after some other suitable
162 ID as selected via --entry-token= described above. The old name of
163 the option is still understood to maximize compatibility.
165 * 'bootctl list' gained support for a new --json= switch to output boot
166 menu entries in JSON format.
168 * 'bootctl is-installed' now supports the --graceful, and various verbs
169 omit output with the new option --quiet.
171 Changes in systemd-homed:
173 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
174 of activated home directories it manages (if the kernel and selected
175 file systems support it). So far it mapped three UID ranges: the
176 range from 0…60000, the user's own UID, and the range 60514…65534,
177 leaving everything else unmapped (in other words, the 16bit UID range
178 is mapped almost fully, with the exception of the UID subrange used
179 for systemd-homed users, with one exception: the user's own UID).
180 Unmapped UIDs may not be used for file ownership in the home
181 directory — any chown() attempts with them will fail. With this
182 release a fourth range is added to these mappings:
183 524288…1879048191. This range is the UID range intended for container
186 https://systemd.io/UIDS-GIDS
188 This range may be used for container managers that place container OS
189 trees in the home directory (which is a questionable approach, for
190 quota, permission, SUID handling and network file system
191 compatibility reasons, but nonetheless apparently commonplace). Note
192 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
193 UID assignments from the range are not managed or mapped by
194 `systemd-homed`, and must be managed with other mechanisms, in the
195 context of the local system.
197 Typically, a better approach to user namespacing in relevant
198 container managers would be to leave container OS trees on disk at
199 UID offset 0, but then map them to a dynamically allocated runtime
200 UID range via another UID mount map at container invocation
201 time. That way user namespace UID ranges become strictly a runtime
202 concept, and do not leak into persistent file systems, persistent
203 user databases or persistent configuration, thus greatly simplifying
204 handling, and improving compatibility with home directories intended
205 to be portable like the ones managed by systemd-homed.
207 Changes in shared libraries:
209 * A new libsystemd-core-<version>.so private shared library is
210 installed under /usr/lib/systemd/system, mirroring the existing
211 libsystemd-shared-<version>.so library. This allows the total
212 installation size to be reduced by binary code reuse.
214 * The <version> tag used in the name of libsystemd-shared.so and
215 libsystemd-core.so can be configured via the meson option
216 'shared-lib-tag'. Distributions may build subsequent versions of the
217 systemd package with unique tags (e.g. the full package version),
218 thus allowing multiple installations of those shared libraries to be
219 available at the same time. This is intended to fix an issue where
220 programs that link to those libraries would fail to execute because
221 they were installed earlier or later than the appropriate version of
224 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
225 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
226 format instead of as a simple series of hex characters.
228 * The sd-device API gained two new calls sd_device_new_from_devname()
229 and sd_device_new_from_path() which permit allocating an sd_device
230 object from a device node name or file system path.
232 * sd-device also gained a new call sd_device_open() which will open the
233 device node associated with a device for which an sd_device object
234 has been allocated. The call is supposed to address races around
235 device nodes being removed/recycled due to hotplug events, or media
236 change events: the call checks internally whether the major/minor of
237 the device node and the "diskseq" (in case of block devices) match
238 with the metadata loaded in the sd_device object, thus ensuring that
239 the device once opened really matches the provided sd_device object.
241 Changes in PID1, systemctl, and systemd-oomd:
243 * A new set of service monitor environment variables will be passed to
244 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
245 handler unit as OnFailure=/OnSuccess=. The variables are:
246 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
247 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
248 handler needs to watch multiple units, use a templated handler.
250 * A new ExtensionDirectories= setting in service unit files allows
251 system extensions to be loaded from a directory. (It is similar to
252 ExtensionImages=, but takes paths to directories, instead of
255 'portablectl attach --extension=' now also accepts directory paths.
257 * The user.delegate and user.invocation_id extended attributes on
258 cgroups are used in addition to trusted.delegate and
259 trusted.invocation_id. The latter pair requires privileges to set,
260 but the former doesn't and can be also set by the unprivileged user
263 (Only supported on kernels ≥5.6.)
265 * Units that were killed by systemd-oomd will now have a service result
266 of 'oom-kill'. The number of times a service was killed is tallied
267 in the 'user.oomd_ooms' extended attribute.
269 The OOMPolicy= unit file setting is now also honoured by
272 * In unit files the new %y/%Y specifiers can be used to refer to
273 normalized unit file path, which is particularly useful for symlinked
276 The new %q specifier resolves to the pretty hostname
277 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
279 The new %d specifier resolves to the credentials directory of a
280 service (same as $CREDENTIALS_DIRECTORY).
282 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
283 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
284 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
285 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
286 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
287 MountFlags= service settings now also work in unprivileged user
288 services, i.e. those run by the user's --user service manager, as long
289 as user namespaces are enabled on the system.
291 * Services with Restart=always and a failing ExecCondition= will no
292 longer be restarted, to bring ExecCondition= behaviour in line with
293 Condition*= settings.
295 * LoadCredential= now accepts a directory as the argument; all files
296 from the directory will be loaded as credentials.
298 * A new D-Bus property ControlGroupId is now exposed on service units,
299 that encapsulates the service's numeric cgroup ID that newer kernels
300 assign to each cgroup.
302 * PID 1 gained support for configuring the "pre-timeout" of watchdog
303 devices and the associated governor, via the new
304 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
305 options in /etc/systemd/system.conf.
307 * systemctl's --timestamp= option gained a new choice "unix", to show
308 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
310 * A new "taint" flag named "old-kernel" is introduced which is set when
311 the kernel systemd runs on is older then the current baseline version
312 (see above). The flag is shown in "systemctl status" output.
314 * Two additional taint flags "short-uid-range" and "short-gid-range"
315 have been added as well, which are set when systemd notices it is run
316 within a userns namespace that does not define the full 0…65535 UID
319 * A new "unmerged-usr" taint flag has been added that is set whenever
320 running on systems where /bin/ + /sbin/ are *not* symlinks to their
321 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
324 * Generators invoked by PID 1 will now have a couple of useful
325 environment variables set describing the execution context a
326 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
327 system service manager, or from the per-user service
328 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
329 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
330 systemd considers the current boot to be a "first"
331 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
332 detected and which type of hypervisor/container
333 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
336 * PID 1 will now automatically pick up system credentials from qemu's
337 fw_cfg interface, thus allowing passing arbitrary data into VM
338 systems similar to how this is already supported for passing them
339 into `systemd-nspawn` containers. Credentials may now also be passed
340 in via the new kernel command line option `systemd.set_credential=`
341 (note that kernel command line options are world-readable during
342 runtime, and only useful for credentials that require no
343 confidentiality). The credentials that can be passed to unified
344 kernels that use the `systemd-stub` UEFI stub are now similarly
345 picked up automatically. Automatic importing of system credentials
346 this way can be turned off via the new
347 `systemd.import_credentials=no` kernel command line option.
349 * LoadCredential= will now automatically look for credentials in the
350 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
351 the argument is not an absolute path. Similarly,
352 LoadCredentialEncrypted= will check the same directories plus
353 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
354 /usr/lib/credstore.encrypted/. The idea is to use those directories
355 as the system-wide location for credentials that services should pick
358 * System and service credentials are described in great detail in a new
361 https://systemd.io/CREDENTIALS
363 Changes in systemd-journald:
365 * The journal JSON export format has been added to listed of stable
366 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
368 * journalctl --list-boots now supports JSON output and the --reverse option.
370 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
371 updated, BUILDING_IMAGES is new:
373 https://systemd.io/JOURNAL_EXPORT_FORMATS
374 https://systemd.io/BUILDING_IMAGES
378 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
379 calculators, etc.), the other AV production devices (DJ tables,
380 keypads, etc.) that should accessible to the seat owner user by
383 * udevadm trigger gained a new --prioritized-subsystem= option to
384 process certain subsystems (and all their parent devices) earlier.
386 systemd-udev-trigger.service now uses this new option to trigger
387 block and TPM devices first, hopefully making the boot a bit faster.
389 * udevadm trigger now implements --type=all, --initialized-match,
390 --initialized-nomatch to trigger both subsystems and devices, only
391 already-initialized devices, and only devices which haven't been
392 initialized yet, respectively.
394 * udevadm gained a new "wait" command for safely waiting for a specific
395 device to show up in the udev device database. This is useful in
396 scripts that asynchronously allocate a block device (e.g. through
397 repartitioning, or allocating a loopback device or similar) and need
398 to synchronize on the creation to complete.
400 * udevadm gained a new "lock" command for locking one or more block
401 devices while formatting it or writing a partition table to it. It is
402 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
403 usable in scripts dealing with block devices.
405 * udevadm info will show a couple of additional device fields in its
406 output, and will not apply a limited set of coloring to line types.
408 * udevadm info --tree will now show a tree of objects (i.e. devices and
409 suchlike) in the /sys/ hierarchy.
411 * Block devices will now get a new set of device symlinks in
412 /dev/disk/by-diskseq/<nr>, which may be used to reference block
413 device nodes via the kernel's "diskseq" value. Note that this does
414 not guarantee that opening a device by a symlink like this will
415 guarantee that the opened device actually matches the specified
416 diskseq value. To be safe against races, the actual diskseq value of
417 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
418 the one in the symlink path.
420 * .link files gained support for setting MDI/MID-X on a link.
422 * .link files gained support for [Match] Firmware= setting to match on
423 the device firmware description string. By mistake, it was previously
424 only supported in .network files.
426 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
427 and [SR-IOV] section to configure SR-IOV virtual functions.
429 Changes in systemd-networkd:
431 * The default scope for unicast routes configured through [Route]
432 section is changed to "link", to make the behavior consistent with
433 "ip route" command. The manual configuration of [Route] Scope= is
436 * A new unit systemd-networkd-wait-online@<interface>.service has been
437 added that can be used to wait for a specific network interface to be
440 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
441 that configures the eponymous kernel attribute on the bridge.
443 * .netdev files now can be used to create virtual WLAN devices, and
444 configure various settings on them, via the [WLAN] section.
446 * .link/.network files gained support for [Match] Kind= setting to match
447 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
449 This value is also shown by 'networkctl status'.
451 * The Local= setting in .netdev files for various virtual network
452 devices gained support for specifying, in addition to the network
453 address, the name of a local interface which must have the specified
456 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
457 files, to configure tunnels in external mode (a.k.a. collect metadata
460 * [Network] L2TP= setting was removed. Please use interface specifier in
461 Local= setting in .netdev files of corresponding L2TP interface.
463 * New [DHCPServer] BootServerName=, BootServerAddress=, and
464 BootFilename= settings can be used to configure the server address,
465 server name, and file name sent in the DHCP packet (e.g. to configure
468 Changes in systemd-resolved:
470 * systemd-resolved is started earlier (in sysinit.target), so it
471 available earlier and will also be started in the initrd if installed
474 Changes in disk encryption:
476 * systemd-cryptenroll can now control whether to require the user to
477 enter a PIN when using TPM-based unlocking of a volume via the new
478 --tpm2-with-pin= option.
480 Option tpm2-pin= can be used in /etc/crypttab.
482 * When unlocking devices via TPM, TPM2 parameter encryption is now
483 used, to ensure that communication between CPU and discrete TPM chips
484 cannot be eavesdropped to acquire disk encryption keys.
486 * A new switch --fido2-credential-algorithm= has been added to
487 systemd-cryptenroll allowing selection of the credential algorithm to
488 use when binding encryption to FIDO2 tokens.
490 Changes in systemd-hostnamed:
492 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
493 to override the values gleaned from the hwdb.
495 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
496 /sys/class/dmi/id) to override the chassis that is reported by
499 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
500 for reading the hardware serial number, as reportd by DMI. It also
501 exposes a new method D-Bus property FirmwareVersion that encode the
502 firmware version of the system.
504 Changes in other components:
506 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
507 handling with the values that were configured during systemd build
508 (if /etc/locale.conf has not been created through some other
509 mechanism). This means that /etc/locale.conf should always have
510 reasonable contents and we avoid a potential mismatch in defaults.
512 * The userdbctl tool will now show UID range information as part of the
515 * A new build-time configuration setting default-user-shell= can be
516 used to set the default shell for user records and nspawn shell
517 invocations (instead of of the default /bin/bash).
519 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
520 information dynamically at runtime via IPC.
522 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
523 whether a functioning TPM2 infrastructure is available, i.e. if
524 firmware, kernel driver and systemd all have TPM2 support enabled and
527 * The systemd-creds tool gained support for generating encrypted
528 credentials that are using an empty encryption key. While this
529 provides no integrity nor confidentiality it's useful to implement
530 codeflows that work the same on TPM-ful and TPM2-less systems. The
531 service manager will only accept credentials "encrypted" that way if
532 a TPM2 device cannot be detected, to ensure that credentials
533 "encrypted" like that cannot be used to trick TPM2 systems.
535 * When deciding whether to colorize output, all systemd programs now
536 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
539 * Meson's new install_tag feature is now in use for several components,
540 allowing to build and install select binaries only: pam, nss, devel
541 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
542 $ meson build systemd-boot
543 $ meson install --tags systemd-boot --no-rebuild
544 https://mesonbuild.com/Installing.html#installation-tags
546 * A new build configuration option has been added, to allow selecting the
547 default compression algorithm used by systemd-journald and systemd-coredump.
548 This allows to build-in support for decompressing all supported formats,
549 but choose a specific one for compression. E.g.:
550 $ meson -Ddefault-compression=xz
552 Experimental features:
554 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
555 loader.conf that implements booting Microsoft Windows from the
556 sd-boot in a way that first reboots the system, to reset the TPM
557 PCRs. This improves compatibility with BitLocker's TPM use, as the
558 PCRs will only record the Windows boot process, and not sd-boot
559 itself, thus retaining the PCR measurements not involving sd-boot.
560 Note that this feature is experimental for now, and is likely going
561 to be generalized and renamed in a future release, without retaining
562 compatibility with the current implementation.
564 * A new systemd-sysupdate component has been added that automatically
565 discovers, downloads, and installs A/B-style updates for the host
566 installation itself, or container images, portable service images,
567 and other assets. See the new systemd-sysupdate man page for updates.
569 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
570 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
571 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
572 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
573 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
574 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
575 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
576 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
577 David, David Bond, Davide Cavalca, David Tardon, davijosw,
578 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
579 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
580 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
581 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
582 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
583 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
584 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
585 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
586 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
587 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
588 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
589 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
590 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
591 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
592 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
593 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
594 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
595 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
596 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
597 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
598 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
599 Simon Ellmann, Sonali Srivastava, Stefan Seering,
600 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
601 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
602 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
603 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
604 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
605 Zbigniew Jędrzejewski-Szmek, наб
607 — Edinburgh, 2022-05-21
611 * Support for encrypted and authenticated credentials has been added.
612 This extends the credential logic introduced with v247 to support
613 non-interactive symmetric encryption and authentication, based on a
614 key that is stored on the /var/ file system or in the TPM2 chip (if
615 available), or the combination of both (by default if a TPM2 chip
616 exists the combination is used, otherwise the /var/ key only). The
617 credentials are automatically decrypted at the moment a service is
618 started, and are made accessible to the service itself in unencrypted
619 form. A new tool 'systemd-creds' encrypts credentials for this
620 purpose, and two new service file settings LoadCredentialEncrypted=
621 and SetCredentialEncrypted= configure such credentials.
623 This feature is useful to store sensitive material such as SSL
624 certificates, passwords and similar securely at rest and only decrypt
625 them when needed, and in a way that is tied to the local OS
626 installation or hardware.
628 * systemd-gpt-auto-generator can now automatically set up discoverable
629 LUKS2 encrypted swap partitions.
631 * The GPT Discoverable Partitions Specification has been substantially
632 extended with support for root and /usr/ partitions for the majority
633 of architectures systemd supports. This includes platforms that do
634 not natively support UEFI, because even though GPT is specified under
635 UEFI umbrella, it is useful on other systems too. Specifically,
636 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
637 Portable Services use the concept without requiring UEFI.
639 * The GPT Discoverable Partitions Specifications has been extended with
640 a new set of partitions that may carry PKCS#7 signatures for Verity
641 partitions, encoded in a simple JSON format. This implements a simple
642 mechanism for building disk images that are fully authenticated and
643 can be tested against a set of cryptographic certificates. This is
644 now implemented for the various systemd tools that can operate with
645 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
646 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
647 The PKCS#7 signatures are passed to the kernel (where they are
648 checked against certificates from the kernel keyring), or can be
649 verified against certificates provided in userspace (via a simple
650 drop-in file mechanism).
652 * systemd-dissect's inspection logic will now report for which uses a
653 disk image is intended. Specifically, it will display whether an
654 image is suitable for booting on UEFI or in a container (using
655 systemd-nspawn's --image= switch), whether it can be used as portable
656 service, or attached as system extension.
658 * The system-extension.d/ drop-in files now support a new field
659 SYSEXT_SCOPE= that may encode which purpose a system extension image
660 is for: one of "initrd", "system" or "portable". This is useful to
661 make images more self-descriptive, and to ensure system extensions
662 cannot be attached in the wrong contexts.
664 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
665 be used in portable service images to indicate which unit prefixes
668 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
669 now is able to decode images for non-native architectures as well.
670 This allows systemd-nspawn to boot images of non-native architectures
671 if the corresponding user mode emulator is installed and
672 systemd-binfmtd is running.
674 * systemd-logind gained new settings HandlePowerKeyLongPress=,
675 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
676 HandleHibernateKeyLongPress= which may be used to configure actions
677 when the relevant keys are pressed for more than 5s. This is useful
678 on devices that only have hardware for a subset of these keys. By
679 default, if the reboot key is pressed long the poweroff operation is
680 now triggered, and when the suspend key is pressed long the hibernate
681 operation is triggered. Long pressing the other two keys currently
682 does not trigger any operation by default.
684 * When showing unit status updates on the console during boot and
685 shutdown, and a service is slow to start so that the cylon animation
686 is shown, the most recent sd_notify() STATUS= text is now shown as
687 well. Services may use this to make the boot/shutdown output easier
688 to understand, and to indicate what precisely a service that is slow
689 to start or stop is waiting for. In particular, the per-user service
690 manager instance now reports what it is doing and which service it is
691 waiting for this way to the system service manager.
693 * The service manager will now re-execute on reception of the
694 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
695 only when running as PID 1. There was no signal to request this when
696 running as per-user service manager, i.e. as any other PID than 1.
697 SIGRTMIN+25 works for both system and user managers.
699 * The hardware watchdog logic in PID 1 gained support for operating
700 with the default timeout configured in the hardware, instead of
701 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
702 request this behavior.
704 * A new kernel command line option systemd.watchdog_sec= is now
705 understood which may be used to override the hardware watchdog
706 time-out for the boot.
708 * A new setting DefaultOOMScoreAdjust= is now supported in
709 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
710 to set the default process OOM score adjustment value for processes
711 started by the service manager. For per-user service managers this
712 now defaults to 100, but for per-system service managers is left as
713 is. This means that by default now services forked off the user
714 service manager are more likely to be killed by the OOM killer than
715 system services or the managers themselves.
717 * A new per-service setting RestrictFileSystems= as been added that
718 restricts the file systems a service has access to by their type.
719 This is based on the new BPF LSM of the Linux kernel. It provides an
720 effective way to make certain API file systems unavailable to
721 services (and thus minimizing attack surface). A new command
722 "systemd-analyze filesystems" has been added that lists all known
723 file system types (and how they are grouped together under useful
726 * Services now support a new setting RestrictNetworkInterfaces= for
727 restricting access to specific network interfaces.
729 * Service unit files gained new settings StartupAllowedCPUs= and
730 StartupAllowedMemoryNodes=. These are similar to their counterparts
731 without the "Startup" prefix and apply during the boot process
732 only. This is useful to improve boot-time behavior of the system and
733 assign resources differently during boot than during regular
734 runtime. This is similar to the preexisting StartupCPUWeight=
737 * Related to this: the various StartupXYZ= settings
738 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
739 during shutdown. The settings not prefixed with "Startup" hence apply
740 during regular runtime, and those that are prefixed like that apply
741 during boot and shutdown.
743 * A new per-unit set of conditions/asserts
744 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
745 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
746 pressure is above the configured threshold, using the kernel PSI
747 feature. For more details see systemd.unit(5) and
748 https://docs.kernel.org/accounting/psi.html
750 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
751 ProtectKernelLogs=yes can now be used.
753 * The default maximum numbers of inodes have been raised from 64k to 1M
754 for /dev/, and from 400k to 1M for /tmp/.
756 * The per-user service manager learnt support for communicating with
757 systemd-oomd to acquire OOM kill information.
759 * A new service setting ExecSearchPath= has been added that allows
760 changing the search path for executables for services. It affects
761 where we look for the binaries specified in ExecStart= and similar,
762 and the specified directories are also added the $PATH environment
763 variable passed to invoked processes.
765 * A new setting RuntimeRandomizedExtraSec= has been added for service
766 and scope units that allows extending the runtime time-out as
767 configured by RuntimeMaxSec= with a randomized amount.
769 * The syntax of the service unit settings RuntimeDirectory=,
770 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
771 if the specified value is now suffixed with a colon, followed by
772 another filename, the latter will be created as symbolic link to the
773 specified directory. This allows creating these service directories
774 together with alias symlinks to make them available under multiple
777 * Service unit files gained two new settings TTYRows=/TTYColumns= for
778 configuring rows/columns of the TTY device passed to
779 stdin/stdout/stderr of the service. This is useful to propagate TTY
780 dimensions to a virtual machine.
782 * A new service unit file setting ExitType= has been added that
783 specifies when to assume a service has exited. By default systemd
784 only watches the main process of a service. By setting
785 ExitType=cgroup it can be told to wait for the last process in a
788 * Automount unit files gained a new setting ExtraOptions= that can be
789 used to configure additional mount options to pass to the kernel when
790 mounting the autofs instance.
792 * "Urlification" (generation of ESC sequences that generate clickable
793 hyperlinks in modern terminals) may now be turned off altogether
796 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
797 settings that default to 200 and 2 s respectively. The ratelimit
798 ensures that a path unit cannot cause PID1 to busy-loop when it is
799 trying to trigger a service that is skipped because of a Condition*=
800 not being satisfied. This matches the configuration and behaviour of
803 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
804 as a plug-in for cryptsetup. This means the plain cryptsetup command
805 may now be used to unlock volumes set up this way.
807 * The TPM2 logic in cryptsetup will now automatically detect systems
808 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
809 updates the SHA1 banks. In such a case PCR policies will be
810 automatically bound to the latter, not the former. This makes the PCR
811 policies reliable, but of course do not provide the same level of
812 trust as SHA256 banks.
814 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
815 RSA primary keys in addition to ECC, improving compatibility with
816 TPM2 chips that do not support ECC. RSA keys are much slower to use
817 than ECC, and hence are only used if ECC is not available.
819 * /etc/crypttab gained support for a new token-timeout= setting for
820 encrypted volumes that allows configuration of the maximum time to
821 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
822 the logic will query the user for a regular passphrase/recovery key
825 * Support for activating dm-integrity volumes at boot via a new file
826 /etc/integritytab and the tool systemd-integritysetup have been
827 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
828 with dm-integrity instead of dm-crypt/dm-verity.
830 * The systemd-veritysetup-generator now understands a new usrhash=
831 kernel command line option for specifying the Verity root hash for
832 the partition backing the /usr/ file system. A matching set of
833 systemd.verity_usr_* kernel command line options has been added as
834 well. These all work similar to the corresponding options for the
837 * The sd-device API gained a new API call sd_device_get_diskseq() to
838 return the DISKSEQ property of a device structure. The "disk
839 sequence" concept is a new feature recently introduced to the Linux
840 kernel that allows detecting reuse cycles of block devices, i.e. can
841 be used to recognize when loopback block devices are reused for a
842 different purpose or CD-ROM drives get their media changed.
844 * A new unit systemd-boot-update.service has been added. If enabled
845 (the default) and the sd-boot loader is detected to be installed, it
846 is automatically updated to the newest version when out of date. This
847 is useful to ensure the boot loader remains up-to-date, and updates
848 automatically propagate from the OS tree in /usr/.
850 * sd-boot will now build with SBAT by default in order to facilitate
851 working with recent versions of Shim that require it to be present.
853 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
854 This is used to robustly generate boot entry titles for Windows.
856 * A new generic target unit factory-reset.target has been added. It is
857 hooked into systemd-logind similar in fashion to
858 reboot/poweroff/suspend/hibernate, and is supposed to be used to
859 initiate a factory reset operation. What precisely this operation
860 entails is up for the implementer to decide, the primary goal of the
861 new unit is provide a framework where to plug in the implementation
862 and how to trigger it.
864 * A new meson build-time option 'clock-valid-range-usec-max' has been
865 added which takes a time in µs and defaults to 15 years. If the RTC
866 time is noticed to be more than the specified time ahead of the
867 built-in epoch of systemd (which by default is the release timestamp
868 of systemd) it is assumed that the RTC is not working correctly, and
869 the RTC is reset to the epoch. (It already is reset to the epoch when
870 noticed to be before it.) This should increase the chance that time
871 doesn't accidentally jump too far ahead due to faulty hardware or
874 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
875 which may be used to automatically save the current system time to
876 disk in regular intervals. This is useful to maintain a roughly
877 monotonic clock even without RTC hardware and with some robustness
878 against abnormal system shutdown.
880 * systemd-analyze verify gained support for a pair of new --image= +
881 --root= switches for verifying units below a specific root
882 directory/image instead of on the host.
884 * systemd-analyze verify gained support for verifying unit files under
885 an explicitly specified unit name, independently of what the filename
888 * systemd-analyze verify gained a new switch --recursive-errors= which
889 controls whether to only fail on errors found in the specified units
890 or recursively any dependent units.
892 * systemd-analyze security now supports a new --offline mode for
893 analyzing unit files stored on disk instead of loaded units. It may
894 be combined with --root=/--image to analyze unit files under a root
895 directory or disk image. It also learnt a new --threshold= parameter
896 for specifying an exposure level threshold: if the exposure level
897 exceeds the specified value the call will fail. It also gained a new
898 --security-policy= switch for configuring security policies to
899 enforce on the units. A policy is a JSON file that lists which tests
900 shall be weighted how much to determine the overall exposure
901 level. Altogether these new features are useful for fully automatic
902 analysis and enforcement of security policies on unit files.
904 * systemd-analyze security gain a new --json= switch for JSON output.
906 * systemd-analyze learnt a new --quiet switch for reducing
907 non-essential output. It's honored by the "dot", "syscall-filter",
908 "filesystems" commands.
910 * systemd-analyze security gained a --profile= option that can be used
911 to take into account a portable profile when analyzing portable
912 services, since a lot of the security-related settings are enabled
915 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
916 files, binaries and executables and prints metadata information,
917 including the build-id and other info described on:
918 https://systemd.io/COREDUMP_PACKAGE_METADATA/
920 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
921 section, for automatically propagating DNS settings from other
924 * The static lease DHCP server logic in systemd-networkd may now serve
925 IP addresses outside of the configured IP pool range for the server.
927 * CAN support in systemd-networkd gained four new settings Loopback=,
928 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
929 control modes. It gained a number of further settings for tweaking
932 * The [CAN] section in .network file gained new TimeQuantaNSec=,
933 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
934 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
935 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
936 DataSyncJumpWidth= settings to control bit-timing processed by the
939 * DHCPv4 client support in systemd-networkd learnt a new Label= option
940 for configuring the address label to apply to configure IPv4
943 * The [IPv6AcceptRA] section of .network files gained support for a new
944 UseMTU= setting that may be used to control whether to apply the
945 announced MTU settings to the local interface.
947 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
948 setting to control whether the DHCPv4 client request and process the
951 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
952 [DHCPPrefixDelegation], as now the prefix delegation is also supported
953 with DHCPv4 protocol by enabling the Use6RD= setting.
955 * The [DHCPPrefixDelegation] section in .network file gained a new
956 setting UplinkInterface= to specify the upstream interface.
958 * The [DHCPv6] section in .network file gained a new setting
959 UseDelegatedPrefix= to control whether the delegated prefixes will be
960 propagated to the downstream interfaces.
962 * The [IPv6AcceptRA] section of .network files now understands two new
963 settings UseGateway=/UseRoutePrefix= for explicitly configuring
964 whether to use the relevant fields from the IPv6 Router Advertisement
967 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
968 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
969 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
970 [IPv6AcceptRA] section to control when the DHCPv6 client is started
971 and how the delegated prefixes are handled by the DHCPv6 client.
973 * The IPv6Token= section in the [Network] section is deprecated, and
974 the [IPv6AcceptRA] section gained the Token= setting for its
975 replacement. The [IPv6Prefix] section also gained the Token= setting.
976 The Token= setting gained 'eui64' mode to explicitly configure an
977 address with the EUI64 algorithm based on the interface MAC address.
978 The 'prefixstable' mode can now optionally take a secret key. The
979 Token= setting in the [DHCPPrefixDelegation] section now supports all
980 algorithms supported by the same settings in the other sections.
982 * The [RoutingPolicyRule] section of .network file gained a new
983 SuppressInterfaceGroup= setting.
985 * The IgnoreCarrierLoss= setting in the [Network] section of .network
986 files now allows a duration to be specified, controlling how long to
987 wait before reacting to carrier loss.
989 * The [DHCPServer] section of .network file gained a new Router=
990 setting to specify the router address.
992 * The [CAKE] section of .network files gained various new settings
993 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
994 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
995 and UseRawPacketSize= for configuring CAKE.
997 * systemd-networkd now ships with new default .network files:
998 80-container-vb.network which matches host-side network bridge device
999 created by systemd-nspawn's --network-bridge or --network-zone
1000 switch, and 80-6rd-tunnel.network which matches automatically created
1001 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
1003 * systemd-networkd's handling of Endpoint= resolution for WireGuard
1004 interfaces has been improved.
1006 * systemd-networkd will now automatically configure routes to addresses
1007 specified in AllowedIPs=. This feature can be controlled via
1008 RouteTable= and RouteMetric= settings in [WireGuard] or
1009 [WireGuardPeer] sections.
1011 * systemd-networkd will now once again automatically generate persistent
1012 MAC addresses for batadv and bridge interfaces. Users can disable this
1013 by using MACAddress=none in .netdev files.
1015 * systemd-networkd and systemd-udevd now support IP over InfiniBand
1016 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
1017 systemd.netdev files gained the [IPoIB] section.
1019 * systemd-networkd and systemd-udevd now support net.ifname-policy=
1020 option on the kernel command-line. This is implemented through the
1021 systemd-network-generator service that automatically generates
1022 appropriate .link, .network, and .netdev files.
1024 * The various systemd-udevd "ethtool" buffer settings now understand
1025 the special value "max" to configure the buffers to the maximum the
1028 * systemd-udevd's .link files may now configure a large variety of
1029 NIC coalescing settings, plus more hardware offload settings.
1031 * .link files gained a new WakeOnLanPassword= setting in the [Link]
1032 section that allows to specify a WoL "SecureOn" password on hardware
1035 * systemd-nspawn's --setenv= switch now supports an additional syntax:
1036 if only a variable name is specified (i.e. without being suffixed by
1037 a '=' character and a value) the current value of the environment
1038 variable is propagated to the container. e.g. --setenv=FOO will
1039 lookup the current value of $FOO in the environment, and pass it down
1040 to the container. Similar behavior has been added to homectl's,
1041 machinectl's and systemd-run's --setenv= switch.
1043 * systemd-nspawn gained a new switch --suppress-sync= which may be used
1044 to optionally suppress the effect of the sync()/fsync()/fdatasync()
1045 system calls for the container payload. This is useful for build
1046 system environments where safety against abnormal system shutdown is
1047 not essential as all build artifacts can be regenerated any time, but
1048 the performance win is beneficial.
1050 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
1051 same value that PID 1 uses for most forked off processes.
1053 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
1054 uidmap/nouidmap options as last parameter. If "uidmap" is used the
1055 bind mounts are created with UID mapping taking place that ensures
1056 the host's file ownerships are mapped 1:1 to container file
1057 ownerships, even if user namespacing is used. This way
1058 files/directories bound into containers will no longer show up as
1059 owned by the nobody user as they typically did if no special care was
1060 taken to shift them manually.
1062 * When discovering Windows installations sd-boot will now attempt to
1063 show the Windows version.
1065 * The color scheme to use in sd-boot may now be configured at
1068 * sd-boot gained the ability to change screen resolution during
1069 boot-time, by hitting the "r" key. This will cycle through available
1070 resolutions and save the last selection.
1072 * sd-boot learnt a new hotkey "f". When pressed the system will enter
1073 firmware setup. This is useful in environments where it is difficult
1074 to hit the right keys early enough to enter the firmware, and works
1075 on any firmware regardless which key it natively uses.
1077 * sd-boot gained support for automatically booting into the menu item
1078 selected on the last boot (using the "@saved" identifier for menu
1081 * sd-boot gained support for automatically loading all EFI drivers
1082 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
1083 Partition (ESP). These drivers are loaded before the menu entries are
1084 loaded. This is useful e.g. to load additional file system drivers
1085 for the XBOOTLDR partition.
1087 * systemd-boot will now paint the input cursor on its own instead of
1088 relying on the firmware to do so, increasing compatibility with broken
1089 firmware that doesn't make the cursor reasonably visible.
1091 * sd-boot now embeds a .osrel PE section like we expect from Boot
1092 Loader Specification Type #2 Unified Kernels. This means sd-boot
1093 itself may be used in place of a Type #2 Unified Kernel. This is
1094 useful for debugging purposes as it allows chain-loading one a
1095 (development) sd-boot instance from another.
1097 * sd-boot now supports a new "devicetree" field in Boot Loader
1098 Specification Type #1 entries: if configured the specified device
1099 tree file is installed before the kernel is invoked. This is useful
1100 for installing/applying new devicetree files without updating the
1103 * Similarly, sd-stub now can read devicetree data from a PE section
1104 ".dtb" and apply it before invoking the kernel.
1106 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
1107 gained the ability to pick up credentials and sysext files, wrap them
1108 in a cpio archive, and pass as an additional initrd to the invoked
1109 Linux kernel, in effect placing those files in the /.extra/ directory
1110 of the initrd environment. This is useful to implement trusted initrd
1111 environments which are fully authenticated but still can be extended
1112 (via sysexts) and parameterized (via encrypted/authenticated
1113 credentials, see above).
1115 Credentials can be located next to the kernel image file (credentials
1116 specific to a single boot entry), or in one of the shared directories
1117 (credentials applicable to multiple boot entries).
1119 * sd-stub now comes with a full man page, that explains its feature set
1120 and how to combine a kernel image, an initrd and the stub to build a
1121 complete EFI unified kernel image, implementing Boot Loader
1122 Specification Type #2.
1124 * sd-stub may now provide the initrd to the executed kernel via the
1125 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
1126 non-x86 architectures.
1128 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
1129 may be used to set the boot menu time-out of the boot loader (for all
1130 or just the subsequent boot).
1132 * bootctl and kernel-install will now read variables
1133 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
1134 /etc/kernel/install.conf. When set, it specifies the layout to use
1135 for installation directories on the boot partition, so that tools
1136 don't need to guess it based on the already-existing directories. The
1137 only value that is defined natively is "bls", corresponding to the
1139 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
1140 kernel-install that implement a different layout can declare other
1141 values for this variable.
1143 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
1144 assumption that if the user installed sd-boot to the ESP, they intend
1145 to use the entry layout understood by sd-boot. It'll also write
1146 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
1147 (and it wasn't specified in the config file yet). Similarly,
1148 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
1149 wasn't specified in the config file yet). Effectively, those changes
1150 mean that the machine-id used for boot loader entry installation is
1151 "frozen" upon first use and becomes independent of the actual
1154 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
1155 images created for distribution ("golden images") are built with no
1156 machine-id, so that a unique machine-id can be created on the first
1157 boot. But those images may contain boot loader entries with the
1158 machine-id used during build included in paths. Using a "frozen"
1159 value allows unambiguously identifying entries that match the
1160 specific installation, while still permitting parallel installations
1163 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
1164 kernel-install to guess the installation layout. This fixes the
1165 problem where a (possibly empty) directory in the boot partition is
1166 created from a different layout causing kernel-install plugins to
1167 assume the wrong layout. A particular example of how this may happen
1168 is the grub2 package in Fedora which includes directories under /boot
1169 directly in its file list. Various other packages pull in grub2 as a
1170 dependency, so it may be installed even if unused, breaking
1171 installations that use the bls layout.
1173 * bootctl and systemd-bless-boot can now be linked statically.
1175 * systemd-sysext now optionally doesn't insist on extension-release.d/
1176 files being placed in the image under the image's file name. If the
1177 file system xattr user.extension-release.strict is set on the
1178 extension release file, it is accepted regardless of its name. This
1179 relaxes security restrictions a bit, as system extension may be
1180 attached under a wrong name this way.
1182 * udevadm's test-builtin command learnt a new --action= switch for
1183 testing the built-in with the specified action (in place of the
1186 * udevadm info gained new switches --property=/--value for showing only
1187 specific udev properties/values instead of all.
1189 * A new hwdb database has been added that contains matches for various
1190 types of signal analyzers (protocol analyzers, logic analyzers,
1191 oscilloscopes, multimeters, bench power supplies, etc.) that should
1192 be accessible to regular users.
1194 * A new hwdb database entry has been added that carries information
1195 about types of cameras (regular or infrared), and in which direction
1196 they point (front or back).
1198 * A new rule to allow console users access to rfkill by default has been
1201 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
1202 now also owned by the system group "sgx".
1204 * A new build-time meson option "extra-net-naming-schemes=" has been
1205 added to define additional naming schemes schemes for udev's network
1206 interface naming logic. This is useful for enterprise distributions
1207 and similar which want to pin the schemes of certain distribution
1208 releases under a specific name and previously had to patch the
1209 sources to introduce new named schemes.
1211 * The predictable naming logic for network interfaces has been extended
1212 to generate stable names from Xen netfront device information.
1214 * hostnamed's chassis property can now be sourced from chassis-type
1215 field encoded in devicetree (in addition to the existing DMI
1218 * systemd-cgls now optionally displays cgroup IDs and extended
1219 attributes for each cgroup. (Controllable via the new --xattr= +
1220 --cgroup-id= switches.)
1222 * coredumpctl gained a new --all switch for operating on all
1223 Journal files instead of just the local ones.
1225 * systemd-coredump will now use libdw/libelf via dlopen() rather than
1226 directly linking, allowing users to easily opt-out of backtrace/metadata
1227 analysis of core files, and reduce image sizes when this is not needed.
1229 * systemd-coredump will now analyze core files with libdw/libelf in a
1230 forked, sandboxed process.
1232 * systemd-homed will now try to unmount an activate home area in
1233 regular intervals once the user logged out fully. Previously this was
1234 attempted exactly once but if the home directory was busy for some
1235 reason it was not tried again.
1237 * systemd-homed's LUKS2 home area backend will now create a BSD file
1238 system lock on the image file while the home area is active
1239 (i.e. mounted). If a home area is found to be locked, logins are
1240 politely refused. This should improve behavior when using home areas
1241 images that are accessible via the network from multiple clients, and
1242 reduce the chance of accidental file system corruption in that case.
1244 * Optionally, systemd-homed will now drop the kernel buffer cache once
1245 a user has fully logged out, configurable via the new --drop-caches=
1248 * systemd-homed now makes use of UID mapped mounts for the home areas.
1249 If the kernel and used file system support it, files are now
1250 internally owned by the "nobody" user (i.e. the user typically used
1251 for indicating "this ownership is not mapped"), and dynamically
1252 mapped to the UID used locally on the system via the UID mapping
1253 mount logic of recent kernels. This makes migrating home areas
1254 between different systems cheaper because recursively chown()ing file
1255 system trees is no longer necessary.
1257 * systemd-homed's CIFS backend now optionally supports CIFS service
1258 names with a directory suffix, in order to place home directories in
1259 a subdirectory of a CIFS share, instead of the top-level directory.
1261 * systemd-homed's CIFS backend gained support for specifying additional
1262 mount options in the JSON user record (cifsExtraMountOptions field,
1263 and --cifs-extra-mount-options= homectl switch). This is for example
1264 useful for configuring mount options such as "noserverino" that some
1265 SMB3 services require (use that to run a homed home directory from a
1266 FritzBox SMB3 share this way).
1268 * systemd-homed will now default to btrfs' zstd compression for home
1269 areas. This is inspired by Fedora's recent decision to switch to zstd
1272 * Additional mount options to use when mounting the file system of
1273 LUKS2 volumes in systemd-homed has been added. Via the
1274 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
1275 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
1276 systemd-homed or via the luksExtraMountOptions user record JSON
1277 property. (Exposed via homectl --luks-extra-mount-options)
1279 * homectl's resize command now takes the special size specifications
1280 "min" and "max" to shrink/grow the home area to the minimum/maximum
1281 size possible, taking disk usage/space constraints and file system
1282 limitations into account. Resizing is now generally graceful: the
1283 logic will try to get as close to the specified size as possible, but
1284 not consider it a failure if the request couldn't be fulfilled
1287 * systemd-homed gained the ability to automatically shrink home areas
1288 on logout to their minimal size and grow them again on next
1289 login. This ensures that while inactive, a home area only takes up
1290 the minimal space necessary, but once activated, it provides
1291 sufficient space for the user's needs. This behavior is only
1292 supported if btrfs is used as file system inside the home area
1293 (because only for btrfs online growing/shrinking is implemented in
1294 the kernel). This behavior is now enabled by default, but may be
1295 controlled via the new --auto-resize-mode= setting of homectl.
1297 * systemd-homed gained support for automatically re-balancing free disk
1298 space among active home areas, in case the LUKS2 backends are used,
1299 and no explicit disk size was requested. This way disk space is
1300 automatically managed and home areas resized in regular intervals and
1301 manual resizing when disk space becomes scarce should not be
1302 necessary anymore. This behavior is only supported if btrfs is used
1303 within the home areas (as only then online shrinking and growing is
1304 supported), and may be configured via the new rebalanceWeight JSON
1305 user record field (as exposed via the new --rebalance-weight= homectl
1306 setting). Re-balancing is mostly automatic, but can also be requested
1307 explicitly via "homectl rebalance", which is synchronous, and thus
1308 may be used to wait until the rebalance run is complete.
1310 * userdbctl gained a --json= switch for configured the JSON formatting
1311 to use when outputting user or group records.
1313 * userdbctl gained a new --multiplexer= switch for explicitly
1314 configuring whether to use the systemd-userdbd server side user
1315 record resolution logic.
1317 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
1318 for chaining up another command to execute after completing the
1319 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
1320 configuration of a single command to invoke, this maybe used to
1321 invoke multiple: first userdbctl's own implementation, and then any
1322 other also configured in the command line.
1324 * The sd-event API gained a new function sd_event_add_inotify_fd() that
1325 is similar to sd_event_add_inotify() but accepts a file descriptor
1326 instead of a path in the file system for referencing the inode to
1329 * The sd-event API gained a new function
1330 sd_event_source_set_ratelimit_expire_callback() that may be used to
1331 define a callback function that is called whenever an event source
1332 leaves the rate limiting phase.
1334 * New documentation has been added explaining which steps are necessary
1335 to port systemd to a new architecture:
1337 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
1339 * The x-systemd.makefs option in /etc/fstab now explicitly supports
1340 ext2, ext3, and f2fs file systems.
1342 * Mount units and units generated from /etc/fstab entries with 'noauto'
1343 are now ordered the same as other units. Effectively, they will be
1344 started earlier (if something actually pulled them in) and stopped
1345 later, similarly to normal mount units that are part of
1346 fs-local.target. This change should be invisible to users, but
1347 should prevent those units from being stopped too early during
1350 * The systemd-getty-generator now honors a new kernel command line
1351 argument systemd.getty_auto= and a new environment variable
1352 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
1353 example useful to turn off gettys inside of containers or similar
1356 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
1357 (in addition to 127.0.0.53, as before). If DNS requests are sent to
1358 this address they are propagated in "bypass" mode only, i.e. are
1359 almost not processed locally, but mostly forwarded as-is to the
1360 current upstream DNS servers. This provides a stable DNS server
1361 address that proxies all requests dynamically to the right upstream
1362 DNS servers even if these dynamically change. This stub does not do
1363 mDNS/LLMNR resolution. However, it will translate look-ups to
1364 DNS-over-TLS if necessary. This new stub is particularly useful in
1365 container/VM environments, or for tethering setups: use DNAT to
1366 redirect traffic to any IP address to this stub.
1368 * systemd-importd now honors new environment variables
1369 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
1370 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
1371 generation, btrfs quota setup and disk synchronization.
1373 * systemd-importd and systemd-resolved can now be optionally built with
1374 OpenSSL instead of libgcrypt.
1376 * systemd-repart no longer requires OpenSSL.
1378 * systemd-sysusers will no longer create the redundant 'nobody' group
1379 by default, as the 'nobody' user is already created with an
1380 appropriate primary group.
1382 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
1384 * systemctl show-environment gained support for --output=json.
1386 * pam_systemd will now first try to use the X11 abstract socket, and
1387 fallback to the socket file in /tmp/.X11-unix/ only if that does not
1390 * systemd-journald will no longer go back to volatile storage
1391 regardless of configuration when its unit is restarted.
1393 * Initial support for the LoongArch architecture has been added (system
1394 call lists, GPT partition table UUIDs, etc).
1396 * systemd-journald's own logging messages are now also logged to the
1397 journal itself when systemd-journald logs to /dev/kmsg.
1399 * systemd-journald now re-enables COW for archived journal files on
1400 filesystems that support COW. One benefit of this change is that
1401 archived journal files will now get compressed on btrfs filesystems
1402 that have compression enabled.
1404 * systemd-journald now deduplicates fields in a single log message
1405 before adding it to the journal. In archived journal files, it will
1406 also punch holes for unused parts and truncate the file as
1407 appropriate, leading to reductions in disk usage.
1409 * journalctl --verify was extended with more informative error
1412 * More of sd-journal's functions are now resistant against journal file
1415 * The shutdown command learnt a new option --show, to display the
1418 * A LICENSES/ directory is now included in the git tree. It contains a
1419 README.md file that explains the licenses used by source files in
1420 this repository. It also contains the text of all applicable
1421 licenses as they appear on spdx.org.
1423 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
1424 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
1425 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
1426 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
1427 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
1428 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
1429 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
1430 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
1431 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
1432 Christian Brauner, Christian Göttsche, Christian Wehrli,
1433 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
1434 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
1435 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
1436 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
1437 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
1438 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
1439 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
1440 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
1441 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
1442 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
1443 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
1444 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
1445 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
1446 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
1447 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
1448 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
1449 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
1450 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
1451 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
1452 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
1453 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
1454 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
1455 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
1456 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
1457 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
1458 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
1459 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
1460 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
1461 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
1462 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
1463 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
1464 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
1465 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
1466 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
1467 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
1468 Дамјан Георгиевски, наб
1470 — Warsaw, 2021-12-23
1474 * When operating on disk images via the --image= switch of various
1475 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
1476 no 'root=' parameter on the kernel command line, and multiple
1477 suitable root or /usr/ partitions exist in the image, then a simple
1478 comparison inspired by strverscmp() is done on the GPT partition
1479 label, and the newest partition is picked. This permits a simple and
1480 generic whole-file-system A/B update logic where new operating system
1481 versions are dropped into partitions whose label is then updated with
1482 a matching version identifier.
1484 * systemd-sysusers now supports querying the passwords to set for the
1485 users it creates via the "credentials" logic introduced in v247: the
1486 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
1487 credentials are consulted for the password to use (either in UNIX
1488 hashed form, or literally). By default these credentials are inherited
1489 down from PID1 (which in turn imports it from a container manager if
1490 there is one). This permits easy configuration of user passwords
1491 during first boot. Example:
1493 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
1495 Note that systemd-sysusers operates in purely additive mode: it
1496 executes no operation if the declared users already exist, and hence
1497 doesn't set any passwords as effect of the command line above if the
1498 specified root user exists already in the image. (Note that
1499 --volatile=yes ensures it doesn't, though.)
1501 * systemd-firstboot now also supports querying various system
1502 parameters via the credential subsystems. Thus, as above this may be
1503 used to initialize important system parameters on first boot of
1504 previously unprovisioned images (i.e. images with a mostly empty
1507 * PID 1 may now show both the unit name and the unit description
1508 strings in its status output during boot. This may be configured with
1509 StatusUnitFormat=combined in system.conf or
1510 systemd.status-unit-format=combined on the kernel command line.
1512 * The systemd-machine-id-setup tool now supports a --image= switch for
1513 provisioning a machine ID file into an OS disk image, similar to how
1514 --root= operates on an OS file tree. This matches the existing switch
1515 of the same name for systemd-tmpfiles, systemd-firstboot, and
1516 systemd-sysusers tools.
1518 * Similarly, systemd-repart gained support for the --image= switch too.
1519 In combination with the existing --size= option, this makes the tool
1520 particularly useful for easily growing disk images in a single
1521 invocation, following the declarative rules included in the image
1524 * systemd-repart's partition configuration files gained support for a
1525 new switch MakeDirectories= which may be used to create arbitrary
1526 directories inside file systems that are created, before registering
1527 them in the partition table. This is useful in particular for root
1528 partitions to create mount point directories for other partitions
1529 included in the image. For example, a disk image that contains a
1530 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
1531 create /home/ and /var/ as empty directories in the root file system
1532 on its creation, so that the resulting image can be mounted
1533 immediately, even in read-only mode.
1535 * systemd-repart's CopyBlocks= setting gained support for the special
1536 value "auto". If used, a suitable matching partition on the booted OS
1537 is found as source to copy blocks from. This is useful when
1538 implementing replicating installers, that are booted from one medium
1539 and then stream their own root partition onto the target medium.
1541 * systemd-repart's partition configuration files gained support for a
1542 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
1543 GPT partition flags for the created partitions: this is useful for
1544 marking newly created partitions as read-only, or as not being
1545 subject for automatic mounting from creation on.
1547 * The /etc/os-release file has been extended with two new (optional)
1548 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
1549 information for OS images that are updated comprehensively and
1550 atomically as one image. Two new specifiers %M, %A now resolve to
1551 these two fields in the various configuration options that resolve
1554 * portablectl gained a new switch --extension= for enabling portable
1555 service images with extensions that follow the extension image
1556 concept introduced with v248, and thus allows layering multiple
1557 images when setting up the root filesystem of the service.
1559 * systemd-coredump will now extract ELF build-id information from
1560 processes dumping core and include it in the coredump report.
1561 Moreover, it will look for ELF .note.package sections with
1562 distribution packaging meta-information about the crashing process.
1563 This is useful to directly embed the rpm or deb (or any other)
1564 package name and version in ELF files, making it easy to match
1565 coredump reports with the specific package for which the software was
1566 compiled. This is particularly useful on environments with ELF files
1567 from multiple vendors, different distributions and versions, as is
1568 common today in our containerized and sand-boxed world. For further
1571 https://systemd.io/COREDUMP_PACKAGE_METADATA
1573 * A new udev hardware database has been added for FireWire devices
1576 * The "net_id" built-in of udev has been updated with three
1577 backwards-incompatible changes:
1579 - PCI hotplug slot names on s390 systems are now parsed as
1580 hexadecimal numbers. They were incorrectly parsed as decimal
1581 previously, or ignored if the name was not a valid decimal
1584 - PCI onboard indices up to 65535 are allowed. Previously, numbers
1585 above 16383 were rejected. This primarily impacts s390 systems,
1586 where values up to 65535 are used.
1588 - Invalid characters in interface names are replaced with "_".
1590 The new version of the net naming scheme is "v249". The previous
1591 scheme can be selected via the "net.naming-scheme=v247" kernel
1592 command line parameter.
1594 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
1595 NULL bus object, for which they will return false. Or in other words,
1596 an unallocated bus connection is neither ready nor open.
1598 * The sd-device API acquired a new API function
1599 sd_device_get_usec_initialized() that returns the monotonic time when
1600 the udev device first appeared in the database.
1602 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
1603 sd_device_get_trigger_uuid(). The former is similar to
1604 sd_device_trigger() but returns a randomly generated UUID that is
1605 associated with the synthetic uevent generated by the call. This UUID
1606 may be read from the sd_device object a monitor eventually receives,
1607 via the sd_device_get_trigger_uuid(). This interface requires kernel
1608 4.13 or above to work, and allows tracking a synthetic uevent through
1609 the entire device management stack. The "udevadm trigger --settle"
1610 logic has been updated to make use of this concept if available to
1611 wait precisely for the uevents it generates. "udevadm trigger" also
1612 gained a new parameter --uuid that prints the UUID for each generated
1615 * sd-device also gained new APIs sd_device_new_from_ifname() and
1616 sd_device_new_from_ifindex() for allocating an sd-device object for
1617 the specified network interface. The former accepts an interface name
1618 (either a primary or an alternative name), the latter an interface
1621 * The native Journal protocol has been documented. Clients may talk
1622 this as alternative to the classic BSD syslog protocol for locally
1623 delivering log records to the Journal. The protocol has been stable
1624 for a long time and in fact been implemented already in a variety
1625 of alternative client libraries. This documentation makes the support
1628 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
1630 * A new BPFProgram= setting has been added to service files. It may be
1631 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
1632 file, or a bind mount or symlink to one. This may be used to upload
1633 and manage BPF programs externally and then hook arbitrary systemd
1636 * The "home.arpa" domain that has been officially declared as the
1637 choice for domain for local home networks per RFC 8375 has been added
1638 to the default NTA list of resolved, since DNSSEC is generally not
1639 available on private domains.
1641 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
1643 * A new ManageForeignRoutingPolicyRules= setting has been added to
1644 .network files which may be used to exclude foreign-created routing
1645 policy rules from systemd-networkd management.
1647 * systemd-network-wait-online gained two new switches -4 and -6 that
1648 may be used to tweak whether to wait for only IPv4 or only IPv6
1651 * .network files gained a new RequiredFamilyForOnline= setting to
1652 fine-tune whether to require an IPv4 or IPv6 address in order to
1653 consider an interface "online".
1655 * networkctl will now show an over-all "online" state in the per-link
1658 * In .network files a new OutgoingInterface= setting has been added to
1659 specify the output interface in bridge FDB setups.
1661 * In .network files the Multipath group ID may now be configured for
1662 [NextHop] entries, via the new Group= setting.
1664 * The DHCP server logic configured in .network files gained a new
1665 setting RelayTarget= that turns the server into a DHCP server relay.
1666 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
1667 to further tweak the DHCP relay behaviour.
1669 * The DHCP server logic also gained a new ServerAddress= setting in
1670 .network files that explicitly specifies the server IP address to
1671 use. If not specified, the address is determined automatically, as
1674 * The DHCP server logic in systemd-networkd gained support for static
1675 DHCP leases, configurable via the [DHCPServerStaticLease]
1676 section. This allows explicitly mapping specific MAC addresses to
1677 fixed IP addresses and vice versa.
1679 * The RestrictAddressFamilies= setting in service files now supports a
1680 new special value "none". If specified sockets of all address
1681 families will be made unavailable to services configured that way.
1683 * systemd-fstab-generator and systemd-repart have been updated to
1684 support booting from disks that carry only a /usr/ partition but no
1685 root partition yet, and where systemd-repart can add it in on the
1686 first boot. This is useful for implementing systems that ship with a
1687 single /usr/ file system, and whose root file system shall be set up
1688 and formatted on a LUKS-encrypted volume whose key is generated
1689 locally (and possibly enrolled in the TPM) during the first boot.
1691 * The [Address] section of .network files now accepts a new
1692 RouteMetric= setting that configures the routing metric to use for
1693 the prefix route created as effect of the address configuration.
1694 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
1695 gained matching settings for their prefix routes. (The option of the
1696 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
1697 it conceptually belongs there; the old option is still understood for
1700 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
1703 * A new udev property ID_NET_DHCP_BROADCAST on network interface
1704 devices is now honoured by systemd-networkd, controlling whether to
1705 issue DHCP offers via broadcasting. This is used to ensure that s390
1706 layer 3 network interfaces work out-of-the-box with systemd-networkd.
1708 * nss-myhostname and systemd-resolved will now synthesize address
1709 records for a new special hostname "_outbound". The name will always
1710 resolve to the local IP addresses most likely used for outbound
1711 connections towards the default routes. On multi-homed hosts this is
1712 useful to have a stable handle referring to "the" local IP address
1713 that matters most, to the point where this is defined.
1715 * The Discoverable Partition Specification has been updated with a new
1716 GPT partition flag "grow-file-system" defined for its partition
1717 types. Whenever partitions with this flag set are automatically
1718 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
1719 of systemd-nspawn or other tools; and as opposed to explicit mounting
1720 via /etc/fstab), the file system within the partition is
1721 automatically grown to the full size of the partition. If the file
1722 system size already matches the partition size this flag has no
1723 effect. Previously, this functionality has been available via the
1724 explicit x-systemd.growfs mount option, and this new flag extends
1725 this to automatically discovered mounts. A new GrowFileSystem=
1726 setting has been added to systemd-repart drop-in files that allows
1727 configuring this partition flag. This new flag defaults to on for
1728 partitions automatically created by systemd-repart, except if they
1729 are marked read-only. See the specification for further details:
1731 https://systemd.io/DISCOVERABLE_PARTITIONS
1733 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
1734 section. If enabled (which is the default), and an NTP server address
1735 is acquired through a DHCP lease on this interface an explicit route
1736 to this address is created on this interface to ensure that NTP
1737 traffic to the NTP server acquired on an interface is also routed
1738 through that interface. The pre-existing RoutesToDNS= setting that
1739 implements the same for DNS servers is now enabled by default.
1741 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
1742 been added that may be used to restrict the network interfaces
1743 sockets created by the service may be bound to. This is implemented
1746 * A new ConditionFirmware= setting has been added to unit files to
1747 conditionalize on certain firmware features. At the moment it may
1748 check whether running on an UEFI system, a device.tree system, or if
1749 the system is compatible with some specified device-tree feature.
1751 * A new ConditionOSRelease= setting has been added to unit files to
1752 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
1753 operators may be used to check if some field has some specific value
1754 or do an alphanumerical comparison. Equality comparisons are useful
1755 for fields like ID, but relative comparisons for fields like
1756 VERSION_ID or IMAGE_VERSION.
1758 * hostnamed gained a new Describe() D-Bus method that returns a JSON
1759 serialization of the host data it exposes. This is exposed via
1760 "hostnamectl --json=" to acquire a host identity description in JSON.
1761 It's our intention to add a similar features to most services and
1762 objects systemd manages, in order to simplify integration with
1763 program code that can consume JSON.
1765 * Similarly, networkd gained a Describe() method on its Manager and
1766 Link bus objects. This is exposed via "networkctl --json=".
1768 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
1769 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
1770 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
1771 that is used both to get the value (when no argument is given), and
1772 to set the value (when an argument is specified). The old names
1773 continue to be supported for compatibility.
1775 * systemd-detect-virt and ConditionVirtualization= are now able to
1776 correctly identify Amazon EC2 environments.
1778 * The LogLevelMax= setting of unit files now applies not only to log
1779 messages generated *by* the service, but also to log messages
1780 generated *about* the service by PID 1. To suppress logs concerning a
1781 specific service comprehensively, set this option to a high log
1784 * bootctl gained support for a new --make-machine-id-directory= switch
1785 that allows precise control on whether to create the top-level
1786 per-machine directory in the boot partition that typically contains
1787 Type 1 boot loader entries.
1789 * During build SBAT data to include in the systemd-boot EFI PE binaries
1790 may be specified now.
1792 * /etc/crypttab learnt a new option "headless". If specified any
1793 requests to query the user interactively for passwords or PINs will
1794 be skipped. This is useful on systems that are headless, i.e. where
1795 an interactive user is generally not present.
1797 * /etc/crypttab also learnt a new option "password-echo=" that allows
1798 configuring whether the encryption password prompt shall echo the
1799 typed password and if so, do so literally or via asterisks. (The
1800 default is the same behaviour as before: provide echo feedback via
1803 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
1804 systemd-homed has been updated to allow explicit configuration of the
1805 "user presence" and "user verification" checks, as well as whether a
1806 PIN is required for authentication, via the new switches
1807 --fido2-with-user-presence=, --fido2-with-user-verification=,
1808 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
1809 features are available, and may be enabled or disabled depends on the
1812 * systemd-nspawn's --private-user= switch now accepts the special value
1813 "identity" which configures a user namespacing environment with an
1814 identity mapping of 65535 UIDs. This means the container UID 0 is
1815 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
1816 this doesn't appear to be useful, however it does reduce the attack
1817 surface a bit, since the resulting container will possess process
1818 capabilities only within its namespace and not on the host.
1820 * systemd-nspawn's --private-user-chown switch has been replaced by a
1821 more generic --private-user-ownership= switch that accepts one of
1822 three values: "chown" is equivalent to the old --private-user-chown,
1823 and "off" is equivalent to the absence of the old switch. The value
1824 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
1825 of files and directories of the underlying image to the chosen UID
1826 range for the container. "auto" is equivalent to "map" if UID mapping
1827 mount are supported, otherwise it is equivalent to "chown". The short
1828 -U switch systemd-nspawn now implies --private-user-ownership=auto
1829 instead of the old --private-user-chown. Effectively this means: if
1830 the backing file system supports UID mapping mounts the feature is
1831 now used by default if -U is used. Generally, it's a good idea to use
1832 UID mapping mounts instead of recursive chown()ing, since it allows
1833 running containers off immutable images (since no modifications of
1834 the images need to take place), and share images between multiple
1835 instances. Moreover, the recursive chown()ing operation is slow and
1836 can be avoided. Conceptually it's also a good thing if transient UID
1837 range uses do not leak into persistent file ownership anymore. TLDR:
1838 finally, the last major drawback of user namespacing has been
1839 removed, and -U should always be used (unless you use btrfs, where
1840 UID mapped mounts do not exist; or your container actually needs
1841 privileges on the host).
1843 * nss-systemd now synthesizes user and group shadow records in addition
1844 to the main user and group records. Thus, hashed passwords managed by
1845 systemd-homed are now accessible via the shadow database.
1847 * The userdb logic (and thus nss-systemd, and so on) now read
1848 additional user/group definitions in JSON format from the drop-in
1849 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
1850 /usr/lib/userdb/. This is a simple and powerful mechanism for making
1851 additional users available to the system, with full integration into
1852 NSS including the shadow databases. Since the full JSON user/group
1853 record format is supported this may also be used to define users with
1854 resource management settings and other runtime settings that
1855 pam_systemd and systemd-logind enforce at login.
1857 * The userdbctl tool gained two new switches --with-dropin= and
1858 --with-varlink= which can be used to fine-tune the sources used for
1859 user database lookups.
1861 * systemd-nspawn gained a new switch --bind-user= for binding a host
1862 user account into the container. This does three things: the user's
1863 home directory is bind mounted from the host into the container,
1864 below the /run/userdb/home/ hierarchy. A free UID is picked in the
1865 container, and a user namespacing UID mapping to the host user's UID
1866 installed. And finally, a minimal JSON user and group record (along
1867 with its hashed password) is dropped into /run/host/userdb/. These
1868 records are picked up automatically by the userdb drop-in logic
1869 describe above, and allow the user to login with the same password as
1870 on the host. Effectively this means: if host and container run new
1871 enough systemd versions making a host user available to the container
1872 is trivially simple.
1874 * systemd-journal-gatewayd now supports the switches --user, --system,
1875 --merge, --file= that are equivalent to the same switches of
1876 journalctl, and permit exposing only the specified subset of the
1879 * The OnFailure= dependency between units is now augmented with a
1880 implicit reverse dependency OnFailureOf= (this new dependency cannot
1881 be configured directly it's only created as effect of an OnFailure=
1882 dependency in the reverse order — it's visible in "systemctl show"
1883 however). Similar, Slice= now has an reverse dependency SliceOf=,
1884 that is also not configurable directly, but useful to determine all
1885 units that are members of a slice.
1887 * A pair of new dependency types between units PropagatesStopTo= +
1888 StopPropagatedFrom= has been added, that allows propagation of unit
1889 stop events between two units. It operates similar to the existing
1890 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
1892 * A new dependency type OnSuccess= has been added (plus the reverse
1893 dependency OnSuccessOf=, which cannot be configured directly, but
1894 exists only as effect of the reverse OnSuccess=). It is similar to
1895 OnFailure=, but triggers in the opposite case: when a service exits
1896 cleanly. This allows "chaining up" of services where one or more
1897 services are started once another service has successfully completed.
1899 * A new dependency type Upholds= has been added (plus the reverse
1900 dependency UpheldBy=, which cannot be configured directly, but exists
1901 only as effect of Upholds=). This dependency type is a stronger form
1902 of Wants=: if a unit has an UpHolds= dependency on some other unit
1903 and the former is active then the latter is started whenever it is
1904 found inactive (and no job is queued for it). This is an alternative
1905 to Restart= inside service units, but less configurable, and the
1906 request to uphold a unit is not encoded in the unit itself but in
1907 another unit that intends to uphold it.
1909 * The systemd-ask-password tool now also supports reading passwords
1910 from the credentials subsystem, via the new --credential= switch.
1912 * The systemd-ask-password tool learnt a new switch --emoji= which may
1913 be used to explicit control whether the lock and key emoji (🔐) is
1914 shown in the password prompt on suitable TTYs.
1916 * The --echo switch of systemd-ask-password now optionally takes a
1917 parameter that controls character echo. It may either show asterisks
1918 (default, as before), turn echo off entirely, or echo the typed
1919 characters literally.
1921 * The systemd-ask-password tool also gained a new -n switch for
1922 suppressing output of a trailing newline character when writing the
1923 acquired password to standard output, similar to /bin/echo's -n
1926 * New documentation has been added that describes the organization of
1927 the systemd source code tree:
1929 https://systemd.io/ARCHITECTURE
1931 * Units using ConditionNeedsUpdate= will no longer be activated in
1934 * It is now possible to list a template unit in the WantedBy= or
1935 RequiredBy= settings of the [Install] section of another template
1936 unit, which will be instantiated using the same instance name.
1938 * A new MemoryAvailable property is available for units. If the unit,
1939 or the slice(s) it is part of, have a memory limit set via MemoryMax=/
1940 MemoryHigh=, MemoryAvailable will indicate how much more memory the
1941 unit can claim before hitting the limit(s).
1943 * systemd-coredump will now try to stay below the cgroup memory limit
1944 placed on itself or one of the slices it runs under, if the storage
1945 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
1946 since files written on such filesystems count toward the cgroup memory
1947 limit. If there is not enough available memory in such cases to store
1948 the core file uncompressed, systemd-coredump will skip to compressed
1949 storage directly (if enabled) and it will avoid analyzing the core file
1950 to print backtrace and metadata in the journal.
1952 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
1953 of a path matches the configured expectations, and remove it if not.
1955 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
1956 specify which of the several available filesystem timestamps (access
1957 time, birth time, change time, modification time) to look at when
1958 deciding whether a path has aged enough to be cleaned.
1960 * A new IPv6StableSecretAddress= setting has been added to .network
1961 files, which takes an IPv6 address to use as secret for IPv6 address
1964 * The [DHCPServer] logic in .network files gained support for a new
1965 UplinkInterface= setting that permits configuration of the uplink
1966 interface name to propagate DHCP lease information from.
1968 * The WakeOnLan= setting in .link files now accepts a list of flags
1969 instead of a single one, to configure multiple wake-on-LAN policies.
1971 * User-space defined tracepoints (USDT) have been added to udev at
1972 strategic locations. This is useful for tracing udev behaviour and
1973 performance with bpftrace and similar tools.
1975 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
1976 setting a network timeout time.
1978 * If a system service is running in a new mount namespace (RootDirectory=
1979 and friends), all file systems will be mounted with MS_NOSUID by
1980 default, unless the system is running with SELinux enabled.
1982 * When enumerating time zones the timedatectl tool will now consult the
1983 'tzdata.zi' file shipped by the IANA time zone database package, in
1984 addition to 'zone1970.tab', as before. This makes sure time zone
1985 aliases are now correctly supported. Some distributions so far did
1986 not install this additional file, most do however. If you
1987 distribution does not install it yet, it might make sense to change
1990 * Intel HID rfkill event is no longer masked, since it's the only
1991 source of rfkill event on newer HP laptops. To have both backward and
1992 forward compatibility, userspace daemon needs to debounce duplicated
1993 events in a short time window.
1995 Contributions from: Aakash Singh, adrian5, Albert Brox,
1996 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
1997 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
1998 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
1999 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
2000 Christian Hesse, Daniel Schaefer, Dan Streetman,
2001 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
2002 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
2003 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
2004 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
2005 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
2006 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
2007 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
2008 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
2009 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
2010 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
2011 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
2012 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
2013 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
2014 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
2015 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
2016 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
2017 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
2018 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
2019 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
2020 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
2021 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
2022 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
2023 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
2024 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
2025 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
2027 — Edinburgh, 2021-07-07
2031 * A concept of system extension images is introduced. Such images may
2032 be used to extend the /usr/ and /opt/ directory hierarchies at
2033 runtime with additional files (even if the file system is read-only).
2034 When a system extension image is activated, its /usr/ and /opt/
2035 hierarchies and os-release information are combined via overlayfs
2036 with the file system hierarchy of the host OS.
2038 A new systemd-sysext tool can be used to merge, unmerge, list, and
2039 refresh system extension hierarchies. See
2040 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
2042 The systemd-sysext.service automatically merges installed system
2043 extensions during boot (before basic.target, but not in very early
2044 boot, since various file systems have to be mounted first).
2046 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
2047 supported system extension level.
2049 * A new ExtensionImages= unit setting can be used to apply the same
2050 system extension image concept from systemd-sysext to the namespaced
2051 file hierarchy of specific services, following the same rules and
2054 * Support for a new special "root=tmpfs" kernel command-line option has
2055 been added. When specified, a tmpfs is mounted on /, and mount.usr=
2056 should be used to point to the operating system implementation.
2058 * A new configuration file /etc/veritytab may be used to configure
2059 dm-verity integrity protection for block devices. Each line is in the
2060 format "volume-name data-device hash-device roothash options",
2061 similar to /etc/crypttab.
2063 * A new kernel command-line option systemd.verity.root_options= may be
2064 used to configure dm-verity behaviour for the root device.
2066 * The key file specified in /etc/crypttab (the third field) may now
2067 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
2068 acquired by connecting to that socket and reading from it. This
2069 allows the implementation of a service to provide key information
2070 dynamically, at the moment when it is needed.
2072 * When the hostname is set explicitly to "localhost", systemd-hostnamed
2073 will respect this. Previously such a setting would be mostly silently
2074 ignored. The goal is to honour configuration as specified by the
2077 * The fallback hostname that will be used by the system manager and
2078 systemd-hostnamed can now be configured in two new ways: by setting
2079 DEFAULT_HOSTNAME= in os-release(5), or by setting
2080 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
2081 also be configured during compilation. The environment variable is
2082 intended for testing and local overrides, the os-release(5) field is
2083 intended to allow customization by different variants of a
2084 distribution that share the same compiled packages.
2086 * The environment block of the manager itself may be configured through
2087 a new ManagerEnvironment= setting in system.conf or user.conf. This
2088 complements existing ways to set the environment block (the kernel
2089 command line for the system manager, the inherited environment and
2090 user@.service unit file settings for the user manager).
2092 * systemd-hostnamed now exports the default hostname and the source of
2093 the configured hostname ("static", "transient", or "default") as
2096 * systemd-hostnamed now exports the "HardwareVendor" and
2097 "HardwareModel" D-Bus properties, which are supposed to contain a
2098 pair of cleaned up, human readable strings describing the system's
2099 vendor and model. It's typically sourced from the firmware's DMI
2100 tables, but may be augmented from a new hwdb database. hostnamectl
2101 shows this in the status output.
2103 * Support has been added to systemd-cryptsetup for extracting the
2104 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
2105 metadata header. This allows the information how to open the
2106 encrypted device to be embedded directly in the device and obviates
2107 the need for configuration in an external file.
2109 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
2110 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
2111 pre-existing support for PKCS#11 security tokens).
2113 * systemd-repart may enroll encrypted partitions using TPM2
2114 hardware. This may be useful for example to create an encrypted /var
2115 partition bound to the machine on first boot.
2117 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
2118 and PKCS#11 security tokens to LUKS volumes, list and destroy
2121 http://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
2123 It also supports enrolling "recovery keys" and regular passphrases.
2125 * The libfido2 dependency is now based on dlopen(), so that the library
2126 is used at runtime when installed, but is not a hard runtime
2129 * systemd-cryptsetup gained support for two new options in
2130 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
2131 request synchronous processing of encryption/decryption IO.
2133 * The manager may be configured at compile time to use the fexecve()
2134 instead of the execve() system call when spawning processes. Using
2135 fexecve() closes a window between checking the security context of an
2136 executable and spawning it, but unfortunately the kernel displays
2137 stale information in the process' "comm" field, which impacts ps
2140 * The configuration option -Dcompat-gateway-hostname has been dropped.
2141 "_gateway" is now the only supported name.
2143 * The ConditionSecurity=tpm2 unit file setting may be used to check if
2144 the system has at least one TPM2 (tpmrm class) device.
2146 * A new ConditionCPUFeature= has been added that may be used to
2147 conditionalize units based on CPU features. For example,
2148 ConditionCPUFeature=rdrand will condition a unit so that it is only
2149 run when the system CPU supports the RDRAND opcode.
2151 * The existing ConditionControlGroupController= setting has been
2152 extended with two new values "v1" and "v2". "v2" means that the
2153 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
2154 hierarchy or the hybrid hierarchy are used.
2156 * A new PrivateIPC= setting on a unit file allows executed processes to
2157 be moved into a private IPC namespace, with separate System V IPC
2158 identifiers and POSIX message queues.
2160 A new IPCNamespacePath= allows the unit to be joined to an existing
2163 * The tables of system calls in seccomp filters are now automatically
2164 generated from kernel lists exported on
2165 https://fedora.juszkiewicz.com.pl/syscalls.html.
2167 The following architectures should now have complete lists:
2168 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
2169 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
2171 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
2172 on /run/ if it is not already a mount point. A writable /run/ has
2173 always been a requirement for a functioning system, but this was not
2174 guaranteed when using a read-only image.
2176 Users can always specify BindPaths= or InaccessiblePaths= as
2177 overrides, and they will take precedence. If the host's root mount
2178 point is used, there is no change in behaviour.
2180 * New bind mounts and file system image mounts may be injected into the
2181 mount namespace of a service (without restarting it). This is exposed
2182 respectively as 'systemctl bind <unit> <path>…' and
2183 'systemctl mount-image <unit> <image>…'.
2185 * The StandardOutput= and StandardError= settings can now specify files
2186 to be truncated for output (as "truncate:<path>").
2188 * The ExecPaths= and NoExecPaths= settings may be used to specify
2189 noexec for parts of the file system.
2191 * sd-bus has a new function sd_bus_open_user_machine() to open a
2192 connection to the session bus of a specific user in a local container
2193 or on the local host. This is exposed in the existing -M switch to
2194 systemctl and similar tools:
2196 systemctl --user -M lennart@foobar start foo
2198 This will connect to the user bus of a user "lennart" in container
2199 "foobar". If no container name is specified, the specified user on
2200 the host itself is connected to
2202 systemctl --user -M lennart@ start quux
2204 * sd-bus also gained a convenience function sd_bus_message_send() to
2205 simplify invocations of sd_bus_send(), taking only a single
2206 parameter: the message to send.
2208 * sd-event allows rate limits to be set on event sources, for dealing
2209 with high-priority event sources that might starve out others. See
2210 the new man page sd_event_source_set_ratelimit(3) for details.
2212 * systemd.link files gained a [Link] Promiscuous= switch, which allows
2213 the device to be raised in promiscuous mode.
2215 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
2216 number of TX and RX queues to be configured.
2218 New [Link] TransmitQueueLength= setting allows the size of the TX
2219 queue to be configured.
2221 New [Link] GenericSegmentOffloadMaxBytes= and
2222 GenericSegmentOffloadMaxSegments= allow capping the packet size and
2223 the number of segments accepted in Generic Segment Offload.
2225 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
2226 wireless routing protocol that operates on ISO/OSI Layer 2 only and
2227 uses ethernet frames to route/bridge packets. This encompasses a new
2228 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
2229 new settings in .netdev files, and a new BatmanAdvanced= setting in
2232 * systemd.network files gained a [Network] RouteTable= configuration
2233 switch to select the routing policy table.
2235 systemd.network files gained a [RoutingPolicyRule] Type=
2236 configuration switch (one of "blackhole, "unreachable", "prohibit").
2238 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
2239 RouteAllowList= settings to ignore/accept route advertisements from
2240 routers matching specified prefixes. The DenyList= setting has been
2241 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
2244 systemd.network files gained a [DHCPv6] UseAddress= setting to
2245 optionally ignore the address provided in the lease.
2247 systemd.network files gained a [DHCPv6PrefixDelegation]
2248 ManageTemporaryAddress= switch.
2250 systemd.network files gained a new ActivationPolicy= setting which
2251 allows configuring how the UP state of an interface shall be managed,
2252 i.e. whether the interface is always upped, always downed, or may be
2253 upped/downed by the user using "ip link set dev".
2255 * The default for the Broadcast= setting in .network files has slightly
2256 changed: the broadcast address will not be configured for wireguard
2259 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
2260 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
2261 configuration options for VLAN packet handling.
2263 * udev rules may now set log_level= option. This allows debug logs to
2264 be enabled for select events, e.g. just for a specific subsystem or
2265 even a single device.
2267 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
2268 DATA_PREPARED_ID properties for block devices with ISO9660 file
2271 * udev now exports decoded DMI information about installed memory slots
2272 as device properties under the /sys/class/dmi/id/ pseudo device.
2274 * /dev/ is not mounted noexec anymore. This didn't provide any
2275 significant security benefits and would conflict with the executable
2276 mappings used with /dev/sgx device nodes. The previous behaviour can
2277 be restored for individual services with NoExecPaths=/dev (or by allow-
2278 listing and excluding /dev from ExecPaths=).
2280 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
2281 and /dev/vhost-net are owned by the kvm group.
2283 * The hardware database has been extended with a list of fingerprint
2284 readers that correctly support USB auto-suspend using data from
2287 * systemd-resolved can now answer DNSSEC questions through the stub
2288 resolver interface in a way that allows local clients to do DNSSEC
2289 validation themselves. For a question with DO+CD set, it'll proxy the
2290 DNS query and respond with a mostly unmodified packet received from
2291 the upstream server.
2293 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
2294 resolved.conf. If true the service will provide caching even for DNS
2295 lookups made to an upstream DNS server on the 127.0.0.1/::1
2296 addresses. By default (and when the option is false) systemd-resolved
2297 will not cache such lookups, in order to avoid duplicate local
2298 caching, under the assumption the local upstream server caches
2301 * systemd-resolved now implements RFC5001 NSID in its local DNS
2302 stub. This may be used by local clients to determine whether they are
2303 talking to the DNS resolver stub or a different DNS server.
2305 * When resolving host names and other records resolvectl will now
2306 report where the data was acquired from (i.e. the local cache, the
2307 network, locally synthesized, …) and whether the network traffic it
2308 effected was encrypted or not. Moreover the tool acquired a number of
2309 new options --cache=, --synthesize=, --network=, --zone=,
2310 --trust-anchor=, --validate= that take booleans and may be used to
2311 tweak a lookup, i.e. whether it may be answered from cached
2312 information, locally synthesized information, information acquired
2313 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
2314 anchor, and whether DNSSEC validation shall be executed for the
2317 * systemd-nspawn gained a new --ambient-capability= setting
2318 (AmbientCapability= in .nspawn files) to configure ambient
2319 capabilities passed to the container payload.
2321 * systemd-nspawn gained the ability to configure the firewall using the
2322 nftables subsystem (in addition to the existing iptables
2323 support). Similarly, systemd-networkd's IPMasquerade= option now
2324 supports nftables as back-end, too. In both cases NAT on IPv6 is now
2325 supported too, in addition to IPv4 (the iptables back-end still is
2328 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
2329 retains its meaning, but has been deprecated. Please switch to either
2330 "ivp4" or "both" (if covering IPv6 is desired).
2332 * systemd-importd will now download .verity and .roothash.p7s files
2333 along with the machine image (as exposed via machinectl pull-raw).
2335 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
2336 setting to configure the time a unit's cgroup needs to exceed memory
2337 pressure limits before action will be taken, and a new
2338 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
2341 systemd-oomd is now considered fully supported (the usual
2342 backwards-compatiblity promises apply). Swap is not required for
2343 operation, but it is still recommended.
2345 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
2346 configures the retry delay when trying to contact servers.
2348 * systemd-stdio-bridge gained --system/--user options to connect to the
2349 system bus (previous default) or the user session bus.
2351 * systemd-localed may now call locale-gen to generate missing locales
2352 on-demand (UTF-8-only). This improves integration with Debian-based
2353 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
2355 * systemctl --check-inhibitors=true may now be used to obey inhibitors
2356 even when invoked non-interactively. The old --ignore-inhibitors
2357 switch is now deprecated and replaced by --check-inhibitors=false.
2359 * systemctl import-environment will now emit a warning when called
2360 without any arguments (i.e. to import the full environment block of
2361 the called program). This command will usually be invoked from a
2362 shell, which means that it'll inherit a bunch of variables which are
2363 specific to that shell, and usually to the TTY the shell is connected
2364 to, and don't have any meaning in the global context of the system or
2365 user service manager. Instead, only specific variables should be
2366 imported into the manager environment block.
2368 Similarly, programs which update the manager environment block by
2369 directly calling the D-Bus API of the manager, should also push
2370 specific variables, and not the full inherited environment.
2372 * systemctl's status output now shows unit state with a more careful
2373 choice of Unicode characters: units in maintenance show a "○" symbol
2374 instead of the usual "●", failed units show "×", and services being
2377 * coredumpctl gained a --debugger-arguments= switch to pass arguments
2378 to the debugger. It also gained support for showing coredump info in
2379 a simple JSON format.
2381 * systemctl/loginctl/machinectl's --signal= option now accept a special
2382 value "list", which may be used to show a brief table with known
2383 process signals and their numbers.
2385 * networkctl now shows the link activation policy in status.
2387 * Various tools gained --pager/--no-pager/--json= switches to
2388 enable/disable the pager and provide JSON output.
2390 * Various tools now accept two new values for the SYSTEMD_COLORS
2391 environment variable: "16" and "256", to configure how many terminal
2392 colors are used in output.
2394 * less 568 or newer is now required for the auto-paging logic of the
2395 various tools. Hyperlink ANSI sequences in terminal output are now
2396 used even if a pager is used, and older versions of less are not able
2397 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
2398 disable this output again.
2400 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
2401 builds, non-merged-usr builds) are now officially deprecated. A
2402 warning is emitted during build. Support is slated to be removed in
2403 about a year (when the Debian Bookworm release development starts).
2405 * Systems with the legacy cgroup v1 hierarchy are now marked as
2406 "tainted", to make it clearer that using the legacy hierarchy is not
2409 * systemd-localed will now refuse to configure a keymap which is not
2410 installed in the file system. This is intended as a bug fix, but
2411 could break cases where systemd-localed was used to configure the
2412 keymap in advanced of it being installed. It is necessary to install
2413 the keymap file first.
2415 * The main git development branch has been renamed to 'main'.
2417 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
2418 for partitions, as in the vast majority of cases they contain none
2419 and are used internally by the bootloader (eg: uboot).
2421 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
2422 spawned processes to the PID of the process itself. This may be used
2423 by programs for detecting whether they were forked off by the service
2424 manager itself or are a process forked off further down the tree.
2426 * The sd-device API gained four new calls: sd_device_get_action() to
2427 determine the uevent add/remove/change/… action the device object has
2428 been seen for, sd_device_get_seqno() to determine the uevent sequence
2429 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
2430 object from stat(2) data of a device node, and sd_device_trigger() to
2431 write to the 'uevent' attribute of a device.
2433 * For most tools the --no-legend= switch has been replaced by
2434 --legend=no and --legend=yes, to force whether tables are shown with
2437 * Units acquired a new property "Markers" that takes a list of zero,
2438 one or two of the following strings: "needs-reload" and
2439 "needs-restart". These markers may be set via "systemctl
2440 set-property". Once a marker is set, "systemctl reload-or-restart
2441 --marked" may be invoked to execute the operation the units are
2442 marked for. This is useful for package managers that want to mark
2443 units for restart/reload while updating, but effect the actual
2444 operations at a later step at once.
2446 * The sd_bus_message_read_strv() API call of sd-bus may now also be
2447 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
2450 * bootctl will now report whether the UEFI firmware used a TPM2 device
2451 and measured the boot process into it.
2453 * systemd-tmpfiles learnt support for a new environment variable
2454 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
2455 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
2456 even if the root fs of the system is not itself a btrfs volume.
2458 * systemd-detect-virt/ConditionVirtualization= will now explicitly
2459 detect Docker/Podman environments where possible. Moreover, they
2460 should be able to generically detect any container manager as long as
2461 it assigns the container a cgroup.
2463 * portablectl gained a new "reattach" verb for detaching/reattaching a
2464 portable service image, useful for updating images on-the-fly.
2466 * Intel SGX enclave device nodes (which expose a security feature of
2467 newer Intel CPUs) will now be owned by a new system group "sgx".
2469 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
2470 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
2471 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
2472 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
2473 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
2474 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
2475 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
2476 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
2477 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
2478 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
2479 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
2480 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
2481 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
2482 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
2483 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
2484 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
2485 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
2486 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
2487 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
2488 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
2489 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
2490 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
2491 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
2492 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
2493 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
2494 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
2495 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
2496 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
2497 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
2498 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
2499 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
2500 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
2501 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
2502 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
2503 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
2504 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
2505 Zmicer Turok, Дамјан Георгиевски
2507 — Berlin, 2021-03-30
2511 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
2512 "bind" and "unbind" to the Linux device model. When this kernel
2513 change was made, systemd-udevd was only minimally updated to handle
2514 and propagate these new event types. The introduction of these new
2515 uevents (which are typically generated for USB devices and devices
2516 needing a firmware upload before being functional) resulted in a
2517 number of issues which we so far didn't address. We hoped the kernel
2518 maintainers would themselves address these issues in some form, but
2519 that did not happen. To handle them properly, many (if not most) udev
2520 rules files shipped in various packages need updating, and so do many
2521 programs that monitor or enumerate devices with libudev or sd-device,
2522 or otherwise process uevents. Please note that this incompatibility
2523 is not fault of systemd or udev, but caused by an incompatible kernel
2524 change that happened back in Linux 4.14, but is becoming more and
2525 more visible as the new uevents are generated by more kernel drivers.
2527 To minimize issues resulting from this kernel change (but not avoid
2528 them entirely) starting with systemd-udevd 247 the udev "tags"
2529 concept (which is a concept for marking and filtering devices during
2530 enumeration and monitoring) has been reworked: udev tags are now
2531 "sticky", meaning that once a tag is assigned to a device it will not
2532 be removed from the device again until the device itself is removed
2533 (i.e. unplugged). This makes sure that any application monitoring
2534 devices that match a specific tag is guaranteed to both see uevents
2535 where the device starts being relevant, and those where it stops
2536 being relevant (the latter now regularly happening due to the new
2537 "unbind" uevent type). The udev tags concept is hence now a concept
2538 tied to a *device* instead of a device *event* — unlike for example
2539 udev properties whose lifecycle (as before) is generally tied to a
2540 device event, meaning that the previously determined properties are
2541 forgotten whenever a new uevent is processed.
2543 With the newly redefined udev tags concept, sometimes it's necessary
2544 to determine which tags are the ones applied by the most recent
2545 uevent/database update, in order to discern them from those
2546 originating from earlier uevents/database updates of the same
2547 device. To accommodate for this a new automatic property CURRENT_TAGS
2548 has been added that works similar to the existing TAGS property but
2549 only lists tags set by the most recent uevent/database
2550 update. Similarly, the libudev/sd-device API has been updated with
2551 new functions to enumerate these 'current' tags, in addition to the
2552 existing APIs that now enumerate the 'sticky' ones.
2554 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
2555 essential that all udev rules files and applications are updated to
2556 handle the new events. Specifically:
2558 • All rule files that currently use a header guard similar to
2559 ACTION!="add|change",GOTO="xyz_end" should be updated to use
2560 ACTION=="remove",GOTO="xyz_end" instead, so that the
2561 properties/tags they add are also applied whenever "bind" (or
2562 "unbind") is seen. (This is most important for all physical device
2563 types — those for which "bind" and "unbind" are currently
2564 generated, for all other device types this change is still
2565 recommended but not as important — but certainly prepares for
2566 future kernel uevent type additions).
2568 • Similarly, all code monitoring devices that contains an 'if' branch
2569 discerning the "add" + "change" uevent actions from all other
2570 uevents actions (i.e. considering devices only relevant after "add"
2571 or "change", and irrelevant on all other events) should be reworked
2572 to instead negatively check for "remove" only (i.e. considering
2573 devices relevant after all event types, except for "remove", which
2574 invalidates the device). Note that this also means that devices
2575 should be considered relevant on "unbind", even though conceptually
2576 this — in some form — invalidates the device. Since the precise
2577 effect of "unbind" is not generically defined, devices should be
2578 considered relevant even after "unbind", however I/O errors
2579 accessing the device should then be handled gracefully.
2581 • Any code that uses device tags for deciding whether a device is
2582 relevant or not most likely needs to be updated to use the new
2583 udev_device_has_current_tag() API (or sd_device_has_current_tag()
2584 in case sd-device is used), to check whether the tag is set at the
2585 moment an uevent is seen (as opposed to the existing
2586 udev_device_has_tag() API which checks if the tag ever existed on
2587 the device, following the API concept redefinition explained
2590 We are very sorry for this breakage and the requirement to update
2591 packages using these interfaces. We'd again like to underline that
2592 this is not caused by systemd/udev changes, but result of a kernel
2595 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
2596 packages have not retriggered devices once the udev package (or any
2597 auxiliary package installing additional udev rules) is updated. We
2598 intend to work with major distributions to change this, so that
2599 "udevadm trigger -a change" is issued on such upgrades, ensuring that
2600 the updated ruleset is applied to the devices already discovered, so
2601 that (asynchronously) after the upgrade completed the udev database
2602 is consistent with the updated rule set. This means udev rules must
2603 be ready to be retriggered with a "change" action any time, and
2604 result in correct and complete udev database entries. While the
2605 majority of udev rule files known to us currently get this right,
2606 some don't. Specifically, there are udev rules files included in
2607 various packages that only set udev properties on the "add" action,
2608 but do not handle the "change" action. If a device matching those
2609 rules is retriggered with the "change" action (as is intended here)
2610 it would suddenly lose the relevant properties. This always has been
2611 problematic, but as soon as all udev devices are triggered on relevant
2612 package upgrades this will become particularly so. It is strongly
2613 recommended to fix offending rules so that they can handle a "change"
2614 action at any time, and acquire all necessary udev properties even
2615 then. Or in other words: the header guard mentioned above
2616 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
2617 this, as it makes sure rules are rerun on "change" correctly, and
2618 accumulate the correct and complete set of udev properties. udev rule
2619 definitions that cannot handle "change" events being triggered at
2620 arbitrary times should be considered buggy.
2622 * The MountAPIVFS= service file setting now defaults to on if
2623 RootImage= and RootDirectory= are used, which means that with those
2624 two settings /proc/, /sys/ and /dev/ are automatically properly set
2625 up for services. Previous behaviour may be restored by explicitly
2626 setting MountAPIVFS=off.
2628 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
2629 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
2630 latter it takes precedence over the former, similar to how most of
2631 systemd's own configuration is handled. Given that PAM stack
2632 definitions are primarily put together by OS vendors/distributions
2633 (though possibly overridden by users), this systemd release moves its
2634 own PAM stack configuration for the "systemd-user" PAM service (i.e.
2635 for the PAM session invoked by the per-user user@.service instance)
2636 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
2637 packages' vendor versions of their PAM stack definitions from
2638 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
2639 desired the location to which systemd installs its PAM stack
2640 configuration may be changed via the -Dpamconfdir Meson option.
2642 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
2643 libpwquality and libcryptsetup have been changed to be based on
2644 dlopen(): instead of regular dynamic library dependencies declared in
2645 the binary ELF headers, these libraries are now loaded on demand
2646 only, if they are available. If the libraries cannot be found the
2647 relevant operations will fail gracefully, or a suitable fallback
2648 logic is chosen. This is supposed to be useful for general purpose
2649 distributions, as it allows minimizing the list of dependencies the
2650 systemd packages pull in, permitting building of more minimal OS
2651 images, while still making use of these "weak" dependencies should
2652 they be installed. Since many package managers automatically
2653 synthesize package dependencies from ELF shared library dependencies,
2654 some additional manual packaging work has to be done now to replace
2655 those (slightly downgraded from "required" to "recommended" or
2656 whatever is conceptually suitable for the package manager). Note that
2657 this change does not alter build-time behaviour: as before the
2658 build-time dependencies have to be installed during build, even if
2659 they now are optional during runtime.
2661 * sd-event.h gained a new call sd_event_add_time_relative() for
2662 installing timers relative to the current time. This is mostly a
2663 convenience wrapper around the pre-existing sd_event_add_time() call
2664 which installs absolute timers.
2666 * sd-event event sources may now be placed in a new "exit-on-failure"
2667 mode, which may be controlled via the new
2668 sd_event_source_get_exit_on_failure() and
2669 sd_event_source_set_exit_on_failure() functions. If enabled, any
2670 failure returned by the event source handler functions will result in
2671 exiting the event loop (unlike the default behaviour of just
2672 disabling the event source but continuing with the event loop). This
2673 feature is useful to set for all event sources that define "primary"
2674 program behaviour (where failure should be fatal) in contrast to
2675 "auxiliary" behaviour (where failure should remain local).
2677 * Most event source types sd-event supports now accept a NULL handler
2678 function, in which case the event loop is exited once the event
2679 source is to be dispatched, using the userdata pointer — converted to
2680 a signed integer — as exit code of the event loop. Previously this
2681 was supported for IO and signal event sources already. Exit event
2682 sources still do not support this (simply because it makes little
2683 sense there, as the event loop is already exiting when they are
2686 * A new per-unit setting RootImageOptions= has been added which allows
2687 tweaking the mount options for any file system mounted as effect of
2688 the RootImage= setting.
2690 * Another new per-unit setting MountImages= has been added, that allows
2691 mounting additional disk images into the file system tree accessible
2694 * Timer units gained a new FixedRandomDelay= boolean setting. If
2695 enabled, the random delay configured with RandomizedDelaySec= is
2696 selected in a way that is stable on a given system (though still
2697 different for different units).
2699 * Socket units gained a new setting Timestamping= that takes "us", "ns"
2700 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
2703 * systemd-repart now generates JSON output when requested with the new
2706 * systemd-machined's OpenMachineShell() bus call will now pass
2707 additional policy metadata data fields to the PolicyKit
2708 authentication request.
2710 * systemd-tmpfiles gained a new -E switch, which is equivalent to
2711 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
2712 --exclude=/sys. It's particularly useful in combination with --root=,
2713 when operating on OS trees that do not have any of these four runtime
2714 directories mounted, as this means no files below these subtrees are
2715 created or modified, since those mount points should probably remain
2718 * systemd-tmpfiles gained a new --image= switch which is like --root=,
2719 but takes a disk image instead of a directory as argument. The
2720 specified disk image is mounted inside a temporary mount namespace
2721 and the tmpfiles.d/ drop-ins stored in the image are executed and
2722 applied to the image. systemd-sysusers similarly gained a new
2723 --image= switch, that allows the sysusers.d/ drop-ins stored in the
2724 image to be applied onto the image.
2726 * Similarly, the journalctl command also gained an --image= switch,
2727 which is a quick one-step solution to look at the log data included
2730 * journalctl's --output=cat option (which outputs the log content
2731 without any metadata, just the pure text messages) will now make use
2732 of terminal colors when run on a suitable terminal, similarly to the
2735 * JSON group records now support a "description" string that may be
2736 used to add a human-readable textual description to such groups. This
2737 is supposed to match the user's GECOS field which traditionally
2738 didn't have a counterpart for group records.
2740 * The "systemd-dissect" tool that may be used to inspect OS disk images
2741 and that was previously installed to /usr/lib/systemd/ has now been
2742 moved to /usr/bin/, reflecting its updated status of an officially
2743 supported tool with a stable interface. It gained support for a new
2744 --mkdir switch which when combined with --mount has the effect of
2745 creating the directory to mount the image to if it is missing
2746 first. It also gained two new commands --copy-from and --copy-to for
2747 copying files and directories in and out of an OS image without the
2748 need to manually mount it. It also acquired support for a new option
2749 --json= to generate JSON output when inspecting an OS image.
2751 * The cgroup2 file system is now mounted with the
2752 "memory_recursiveprot" mount option, supported since kernel 5.7. This
2753 means that the MemoryLow= and MemoryMin= unit file settings now apply
2754 recursively to whole subtrees.
2756 * systemd-homed now defaults to using the btrfs file system — if
2757 available — when creating home directories in LUKS volumes. This may
2758 be changed with the DefaultFileSystemType= setting in homed.conf.
2759 It's now the default file system in various major distributions and
2760 has the major benefit for homed that it can be grown and shrunk while
2761 mounted, unlike the other contenders ext4 and xfs, which can both be
2762 grown online, but not shrunk (in fact xfs is the technically most
2763 limited option here, as it cannot be shrunk at all).
2765 * JSON user records managed by systemd-homed gained support for
2766 "recovery keys". These are basically secondary passphrases that can
2767 unlock user accounts/home directories. They are computer-generated
2768 rather than user-chosen, and typically have greater entropy.
2769 homectl's --recovery-key= option may be used to add a recovery key to
2770 a user account. The generated recovery key is displayed as a QR code,
2771 so that it can be scanned to be kept in a safe place. This feature is
2772 particularly useful in combination with systemd-homed's support for
2773 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
2774 security tokens are lost. Recovery keys may be entered wherever the
2775 system asks for a password.
2777 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
2778 home directory which indicates that a home directory has not been
2779 deactivated cleanly when offline. This flag is useful to identify
2780 home directories for which the offline discard logic did not run when
2781 offlining, and where it would be a good idea to log in again to catch
2784 * systemctl gained a new parameter --timestamp= which may be used to
2785 change the style in which timestamps are output, i.e. whether to show
2786 them in local timezone or UTC, or whether to show µs granularity.
2788 * Alibaba's "pouch" container manager is now detected by
2789 systemd-detect-virt, ConditionVirtualization= and similar
2790 constructs. Similar, they now also recognize IBM PowerVM machine
2793 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
2794 place to use for propagating external mounts into the
2795 container. Similarly /run/host/notify is now used as the socket path
2796 for container payloads to communicate with the container manager
2797 using sd_notify(). The container manager now uses the
2798 /run/host/inaccessible/ directory to place "inaccessible" file nodes
2799 of all relevant types which may be used by the container payload as
2800 bind mount source to over-mount inodes to make them inaccessible.
2801 /run/host/container-manager will now be initialized with the same
2802 string as the $container environment variable passed to the
2803 container's PID 1. /run/host/container-uuid will be initialized with
2804 the same string as $container_uuid. This means the /run/host/
2805 hierarchy is now the primary way to make host resources available to
2806 the container. The Container Interface documents these new files and
2809 https://systemd.io/CONTAINER_INTERFACE
2811 * Support for the "ConditionNull=" unit file condition has been
2812 deprecated and undocumented for 6 years. systemd started to warn
2813 about its use 1.5 years ago. It has now been removed entirely.
2815 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
2816 a sd_bus_error struct and a list of error names, and checks if the
2817 error matches one of these names. It's a convenience wrapper that is
2818 useful in cases where multiple errors shall be handled the same way.
2820 * A new system call filter list "@known" has been added, that contains
2821 all system calls known at the time systemd was built.
2823 * Behaviour of system call filter allow lists has changed slightly:
2824 system calls that are contained in @known will result in EPERM by
2825 default, while those not contained in it result in ENOSYS. This
2826 should improve compatibility because known system calls will thus be
2827 communicated as prohibited, while unknown (and thus newer ones) will
2828 be communicated as not implemented, which hopefully has the greatest
2829 chance of triggering the right fallback code paths in client
2832 * "systemd-analyze syscall-filter" will now show two separate sections
2833 at the bottom of the output: system calls known during systemd build
2834 time but not included in any of the filter groups shown above, and
2835 system calls defined on the local kernel but known during systemd
2838 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
2839 systemd-nspawn all system call filter violations will be logged by
2840 the kernel (audit). This is useful for tracking down system calls
2841 invoked by container payloads that are prohibited by the container's
2842 system call filter policy.
2844 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
2845 systemd-nspawn (and other programs that use seccomp) all seccomp
2846 filtering is turned off.
2848 * Two new unit file settings ProtectProc= and ProcSubset= have been
2849 added that expose the hidepid= and subset= mount options of procfs.
2850 All processes of the unit will only see processes in /proc that are
2851 are owned by the unit's user. This is an important new sandboxing
2852 option that is recommended to be set on all system services. All
2853 long-running system services that are included in systemd itself set
2854 this option now. This option is only supported on kernel 5.8 and
2855 above, since the hidepid= option supported on older kernels was not a
2856 per-mount option but actually applied to the whole PID namespace.
2858 * Socket units gained a new boolean setting FlushPending=. If enabled
2859 all pending socket data/connections are flushed whenever the socket
2860 unit enters the "listening" state, i.e. after the associated service
2863 * The unit file setting NUMAMask= gained a new "all" value: when used,
2864 all existing NUMA nodes are added to the NUMA mask.
2866 * A new "credentials" logic has been added to system services. This is
2867 a simple mechanism to pass privileged data to services in a safe and
2868 secure way. It's supposed to be used to pass per-service secret data
2869 such as passwords or cryptographic keys but also associated less
2870 private information such as user names, certificates, and similar to
2871 system services. Each credential is identified by a short user-chosen
2872 name and may contain arbitrary binary data. Two new unit file
2873 settings have been added: SetCredential= and LoadCredential=. The
2874 former allows setting a credential to a literal string, the latter
2875 sets a credential to the contents of a file (or data read from a
2876 user-chosen AF_UNIX stream socket). Credentials are passed to the
2877 service via a special credentials directory, one file for each
2878 credential. The path to the credentials directory is passed in a new
2879 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
2880 are passed in the file system they may be easily referenced in
2881 ExecStart= command lines too, thus no explicit support for the
2882 credentials logic in daemons is required (though ideally daemons
2883 would look for the bits they need in $CREDENTIALS_DIRECTORY
2884 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
2885 backed by unswappable memory if privileges allow it, immutable if
2886 privileges allow it, is accessible only to the service's UID, and is
2887 automatically destroyed when the service stops.
2889 * systemd-nspawn supports the same credentials logic. It can both
2890 consume credentials passed to it via the aforementioned
2891 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
2892 to its payload. The service manager/PID 1 has been updated to match
2893 this: it can also accept credentials from the container manager that
2894 invokes it (in fact: any process that invokes it), and passes them on
2895 to its services. Thus, credentials can be propagated recursively down
2896 the tree: from a system's service manager to a systemd-nspawn
2897 service, to the service manager that runs as container payload and to
2898 the service it runs below. Credentials may also be added on the
2899 systemd-nspawn command line, using new --set-credential= and
2900 --load-credential= command line switches that match the
2901 aforementioned service settings.
2903 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
2904 the partition drop-ins which may be used to format/LUKS
2905 encrypt/populate any created partitions. The partitions are
2906 encrypted/formatted/populated before they are registered in the
2907 partition table, so that they appear atomically: either the
2908 partitions do not exist yet or they exist fully encrypted, formatted,
2909 and populated — there is no time window where they are
2910 "half-initialized". Thus the system is robust to abrupt shutdown: if
2911 the tool is terminated half-way during its operations on next boot it
2912 will start from the beginning.
2914 * systemd-repart's --size= operation gained a new "auto" value. If
2915 specified, and operating on a loopback file it is automatically sized
2916 to the minimal size the size constraints permit. This is useful to
2917 use "systemd-repart" as an image builder for minimally sized images.
2919 * systemd-resolved now gained a third IPC interface for requesting name
2920 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
2921 interface is now supported. The nss-resolve NSS module has been
2922 modified to use this new interface instead of D-Bus. Using Varlink
2923 has a major benefit over D-Bus: it works without a broker service,
2924 and thus already during earliest boot, before the dbus daemon has
2925 been started. This means name resolution via systemd-resolved now
2926 works at the same time systemd-networkd operates: from earliest boot
2927 on, including in the initrd.
2929 * systemd-resolved gained support for a new DNSStubListenerExtra=
2930 configuration file setting which may be used to specify additional IP
2931 addresses the built-in DNS stub shall listen on, in addition to the
2932 main one on 127.0.0.53:53.
2934 * Name lookups issued via systemd-resolved's D-Bus and Varlink
2935 interfaces (and thus also via glibc NSS if nss-resolve is used) will
2936 now honour a trailing dot in the hostname: if specified the search
2937 path logic is turned off. Thus "resolvectl query foo." is now
2938 equivalent to "resolvectl query --search=off foo.".
2940 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
2941 exposes how /etc/resolv.conf is currently managed: by resolved (and
2942 in which mode if so) or another subsystem. "resolvctl" will display
2943 this property in its status output.
2945 * The resolv.conf snippets systemd-resolved provides will now set "."
2946 as the search domain if no other search domain is known. This turns
2947 off the derivation of an implicit search domain by nss-dns for the
2948 hostname, when the hostname is set to an FQDN. This change is done to
2949 make nss-dns using resolv.conf provided by systemd-resolved behave
2950 more similarly to nss-resolve.
2952 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
2953 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
2954 "birth" time (btime) of a file in addition to the atime, mtime, and
2957 * systemd-analyze gained a new verb "capability" that lists all known
2958 capabilities by the systemd build and by the kernel.
2960 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
2961 advance the system clock to it at boot if it is noticed to be before
2962 that time. Previously, PID 1 would only advance the time to an epoch
2963 time that is set during build-time. With this new file OS builders
2964 can change this epoch timestamp on individual OS images without
2965 having to rebuild systemd.
2967 * systemd-logind will now listen to the KEY_RESTART key from the Linux
2968 input layer and reboot the system if it is pressed, similarly to how
2969 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
2970 was originally defined in the Multimedia context (to restart playback
2971 of a song or film), but is now primarily used in various embedded
2972 devices for "Reboot" buttons. Accordingly, systemd-logind will now
2973 honour it as such. This may configured in more detail via the new
2974 HandleRebootKey= and RebootKeyIgnoreInhibited=.
2976 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
2977 copying OS trees, for example in "systemd-nspawn --ephemeral",
2978 "systemd-nspawn --template=", "machinectl clone" and similar. This is
2979 useful when operating with OSTree images, which use hardlinks heavily
2980 throughout, and where such copies previously resulting in "exploding"
2983 * systemd-nspawn's --console= setting gained support for a new
2984 "autopipe" value, which is identical to "interactive" when invoked on
2985 a TTY, and "pipe" otherwise.
2987 * systemd-networkd's .network files gained support for explicitly
2988 configuring the multicast membership entries of bridge devices in the
2989 [BridgeMDB] section. It also gained support for the PIE queuing
2990 discipline in the [FlowQueuePIE] sections.
2992 * systemd-networkd's .netdev files may now be used to create "BareUDP"
2993 tunnels, configured in the new [BareUDP] setting.
2995 * systemd-networkd's Gateway= setting in .network files now accepts the
2996 special values "_dhcp4" and "_ipv6ra" to configure additional,
2997 locally defined, explicit routes to the gateway acquired via DHCP or
2998 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
2999 but still accepted for backwards compatibility.
3001 * systemd-networkd's [IPv6PrefixDelegation] section and
3002 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
3003 IPv6SendRA= (the old names are still accepted for backwards
3006 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
3007 boolean setting in [Network] section. If enabled, the delegated prefix
3008 gained by another link will be configured, and an address within the
3009 prefix will be assigned.
3011 * systemd-networkd's .network files gained the Announce= boolean setting
3012 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
3013 prefix will be announced through IPv6 router advertisement (IPv6 RA).
3014 The setting is enabled by default.
3016 * VXLAN tunnels may now be marked as independent of any underlying
3017 network interface via the new Independent= boolean setting.
3019 * systemctl gained support for two new verbs: "service-log-level" and
3020 "service-log-target" may be used on services that implement the
3021 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
3022 adjust the log level and target. All of systemd's long-running
3023 services support this now, but ideally all system services would
3024 implement this interface to make the system more uniformly
3027 * The SystemCallErrorNumber= unit file setting now accepts the new
3028 "kill" and "log" actions, in addition to arbitrary error number
3029 specifications as before. If "kill" the processes are killed on the
3030 event, if "log" the offending system call is audit logged.
3032 * A new SystemCallLog= unit file setting has been added that accepts a
3033 list of system calls that shall be logged about (audit).
3035 * The OS image dissection logic (as used by RootImage= in unit files or
3036 systemd-nspawn's --image= switch) has gained support for identifying
3037 and mounting explicit /usr/ partitions, which are now defined in the
3038 discoverable partition specification. This should be useful for
3039 environments where the root file system is
3040 generated/formatted/populated dynamically on first boot and combined
3041 with an immutable /usr/ tree that is supplied by the vendor.
3043 * In the final phase of shutdown, within the systemd-shutdown binary
3044 we'll now try to detach MD devices (i.e software RAID) in addition to
3045 loopback block devices and DM devices as before. This is supposed to
3046 be a safety net only, in order to increase robustness if things go
3047 wrong. Storage subsystems are expected to properly detach their
3048 storage volumes during regular shutdown already (or in case of
3049 storage backing the root file system: in the initrd hook we return to
3052 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
3053 will now log the thread ID in their log output. This is useful when
3054 working with heavily threaded programs.
3056 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
3057 not use the RDRAND CPU instruction. This is useful in environments
3058 such as replay debuggers where non-deterministic behaviour is not
3061 * The autopaging logic in systemd's various tools (such as systemctl)
3062 has been updated to turn on "secure" mode in "less"
3063 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
3064 detected. This disables invoking external programs from the pager,
3065 via the pipe logic. This behaviour may be overridden via the new
3066 $SYSTEMD_PAGERSECURE environment variable.
3068 * Units which have resource limits (.service, .mount, .swap, .slice,
3069 .socket, and .slice) gained new configuration settings
3070 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
3071 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
3072 limits and optional action taken by systemd-oomd.
3074 * A new service systemd-oomd has been added. It monitors resource
3075 contention for selected parts of the unit hierarchy using the PSI
3076 information reported by the kernel, and kills processes when memory
3077 or swap pressure is above configured limits. This service is only
3078 enabled by default in developer mode (see below) and should be
3079 considered a preview in this release. Behaviour details and option
3080 names are subject to change without the usual backwards-compatibility
3083 * A new helper oomctl has been added to introspect systemd-oomd state.
3084 It is only enabled by default in developer mode and should be
3085 considered a preview without the usual backwards-compatibility
3088 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
3089 enabled, systemd reads the system UID boundaries from /etc/login.defs
3090 at runtime, instead of using the built-in values selected during
3091 build. This is an option to improve compatibility for upgrades from
3092 old systems. It's strongly recommended not to make use of this
3093 functionality on new systems (or even enable it during build), as it
3094 makes something runtime-configurable that is mostly an implementation
3095 detail of the OS, and permits avoidable differences in deployments
3096 that create all kinds of problems in the long run.
3098 * New meson option '-Dmode=developer|release' has been added. When
3099 'developer', additional checks and features are enabled that are
3100 relevant during upstream development, e.g. verification that
3101 semi-automatically-generated documentation has been properly updated
3102 following API changes. Those checks are considered hints for
3103 developers and are not actionable in downstream builds. In addition,
3104 extra features that are not ready for general consumption may be
3105 enabled in developer mode. It is thus recommended to set
3106 '-Dmode=release' in end-user and distro builds.
3108 * systemd-cryptsetup gained support for processing detached LUKS
3109 headers specified on the kernel command line via the header=
3110 parameter of the luks.options= kernel command line option. The same
3111 device/path syntax as for key files is supported for header files
3114 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
3115 slot index data for devices that are connected through a PCI bridge
3116 where the _SUN index is associated with the bridge instead of the
3117 network device itself. Previously this would create ambiguous device
3118 naming if multiple network interfaces were connected to the same PCI
3119 bridge. Since this is a naming scheme incompatibility on systems that
3120 possess hardware like this it has been introduced as new naming
3121 scheme "v247". The previous scheme can be selected via the
3122 "net.naming-scheme=v245" kernel command line parameter.
3124 * ConditionFirstBoot= semantics have been modified to be safe towards
3125 abnormal system power-off during first boot. Specifically, the
3126 "systemd-machine-id-commit.service" service now acts as boot
3127 milestone indicating when the first boot process is sufficiently
3128 complete in order to not consider the next following boot also a
3129 first boot. If the system is reset before this unit is reached the
3130 first time, the next boot will still be considered a first boot; once
3131 it has been reached, no further boots will be considered a first
3132 boot. The "first-boot-complete.target" unit now acts as official hook
3133 point to order against this. If a service shall be run on every boot
3134 until the first boot fully succeeds it may thus be ordered before
3135 this target unit (and pull it in) and carry ConditionFirstBoot=
3138 * bootctl's set-default and set-oneshot commands now accept the three
3139 special strings "@default", "@oneshot", "@current" in place of a boot
3140 entry id. These strings are resolved to the current default and
3141 oneshot boot loader entry, as well as the currently booted one. Thus
3142 a command "bootctl set-default @current" may be used to make the
3143 currently boot menu item the new default for all subsequent boots.
3145 * "systemctl edit" has been updated to show the original effective unit
3146 contents in commented form in the text editor.
3148 * Units in user mode are now segregated into three new slices:
3149 session.slice (units that form the core of graphical session),
3150 app.slice ("normal" user applications), and background.slice
3151 (low-priority tasks). Unless otherwise configured, user units are
3152 placed in app.slice. The plan is to add resource limits and
3153 protections for the different slices in the future.
3155 * New GPT partition types for RISCV32/64 for the root and /usr
3156 partitions, and their associated Verity partitions have been defined,
3157 and are now understood by systemd-gpt-auto-generator, and the OS
3158 image dissection logic.
3160 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
3161 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
3162 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
3163 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
3164 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
3165 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
3166 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
3167 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
3168 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
3169 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
3170 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
3171 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
3172 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
3173 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
3174 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
3175 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
3176 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
3177 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
3178 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
3179 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
3180 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
3181 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
3182 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
3183 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
3184 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
3185 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
3186 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
3187 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
3188 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
3189 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
3190 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
3191 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
3192 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
3194 – Warsaw, 2020-11-26
3198 * The service manager gained basic support for cgroup v2 freezer. Units
3199 can now be suspended or resumed either using new systemctl verbs,
3200 freeze and thaw respectively, or via D-Bus.
3202 * PID 1 may now automatically load pre-compiled AppArmor policies from
3203 /etc/apparmor/earlypolicy during early boot.
3205 * The CPUAffinity= setting in service unit files now supports a new
3206 special value "numa" that causes the CPU affinity masked to be set
3207 based on the NUMA mask.
3209 * systemd will now log about all left-over processes remaining in a
3210 unit when the unit is stopped. It will now warn about services using
3211 KillMode=none, as this is generally an unsafe thing to make use of.
3213 * Two new unit file settings
3214 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
3215 added. They may be used to check whether a specific file system path
3216 resides on a block device that is encrypted on the block level
3217 (i.e. using dm-crypt/LUKS).
3219 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
3220 has been added that may be used for simple environment checks. This
3221 is particularly useful when passing in environment variables from a
3222 container manager (or from PAM in case of the systemd --user
3225 * .service unit files now accept a new setting CoredumpFilter= which
3226 allows configuration of the memory sections coredumps of the
3227 service's processes shall include.
3229 * .mount units gained a new ReadWriteOnly= boolean option. If set
3230 it will not be attempted to mount a file system read-only if mounting
3231 in read-write mode doesn't succeed. An option x-systemd.rw-only is
3232 available in /etc/fstab to control the same.
3234 * .socket units gained a new boolean setting PassPacketInfo=. If
3235 enabled, the kernel will attach additional per-packet metadata to all
3236 packets read from the socket, as an ancillary message. This controls
3237 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
3238 depending on socket type.
3240 * .service units gained a new setting RootHash= which may be used to
3241 specify the root hash for verity enabled disk images which are
3242 specified in RootImage=. RootVerity= may be used to specify a path to
3243 the Verity data matching a RootImage= file system. (The latter is
3244 only useful for images that do not contain the Verity data embedded
3245 into the same image that carries a GPT partition table following the
3246 Discoverable Partition Specification). Similarly, systemd-nspawn
3247 gained a new switch --verity-data= that takes a path to a file with
3248 the verity data of the disk image supplied in --image=, if the image
3249 doesn't contain the verity data itself.
3251 * .service units gained a new setting RootHashSignature= which takes
3252 either a base64 encoded PKCS#7 signature of the root hash specified
3253 with RootHash=, or a path to a file to read the signature from. This
3254 allows validation of the root hash against public keys available in
3255 the kernel keyring, and is only supported on recent kernels
3256 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
3257 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
3258 this mechanism has also been added to systemd-veritysetup.
3260 * .service unit files gained two new options
3261 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
3262 tune behaviour if a start or stop timeout is hit, i.e. whether to
3263 terminate the service with SIGTERM, SIGABRT or SIGKILL.
3265 * Most options in systemd that accept hexadecimal values prefixed with
3266 0x in additional to the usual decimal notation now also support octal
3267 notation when the 0o prefix is used and binary notation if the 0b
3270 * Various command line parameters and configuration file settings that
3271 configure key or certificate files now optionally take paths to
3272 AF_UNIX sockets in the file system. If configured that way a stream
3273 connection is made to the socket and the required data read from
3274 it. This is a simple and natural extension to the existing regular
3275 file logic, and permits other software to provide keys or
3276 certificates via simple IPC services, for example when unencrypted
3277 storage on disk is not desired. Specifically, systemd-networkd's
3278 Wireguard and MACSEC key file settings as well as
3279 systemd-journal-gatewayd's and systemd-journal-remote's PEM
3280 key/certificate parameters support this now.
3282 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
3283 configuration files that support specifier expansion learnt six new
3284 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
3285 resolve to the various ID fields from /etc/os-release, %l resolves to
3286 the "short" hostname of the system, i.e. the hostname configured in
3287 the kernel truncated at the first dot.
3289 * Support for the .include syntax in unit files has been removed. The
3290 concept has been obsolete for 6 years and we started warning about
3291 its pending removal 2 years ago (also see NEWS file below). It's
3294 * StandardError= and StandardOutput= in unit files no longer support
3295 the "syslog" and "syslog-console" switches. They were long removed
3296 from the documentation, but will now result in warnings when used,
3297 and be converted to "journal" and "journal+console" automatically.
3299 * If the service setting User= is set to the "nobody" user, a warning
3300 message is now written to the logs (but the value is nonetheless
3301 accepted). Setting User=nobody is unsafe, since the primary purpose
3302 of the "nobody" user is to own all files whose owner cannot be mapped
3303 locally. It's in particular used by the NFS subsystem and in user
3304 namespacing. By running a service under this user's UID it might get
3305 read and even write access to all these otherwise unmappable files,
3306 which is quite likely a major security problem.
3308 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
3309 and others) now have a size and inode limits applied (50% of RAM for
3310 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
3311 that the implicit kernel default is 50% too, so there is no change
3312 in the size limit for /tmp and /dev/shm.
3314 * nss-mymachines lost support for resolution of users and groups, and
3315 now only does resolution of hostnames. This functionality is now
3316 provided by nss-systemd. Thus, the 'mymachines' entry should be
3317 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
3318 (and 'systemd' added if it is not already there).
3320 * A new kernel command line option systemd.hostname= has been added
3321 that allows controlling the hostname that is initialized early during
3324 * A kernel command line option "udev.blockdev_read_only" has been
3325 added. If specified all hardware block devices that show up are
3326 immediately marked as read-only by udev. This option is useful for
3327 making sure that a specific boot under no circumstances modifies data
3328 on disk. Use "blockdev --setrw" to undo the effect of this, per
3331 * A new boolean kernel command line option systemd.swap= has been
3332 added, which may be used to turn off automatic activation of swap
3333 devices listed in /etc/fstab.
3335 * New kernel command line options systemd.condition-needs-update= and
3336 systemd.condition-first-boot= have been added, which override the
3337 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
3340 * A new kernel command line option systemd.clock-usec= has been added
3341 that allows setting the system clock to the specified time in µs
3342 since Jan 1st, 1970 early during boot. This is in particular useful
3343 in order to make test cases more reliable.
3345 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
3346 systemd-coredump to save core files for suid processes. When saving
3347 the core file, systemd-coredump will use the effective uid and gid of
3348 the process that faulted.
3350 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
3351 now automatically set to "Y" at boot, in order to enable pstore
3352 generation for collection with systemd-pstore.
3354 * We provide a set of udev rules to enable auto-suspend on PCI and USB
3355 devices that were tested to correctly support it. Previously, this
3356 was distributed as a set of udev rules, but has now been replaced by
3357 by a set of hwdb entries (and a much shorter udev rule to take action
3358 if the device modalias matches one of the new hwdb entries).
3360 As before, entries are periodically imported from the database
3361 maintained by the ChromiumOS project. If you have a device that
3362 supports auto-suspend correctly and where it should be enabled by
3363 default, please submit a patch that adds it to the database (see
3364 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
3366 * systemd-udevd gained the new configuration option timeout_signal= as well
3367 as a corresponding kernel command line option udev.timeout_signal=.
3368 The option can be used to configure the UNIX signal that the main
3369 daemon sends to the worker processes on timeout. Setting the signal
3370 to SIGABRT is useful for debugging.
3372 * .link files managed by systemd-udevd gained options RxFlowControl=,
3373 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
3374 order to configure various flow control parameters. They also gained
3375 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
3376 frame ring buffer sizes.
3378 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
3379 enabled systemd-networkd manages all routes configured by other tools.
3381 * .network files managed by systemd-networkd gained a new section
3382 [SR-IOV], in order to configure SR-IOV capable network devices.
3384 * systemd-networkd's [IPv6Prefix] section in .network files gained a
3385 new boolean setting Assign=. If enabled an address from the prefix is
3386 automatically assigned to the interface.
3388 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
3389 controls delegated prefixes assigned by DHCPv6 client. The section
3390 has three settings: SubnetID=, Assign=, and Token=. The setting
3391 SubnetID= allows explicit configuration of the preferred subnet that
3392 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
3393 Assign= is enabled (which is the default) an address from any acquired
3394 delegated prefix is automatically chosen and assigned to the
3395 interface. The setting Token= specifies an optional address generation
3398 * systemd-networkd's [Network] section gained a new setting
3399 IPv4AcceptLocal=. If enabled the interface accepts packets with local
3402 * systemd-networkd gained support for configuring the HTB queuing
3403 discipline in the [HierarchyTokenBucket] and
3404 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
3405 be configured in the [PFIFO] section, "GRED" in
3406 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
3407 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
3408 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
3409 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
3410 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
3411 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
3413 * systemd-networkd gained support for a new Termination= setting in the
3414 [CAN] section for configuring the termination resistor. It also
3415 gained a new ListenOnly= setting for controlling whether to only
3416 listen on CAN interfaces, without interfering with traffic otherwise
3417 (which is useful for debugging/monitoring CAN network
3418 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
3419 been added to configure various CAN-FD aspects.
3421 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
3422 When enabled, DHCPv6 will be attempted right-away without requiring an
3423 Router Advertisement packet suggesting it first (i.e. without the 'M'
3424 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
3425 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
3426 the RA packets suggest it.
3428 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
3429 which may be used to turn off use of the gateway information provided
3430 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
3431 used to configure how to process leases that lack a lifetime option.
3433 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
3434 setting SendVendorOption= allowing configuration of additional vendor
3435 options to send in the DHCP requests/responses. The [DHCPv6] section
3436 gained a new SendOption= setting for sending arbitrary DHCP
3437 options. RequestOptions= has been added to request arbitrary options
3438 from the server. UserClass= has been added to set the DHCP user class
3441 * systemd-networkd's [DHCPServer] section gained a new set of options
3442 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
3443 information about these three protocols in the DHCP lease. It also
3444 gained support for including "MUD" URLs ("Manufacturer Usage
3445 Description"). Support for "MUD" URLs was also added to the LLDP
3446 stack, configurable in the [LLDP] section in .network files.
3448 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
3449 mode. Also, the sections now support a new setting SourceMACAddress=.
3451 * systemd-networkd's .netdev files now support a new setting
3452 VLANProtocol= in the [Bridge] section that allows configuration of
3453 the VLAN protocol to use.
3455 * systemd-networkd supports a new Group= setting in the [Link] section
3456 of the .network files, to control the link group.
3458 * systemd-networkd's [Network] section gained a new
3459 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
3460 link local address is generated.
3462 * A new default .network file is now shipped that matches TUN/TAP
3463 devices that begin with "vt-" in their name. Such interfaces will
3464 have IP routing onto the host links set up automatically. This is
3465 supposed to be used by VM managers to trivially acquire a network
3466 interface which is fully set up for host communication, simply by
3467 carefully picking an interface name to use.
3469 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
3470 which sets the route priority for routes specified by the DHCP server.
3472 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
3473 which configures the vendor class information sent to DHCP server.
3475 * The BlackList= settings in .network files' [DHCPv4] and
3476 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
3477 are still understood to provide compatibility.
3479 * networkctl gained the new "forcerenew" command for forcing all DHCP
3480 server clients to renew their lease. The interface "status" output
3481 will now show numerous additional fields of information about an
3482 interface. There are new "up" and "down" commands to bring specific
3483 interfaces up or down.
3485 * systemd-resolved's DNS= configuration option now optionally accepts a
3486 port number (after ":") and a host name (after "#"). When the host
3487 name is specified, the DNS-over-TLS certificate is validated to match
3488 the specified hostname. Additionally, in case of IPv6 addresses, an
3489 interface may be specified (after "%").
3491 * systemd-resolved may be configured to forward single-label DNS names.
3492 This is not standard-conformant, but may make sense in setups where
3493 public DNS servers are not used.
3495 * systemd-resolved's DNS-over-TLS support gained SNI validation.
3497 * systemd-nspawn's --resolv-conf= switch gained a number of new
3498 supported values. Specifically, options starting with "replace-" are
3499 like those prefixed "copy-" but replace any existing resolv.conf
3500 file. And options ending in "-uplink" and "-stub" can now be used to
3501 propagate other flavours of resolv.conf into the container (as
3502 defined by systemd-resolved).
3504 * The various programs included in systemd can now optionally output
3505 their log messages on stderr prefixed with a timestamp, controlled by
3506 the $SYSTEMD_LOG_TIME environment variable.
3508 * systemctl gained a new "-P" switch that is a shortcut for "--value
3511 * "systemctl list-units" and "systemctl list-machines" no longer hide
3512 their first output column with --no-legend. To hide the first column,
3515 * "systemctl reboot" takes the option "--reboot-argument=".
3516 The optional positional argument to "systemctl reboot" is now
3517 being deprecated in favor of this option.
3519 * systemd-run gained a new switch --slice-inherit. If specified the
3520 unit it generates is placed in the same slice as the systemd-run
3523 * systemd-journald gained support for zstd compression of large fields
3524 in journal files. The hash tables in journal files have been hardened
3525 against hash collisions. This is an incompatible change and means
3526 that journal files created with new systemd versions are not readable
3527 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
3528 environment variable for systemd-journald.service is set to 0 this
3529 new hardening functionality may be turned off, so that generated
3530 journal files remain compatible with older journalctl
3533 * journalctl will now include a clickable link in the default output for
3534 each log message for which an URL with further documentation is
3535 known. This is only supported on terminal emulators that support
3536 clickable hyperlinks, and is turned off if a pager is used (since
3537 "less" still doesn't support hyperlinks,
3538 unfortunately). Documentation URLs may be included in log messages
3539 either by including a DOCUMENTATION= journal field in it, or by
3540 associating a journal message catalog entry with the log message's
3541 MESSAGE_ID, which then carries a "Documentation:" tag.
3543 * journald.conf gained a new boolean setting Audit= that may be used to
3544 control whether systemd-journald will enable audit during
3547 * when systemd-journald's log stream is broken up into multiple lines
3548 because the PID of the sender changed this is indicated in the
3549 generated log records via the _LINE_BREAK=pid-change field.
3551 * journalctl's "-o cat" output mode will now show one or more journal
3552 fields specified with --output-fields= instead of unconditionally
3553 MESSAGE=. This is useful to retrieve a very specific set of fields
3554 without any decoration.
3556 * The sd-journal.h API gained two new functions:
3557 sd_journal_enumerate_available_unique() and
3558 sd_journal_enumerate_available_data() that operate like their
3559 counterparts that lack the _available_ in the name, but skip items
3560 that cannot be read and processed by the local implementation
3561 (i.e. are compressed in an unsupported format or such),
3563 * coredumpctl gained a new --file= switch, matching the same one in
3564 journalctl: a specific journal file may be specified to read the
3567 * coredumps collected by systemd-coredump may now be compressed using
3570 * systemd-binfmt gained a new switch --unregister for unregistering all
3571 registered entries at once. This is now invoked automatically at
3572 shutdown, so that binary formats registered with the "F" flag will
3573 not block clean file system unmounting.
3575 * systemd-notify's --pid= switch gained new values: "parent", "self",
3576 "auto" for controlling which PID to send to the service manager: the
3577 systemd-notify process' PID, or the one of the process invoking it.
3579 * systemd-logind's Session bus object learnt a new method call
3580 SetType() for temporarily updating the session type of an already
3581 allocated session. This is useful for upgrading tty sessions to
3582 graphical ones once a compositor is invoked.
3584 * systemd-socket-proxy gained a new switch --exit-idle-time= for
3585 configuring an exit-on-idle time.
3587 * systemd-repart's --empty= setting gained a new value "create". If
3588 specified a new empty regular disk image file is created under the
3589 specified name. Its size may be specified with the new --size=
3590 option. The latter is also supported without the "create" mode, in
3591 order to grow existing disk image files to the specified size. These
3592 two new options are useful when creating or manipulating disk images
3593 instead of operating on actual block devices.
3595 * systemd-repart drop-ins now support a new UUID= setting to control
3596 the UUID to assign to a newly created partition.
3598 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
3601 * systemd-repart's Label= setting now support the usual, simple
3602 specifier expansion.
3604 * systemd-homed's LUKS backend gained the ability to discard empty file
3605 system blocks automatically when the user logs out. This is enabled
3606 by default to ensure that home directories take minimal space when
3607 logged out but get full size guarantees when logged in. This may be
3608 controlled with the new --luks-offline-discard= switch to homectl.
3610 * If systemd-homed detects that /home/ is encrypted as a whole it will
3611 now default to the directory or subvolume backends instead of the
3612 LUKS backend, in order to avoid double encryption. The default
3613 storage and file system may now be configured explicitly, too, via
3614 the new /etc/systemd/homed.conf configuration file.
3616 * systemd-homed now supports unlocking home directories with FIDO2
3617 security tokens that support the 'hmac-secret' extension, in addition
3618 to the existing support for PKCS#11 security token unlocking
3619 support. Note that many recent hardware security tokens support both
3620 interfaces. The FIDO2 support is accessible via homectl's
3621 --fido2-device= option.
3623 * homectl's --pkcs11-uri= setting now accepts two special parameters:
3624 if "auto" is specified and only one suitable PKCS#11 security token
3625 is plugged in, its URL is automatically determined and enrolled for
3626 unlocking the home directory. If "list" is specified a brief table of
3627 suitable PKCS#11 security tokens is shown. Similar, the new
3628 --fido2-device= option also supports these two special values, for
3629 automatically selecting and listing suitable FIDO2 devices.
3631 * The /etc/crypttab tmp option now optionally takes an argument
3632 selecting the file system to use. Moreover, the default is now
3633 changed from ext2 to ext4.
3635 * There's a new /etc/crypttab option "keyfile-erase". If specified the
3636 key file listed in the same line is removed after use, regardless if
3637 volume activation was successful or not. This is useful if the key
3638 file is only acquired transiently at runtime and shall be erased
3639 before the system continues to boot.
3641 * There's also a new /etc/crypttab option "try-empty-password". If
3642 specified, before asking the user for a password it is attempted to
3643 unlock the volume with an empty password. This is useful for
3644 installing encrypted images whose password shall be set on first boot
3645 instead of at installation time.
3647 * systemd-cryptsetup will now attempt to load the keys to unlock
3648 volumes with automatically from files in
3649 /etc/cryptsetup-keys.d/<volume>.key and
3650 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
3652 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
3653 /etc/crypttab, during boot.
3655 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
3656 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
3659 * A new generator systemd-xdg-autostart-generator has been added. It
3660 generates systemd unit files from XDG autostart .desktop files, and
3661 may be used to let the systemd user instance manage services that are
3662 started automatically as part of the desktop session.
3664 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
3665 to query and change the firmware's 'reboot into firmware' setup flag.
3667 * systemd-firstboot gained a new switch --kernel-command-line= that may
3668 be used to initialize the /etc/kernel/cmdline file of the image. It
3669 also gained a new switch --root-password-hashed= which is like
3670 --root-password= but accepts a pre-hashed UNIX password as
3671 argument. The new option --delete-root-password may be used to unset
3672 any password for the root user (dangerous!). The --root-shell= switch
3673 may be used to control the shell to use for the root account. A new
3674 --force option may be used to override any already set settings with
3675 the parameters specified on the command line (by default, the tool
3676 will not override what has already been set before, i.e. is purely
3679 * systemd-firstboot gained support for a new --image= switch, which is
3680 similar to --root= but accepts the path to a disk image file, on
3681 which it then operates.
3683 * A new sd-path.h API has been added to libsystemd. It provides a
3684 simple API for retrieving various search paths and primary
3685 directories for various resources.
3687 * A new call sd_notify_barrier() has been added to the sd-daemon.h
3688 API. The call will block until all previously sent sd_notify()
3689 messages have been processed by the service manager. This is useful
3690 to remove races caused by a process already having disappeared at the
3691 time a notification message is processed by the service manager,
3692 making correct attribution impossible. The systemd-notify tool will
3693 now make use of this call implicitly, but this can be turned off again
3694 via the new --no-block switch.
3696 * When sending a file descriptor (fd) to the service manager to keep
3697 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
3698 may be specified. If passed the service manager will refrain from
3699 poll()ing on the file descriptor. Traditionally (and when the
3700 parameter is not specified), the service manager will poll it for
3701 POLLHUP or POLLERR events, and immediately close the fds in that
3704 * The service manager (PID1) gained a new D-Bus method call
3705 SetShowStatus() which may be used to control whether it shall show
3706 boot-time status output on the console. This method has a similar
3707 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
3709 * The sd-bus API gained a number of convenience functions that take
3710 va_list arguments rather than "...". For example, there's now
3711 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
3712 it easier to build wrappers that accept variadic arguments and want
3713 to pass a ready va_list structure to sd-bus.
3715 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
3716 flag which alters how the userdata pointer to pass to the callbacks
3717 is determined. When the flag is set, the offset field is converted
3718 as-is into a pointer, without adding it to the object pointer the
3719 vtable is associated with.
3721 * sd-bus now exposes four new functions:
3722 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
3723 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
3724 validate strings to check if they qualify as various D-Bus concepts.
3726 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
3727 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
3728 that simplify adding argument names to D-Bus methods and signals.
3730 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
3732 * Various D-Bus APIs of systemd daemons now have man pages that
3733 document the methods, signals and properties.
3735 * The expectations on user/group name syntax are now documented in
3736 detail; documentation on how classic home directories may be
3737 converted into home directories managed by homed has been added;
3738 documentation regarding integration of homed/userdb functionality in
3739 desktops has been added:
3741 https://systemd.io/USER_NAMES
3742 https://systemd.io/CONVERTING_TO_HOMED
3743 https://systemd.io/USERDB_AND_DESKTOPS
3745 * Documentation for the on-disk Journal file format has been updated
3746 and has now moved to:
3748 https://systemd.io/JOURNAL_FILE_FORMAT
3750 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
3751 has been extended by a set of environment variables that expose
3752 select fields from the host's os-release file to the container
3753 payload. Similarly, host's os-release files can be mounted into the
3754 container underneath /run/host. Together, those mechanisms provide a
3755 standardized way to expose information about the host to the
3756 container payload. Both interfaces are implemented in systemd-nspawn.
3758 * All D-Bus services shipped in systemd now implement the generic
3759 LogControl1 D-Bus API which allows clients to change log level +
3760 target of the service during runtime.
3762 * Only relevant for developers: the mkosi.default symlink has been
3763 dropped from version control. Please create a symlink to one of the
3764 distribution-specific defaults in .mkosi/ based on your preference.
3766 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
3767 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
3768 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
3769 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
3770 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
3771 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
3772 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
3773 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
3774 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
3775 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
3776 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
3777 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
3778 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
3779 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
3780 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
3781 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
3782 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
3783 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
3784 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
3785 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
3786 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
3787 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
3788 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
3789 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
3790 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
3791 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
3792 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
3793 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
3794 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
3795 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
3796 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
3797 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
3798 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
3799 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
3800 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
3801 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
3802 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
3803 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
3805 – Warsaw, 2020-07-30
3809 * A new tool "systemd-repart" has been added, that operates as an
3810 idempotent declarative repartitioner for GPT partition tables.
3811 Specifically, a set of partitions that must or may exist can be
3812 configured via drop-in files, and during every boot the partition
3813 table on disk is compared with these files, creating missing
3814 partitions or growing existing ones based on configurable relative
3815 and absolute size constraints. The tool is strictly incremental,
3816 i.e. does not delete, shrink or move partitions, but only adds and
3817 grows them. The primary use-case is OS images that ship in minimized
3818 form, that on first boot are grown to the size of the underlying
3819 block device or augmented with additional partitions. For example,
3820 the root partition could be extended to cover the whole disk, or a
3821 swap or /home partitions could be added on first boot. It can also be
3822 used for systems that use an A/B update scheme but ship images with
3823 just the A partition, with B added on first boot. The tool is
3824 primarily intended to be run in the initrd, shortly before
3825 transitioning into the host OS, but can also be run after the
3826 transition took place. It automatically discovers the disk backing
3827 the root file system, and should hence not require any additional
3828 configuration besides the partition definition drop-ins. If no
3829 configuration drop-ins are present, no action is taken.
3831 * A new component "userdb" has been added, along with a small daemon
3832 "systemd-userdbd.service" and a client tool "userdbctl". The framework
3833 allows defining rich user and group records in a JSON format,
3834 extending on the classic "struct passwd" and "struct group"
3835 structures. Various components in systemd have been updated to
3836 process records in this format, including systemd-logind and
3837 pam-systemd. The user records are intended to be extensible, and
3838 allow setting various resource management, security and runtime
3839 parameters that shall be applied to processes and sessions of the
3840 user as they log in. This facility is intended to allow associating
3841 such metadata directly with user/group records so that they can be
3842 produced, extended and consumed in unified form. We hope that
3843 eventually frameworks such as sssd will generate records this way, so
3844 that for the first time resource management and various other
3845 per-user settings can be configured in LDAP directories and then
3846 provided to systemd (specifically to systemd-logind and pam-system)
3847 to apply on login. For further details see:
3849 https://systemd.io/USER_RECORD
3850 https://systemd.io/GROUP_RECORD
3851 https://systemd.io/USER_GROUP_API
3853 * A small new service systemd-homed.service has been added, that may be
3854 used to securely manage home directories with built-in encryption.
3855 The complete user record data is unified with the home directory,
3856 thus making home directories naturally migratable. Its primary
3857 back-end is based on LUKS volumes, but fscrypt, plain directories,
3858 and other storage schemes are also supported. This solves a couple of
3859 problems we saw with traditional ways to manage home directories, in
3860 particular when it comes to encryption. For further discussion of
3861 this, see the video of Lennart's talk at AllSystemsGo! 2019:
3863 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
3865 For further details about the format and expectations on home
3866 directories this new daemon makes, see:
3868 https://systemd.io/HOME_DIRECTORY
3870 * systemd-journald is now multi-instantiable. In addition to the main
3871 instance systemd-journald.service there's now a template unit
3872 systemd-journald@.service, with each instance defining a new named
3873 log 'namespace' (whose name is specified via the instance part of the
3874 unit name). A new unit file setting LogNamespace= has been added,
3875 taking such a namespace name, that assigns services to the specified
3876 log namespaces. As each log namespace is serviced by its own
3877 independent journal daemon, this functionality may be used to improve
3878 performance and increase isolation of applications, at the price of
3879 losing global message ordering. Each instance of journald has a
3880 separate set of configuration files, with possibly different disk
3881 usage limitations and other settings.
3883 journalctl now takes a new option --namespace= to show logs from a
3884 specific log namespace. The sd-journal.h API gained
3885 sd_journal_open_namespace() for opening the log stream of a specific
3886 log namespace. systemd-journald also gained the ability to exit on
3887 idle, which is useful in the context of log namespaces, as this means
3888 log daemons for log namespaces can be activated automatically on
3889 demand and will stop automatically when no longer used, minimizing
3892 * When systemd-tmpfiles copies a file tree using the 'C' line type it
3893 will now label every copied file according to the SELinux database.
3895 * When systemd/PID 1 detects it is used in the initrd it will now boot
3896 into initrd.target rather than default.target by default. This should
3897 make it simpler to build initrds with systemd as for many cases the
3898 only difference between a host OS image and an initrd image now is
3899 the presence of the /etc/initrd-release file.
3901 * A new kernel command line option systemd.cpu_affinity= is now
3902 understood. It's equivalent to the CPUAffinity= option in
3903 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
3904 itself and the default for all other processes.
3906 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
3907 equivalent), the SELinux database is now reloaded, ensuring that
3908 sockets and other file system objects are generated taking the new
3909 database into account.
3911 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
3912 "quiet" has been changed to imply that instead of
3913 "systemd.show-status=auto". In this mode, only messages about errors
3914 and significant delays in boot are shown on the console.
3916 * The sd-event.h API gained native support for the new Linux "pidfd"
3917 concept. This permits watching processes using file descriptors
3918 instead of PID numbers, which fixes a number of races and makes
3919 process supervision more robust and efficient. All of systemd's
3920 components will now use pidfds if the kernel supports it for process
3921 watching, with the exception of PID 1 itself, unfortunately. We hope
3922 to move PID 1 to exclusively using pidfds too eventually, but this
3923 requires some more kernel work first. (Background: PID 1 watches
3924 processes using waitid() with the P_ALL flag, and that does not play
3925 together nicely with pidfds yet.)
3927 * Closely related to this, the sd-event.h API gained two new calls
3928 sd_event_source_send_child_signal() (for sending a signal to a
3929 watched process) and sd_event_source_get_child_process_own() (for
3930 marking a process so that it is killed automatically whenever the
3931 event source watching it is freed).
3933 * systemd-networkd gained support for configuring Token Bucket Filter
3934 (TBF) parameters in its qdisc configuration support. Similarly,
3935 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
3936 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
3938 * systemd-networkd gained support for Intermediate Functional Block
3939 (IFB) network devices.
3941 * systemd-networkd gained support for configuring multi-path IP routes,
3942 using the new MultiPathRoute= setting in the [Route] section.
3944 * systemd-networkd's DHCPv4 client has been updated to support a new
3945 SendDecline= option. If enabled, duplicate address detection is done
3946 after a DHCP offer is received from the server. If a conflict is
3947 detected, the address is declined. The DHCPv4 client also gained
3948 support for a new RouteMTUBytes= setting that allows to configure the
3949 MTU size to be used for routes generated from DHCPv4 leases.
3951 * The PrefixRoute= setting in systemd-networkd's [Address] section of
3952 .network files has been deprecated, and replaced by AddPrefixRoute=,
3953 with its sense inverted.
3955 * The Gateway= setting of [Route] sections of .network files gained
3956 support for a special new value "_dhcp". If set, the configured
3957 static route uses the gateway host configured via DHCP.
3959 * New User= and SuppressPrefixLength= settings have been implemented
3960 for the [RoutingPolicyRule] section of .network files to configure
3961 source routing based on UID ranges and prefix length, respectively.
3963 * The Type= match property of .link files has been generalized to
3964 always match the device type shown by 'networkctl status', even for
3965 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
3968 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
3969 D-Bus message object as "sensitive". Those objects are erased from
3970 memory when they are freed. This concept is intended to be used for
3971 messages that contain security sensitive data. A new flag
3972 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
3973 in sd-bus vtables, causing any incoming and outgoing messages of
3974 those methods to be implicitly marked as "sensitive".
3976 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
3977 contents of a message (or parts thereof) to standard output for
3980 * systemd-sysusers gained support for creating users with the primary
3981 group named differently than the user.
3983 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
3984 gained support for growing XFS partitions. Previously it supported
3985 only ext4 and btrfs partitions.
3987 * The support for /etc/crypttab gained a new x-initrd.attach option. If
3988 set, the specified encrypted volume is unlocked already in the
3989 initrd. This concept corresponds to the x-initrd.mount option in
3992 * systemd-cryptsetup gained native support for unlocking encrypted
3993 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
3994 encryption of volumes to YubiKeys. This is exposed in the new
3995 pkcs11-uri= option in /etc/crypttab.
3997 * The /etc/fstab support in systemd now supports two new mount options
3998 x-systemd.{required,wanted}-by=, for explicitly configuring the units
3999 that the specified mount shall be pulled in by, in place of
4000 the usual local-fs.target/remote-fs.target.
4002 * The https://systemd.io/ web site has been relaunched, directly
4003 populated with most of the documentation included in the systemd
4004 repository. systemd also acquired a new logo, thanks to Tobias
4007 * systemd-udevd gained support for managing "alternative" network
4008 interface names, as supported by new Linux kernels. For the first
4009 time this permits assigning multiple (and longer!) names to a network
4010 interface. systemd-udevd will now by default assign the names
4011 generated via all supported naming schemes to each interface. This
4012 may be further tweaked with .link files and the AlternativeName= and
4013 AlternativeNamesPolicy= settings. Other components of systemd have
4014 been updated to support the new alternative names wherever
4015 appropriate. For example, systemd-nspawn will now generate
4016 alternative interface names for the host-facing side of container
4017 veth links based on the full container name without truncation.
4019 * systemd-nspawn interface naming logic has been updated in another way
4020 too: if the main interface name (i.e. as opposed to new-style
4021 "alternative" names) based on the container name is truncated, a
4022 simple hashing scheme is used to give different interface names to
4023 multiple containers whose names all begin with the same prefix. Since
4024 this changes the primary interface names pointing to containers if
4025 truncation happens, the old scheme may still be requested by
4026 selecting an older naming scheme, via the net.naming-scheme= kernel
4027 command line option.
4029 * PrivateUsers= in service files now works in services run by the
4030 systemd --user per-user instance of the service manager.
4032 * A new per-service sandboxing option ProtectClock= has been added that
4033 locks down write access to the system clock. It takes away device
4034 node access to /dev/rtc as well as the system calls that set the
4035 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
4036 Note that this option does not affect access to auxiliary services
4037 that allow changing the clock, for example access to
4040 * The systemd-id128 tool gained a new "show" verb for listing or
4041 resolving a number of well-known UUIDs/128bit IDs, currently mostly
4042 GPT partition table types.
4044 * The Discoverable Partitions Specification has been updated to support
4045 /var and /var/tmp partition discovery. Support for this has been
4046 added to systemd-gpt-auto-generator. For details see:
4048 https://systemd.io/DISCOVERABLE_PARTITIONS
4050 * "systemctl list-unit-files" has been updated to show a new column
4051 with the suggested enablement state based on the vendor preset files
4052 for the respective units.
4054 * "systemctl" gained a new option "--with-dependencies". If specified
4055 commands such as "systemctl status" or "systemctl cat" will now show
4056 all specified units along with all units they depend on.
4058 * networkctl gained support for showing per-interface logs in its
4061 * systemd-networkd-wait-online gained support for specifying the maximum
4062 operational state to wait for, and to wait for interfaces to
4065 * The [Match] section of .link and .network files now supports a new
4066 option PermanentMACAddress= which may be used to check against the
4067 permanent MAC address of a network device even if a randomized MAC
4070 * The [TrafficControlQueueingDiscipline] section in .network files has
4071 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
4072 dropped from the individual setting names.
4074 * Any .link and .network files that have an empty [Match] section (this
4075 also includes empty and commented-out files) will now be
4076 rejected. systemd-udev and systemd-networkd started warning about
4077 such files in version 243.
4079 * systemd-logind will now validate access to the operation of changing
4080 the virtual terminal via a polkit action. By default, only users
4081 with at least one session on a local VT are granted permission.
4083 * When systemd sets up PAM sessions that invoked service processes
4084 shall run in, the pam_setcred() API is now invoked, thus permitting
4085 PAM modules to set additional credentials for the processes.
4087 * portablectl attach/detach verbs now accept --now and --enable options
4088 to combine attachment with enablement and invocation, or detachment
4089 with stopping and disablement.
4091 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
4092 fixed, which in turn exposed bugs in unit configuration of services
4093 which have Type=oneshot and should only run once, but do not have
4094 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
4095 service may be started again after exiting successfully, for example
4096 as a dependency in another transaction. Affected services included
4097 some internal systemd services (most notably
4098 systemd-vconsole-setup.service, which was updated to have
4099 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
4100 plymouth has been suitably updated or patched before upgrading to
4101 this systemd release. See
4102 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
4103 additional discussion.
4105 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
4106 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
4107 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
4108 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
4109 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
4110 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
4111 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
4112 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
4113 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
4114 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
4115 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
4116 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
4117 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
4118 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
4119 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
4120 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
4121 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
4122 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
4123 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
4124 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
4125 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
4126 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
4127 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
4128 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
4131 – Warsaw, 2020-03-06
4135 * Support for the cpuset cgroups v2 controller has been added.
4136 Processes may be restricted to specific CPUs using the new
4137 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
4138 AllowedMemoryNodes= setting.
4140 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
4141 now be configured using a new RestartKillSignal= setting. This
4142 allows units which signals to request termination to implement
4143 different behaviour when stopping in preparation for a restart.
4145 * "systemctl clean" may now be used also for socket, mount, and swap
4148 * systemd will also read configuration options from the EFI variable
4149 SystemdOptions. This may be used to configure systemd behaviour when
4150 modifying the kernel command line is inconvenient, but configuration
4151 on disk is read too late, for example for the options related to
4152 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
4153 set the EFI variable.
4155 * systemd will now disable printk ratelimits in early boot. This should
4156 allow us to capture more logs from the early boot phase where normal
4157 storage is not available and the kernel ring buffer is used for
4158 logging. Configuration on the kernel command line has higher priority
4159 and overrides the systemd setting.
4161 systemd programs which log to /dev/kmsg directly use internal
4162 ratelimits to prevent runaway logging. (Normally this is only used
4163 during early boot, so in practice this change has very little
4166 * Unit files now support top level dropin directories of the form
4167 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
4168 that affects all corresponding unit files.
4170 * systemctl gained support for 'stop --job-mode=triggering' which will
4171 stop the specified unit and any units which could trigger it.
4173 * Unit status display now includes units triggering and triggered by
4174 the unit being shown.
4176 * The RuntimeMaxSec= setting is now supported by scopes, not just
4177 .service units. This is particularly useful for PAM sessions which
4178 create a scope unit for the user login. systemd.runtime_max_sec=
4179 setting may used with the pam_systemd module to limit the duration
4180 of the PAM session, for example for time-limited logins.
4182 * A new @pkey system call group is now defined to make it easier to
4183 allow-list memory protection syscalls for containers and services
4184 which need to use them.
4186 * systemd-udevd: removed the 30s timeout for killing stale workers on
4187 exit. systemd-udevd now waits for workers to finish. The hard-coded
4188 exit timeout of 30s was too short for some large installations, where
4189 driver initialization could be prematurely interrupted during initrd
4190 processing if the root file system had been mounted and init was
4191 preparing to switch root. If udevd is run without systemd and workers
4192 are hanging while udevd receives an exit signal, udevd will now exit
4193 when udev.event_timeout is reached for the last hanging worker. With
4194 systemd, the exit timeout can additionally be configured using
4195 TimeoutStopSec= in systemd-udevd.service.
4197 * udev now provides a program (fido_id) that identifies FIDO CTAP1
4198 ("U2F")/CTAP2 security tokens based on the usage declared in their
4199 report and descriptor and outputs suitable environment variables.
4200 This replaces the externally maintained allow lists of all known
4201 security tokens that were used previously.
4203 * Automatically generated autosuspend udev rules for allow-listed
4204 devices have been imported from the Chromium OS project. This should
4205 improve power saving with many more devices.
4207 * udev gained a new "CONST{key}=value" setting that allows matching
4208 against system-wide constants without forking a helper binary.
4209 Currently "arch" and "virt" keys are supported.
4211 * udev now opens CDROMs in non-exclusive mode when querying their
4212 capabilities. This should fix issues where other programs trying to
4213 use the CDROM cannot gain access to it, but carries a risk of
4214 interfering with programs writing to the disk, if they did not open
4215 the device in exclusive mode as they should.
4217 * systemd-networkd does not create a default route for IPv4 link local
4218 addressing anymore. The creation of the route was unexpected and was
4219 breaking routing in various cases, but people who rely on it being
4220 created implicitly will need to adjust. Such a route may be requested
4221 with DefaultRouteOnDevice=yes.
4223 Similarly, systemd-networkd will not assign a link-local IPv6 address
4224 when IPv6 link-local routing is not enabled.
4226 * Receive and transmit buffers may now be configured on links with
4227 the new RxBufferSize= and TxBufferSize= settings.
4229 * systemd-networkd may now advertise additional IPv6 routes. A new
4230 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
4233 * systemd-networkd may now configure "next hop" routes using the
4234 [NextHop] section and Gateway= and Id= settings.
4236 * systemd-networkd will now retain DHCP config on restarts by default
4237 (but this may be overridden using the KeepConfiguration= setting).
4238 The default for SendRelease= has been changed to true.
4240 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
4241 received from the server.
4243 The client will use the received SIP server list if UseSIP=yes is
4246 The client may be configured to request specific options from the
4247 server using a new RequestOptions= setting.
4249 The client may be configured to send arbitrary options to the server
4250 using a new SendOption= setting.
4252 A new IPServiceType= setting has been added to configure the "IP
4253 service type" value used by the client.
4255 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
4256 request prefix hints in the DHCPv6 solicitation.
4258 * The DHCPv4 server may be configured to send arbitrary options using
4259 a new SendOption= setting.
4261 * The DHCPv4 server may now be configured to emit SIP server list using
4262 the new EmitSIP= and SIP= settings.
4264 * systemd-networkd and networkctl may now renew DHCP leases on demand.
4265 networkctl has a new 'networkctl renew' verb.
4267 * systemd-networkd may now reconfigure links on demand. networkctl
4268 gained two new verbs: "reload" will reload the configuration, and
4269 "reconfigure DEVICE…" will reconfigure one or more devices.
4271 * .network files may now match on SSID and BSSID of a wireless network,
4272 i.e. the access point name and hardware address using the new SSID=
4273 and BSSID= options. networkctl will display the current SSID and
4274 BSSID for wireless links.
4276 .network files may also match on the wireless network type using the
4277 new WLANInterfaceType= option.
4279 * systemd-networkd now includes default configuration that enables
4280 link-local addressing when connected to an ad-hoc wireless network.
4282 * systemd-networkd may configure the Traffic Control queueing
4283 disciplines in the kernel using the new
4284 [TrafficControlQueueingDiscipline] section and Parent=,
4285 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
4286 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
4287 NetworkEmulatorDuplicateRate= settings.
4289 * systemd-tmpfiles gained a new w+ setting to append to files.
4291 * systemd-analyze dump will now report when the memory configuration in
4292 the kernel does not match what systemd has configured (usually,
4293 because some external program has modified the kernel configuration
4296 * systemd-analyze gained a new --base-time= switch instructs the
4297 'calendar' verb to resolve times relative to that timestamp instead
4298 of the present time.
4300 * journalctl --update-catalog now produces deterministic output (making
4301 reproducible image builds easier).
4303 * A new devicetree-overlay setting is now documented in the Boot Loader
4306 * The default value of the WatchdogSec= setting used in systemd
4307 services (the ones bundled with the project itself) may be set at
4308 configuration time using the -Dservice-watchdog= setting. If set to
4309 empty, the watchdogs will be disabled.
4311 * systemd-resolved validates IP addresses in certificates now when GnuTLS
4314 * libcryptsetup >= 2.0.1 is now required.
4316 * A configuration option -Duser-path= may be used to override the $PATH
4317 used by the user service manager. The default is again to use the same
4318 path as the system manager.
4320 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
4321 outputting the 128bit IDs in UUID format (i.e. in the "canonical
4324 * Service units gained a new sandboxing option ProtectKernelLogs= which
4325 makes sure the program cannot get direct access to the kernel log
4326 buffer anymore, i.e. the syslog() system call (not to be confused
4327 with the API of the same name in libc, which is not affected), the
4328 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
4329 inaccessible to the service. It's recommended to enable this setting
4330 for all services that should not be able to read from or write to the
4331 kernel log buffer, which are probably almost all.
4333 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
4334 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
4335 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
4336 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
4337 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
4338 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
4339 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
4340 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
4341 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
4342 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
4343 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
4344 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
4345 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
4346 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
4347 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
4348 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
4349 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
4350 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
4351 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
4352 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
4353 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
4354 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
4355 Zach Smith, Zbigniew Jędrzejewski-Szmek
4357 – Warsaw, 2019-11-29
4361 * This release enables unprivileged programs (i.e. requiring neither
4362 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
4363 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
4364 kernel for the whole UNIX group range, i.e. all processes. This
4365 change should be reasonably safe, as the kernel support for it was
4366 specifically implemented to allow safe access to ICMP Echo for
4367 processes lacking any privileges. If this is not desirable, it can be
4368 disabled again by setting the parameter to "1 0".
4370 * Previously, filters defined with SystemCallFilter= would have the
4371 effect that any calling of an offending system call would terminate
4372 the calling thread. This behaviour never made much sense, since
4373 killing individual threads of unsuspecting processes is likely to
4374 create more problems than it solves. With this release the default
4375 action changed from killing the thread to killing the whole
4376 process. For this to work correctly both a kernel version (>= 4.14)
4377 and a libseccomp version (>= 2.4.0) supporting this new seccomp
4378 action is required. If an older kernel or libseccomp is used the old
4379 behaviour continues to be used. This change does not affect any
4380 services that have no system call filters defined, or that use
4381 SystemCallErrorNumber= (and thus see EPERM or another error instead
4382 of being killed when calling an offending system call). Note that
4383 systemd documentation always claimed that the whole process is
4384 killed. With this change behaviour is thus adjusted to match the
4387 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
4388 4194304 by default, i.e. the full 22bit range the kernel allows, up
4389 from the old 16bit range. This should improve security and
4390 robustness, as PID collisions are made less likely (though certainly
4391 still possible). There are rumours this might create compatibility
4392 problems, though at this moment no practical ones are known to
4393 us. Downstream distributions are hence advised to undo this change in
4394 their builds if they are concerned about maximum compatibility, but
4395 for everybody else we recommend leaving the value bumped. Besides
4396 improving security and robustness this should also simplify things as
4397 the maximum number of allowed concurrent tasks was previously bounded
4398 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
4399 only a single knob is left ("kernel.threads-max"). There have been
4400 concerns that usability is affected by this change because larger PID
4401 numbers are harder to type, but we believe the change from 5 digits
4402 to 7 digits doesn't hamper usability.
4404 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
4405 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
4406 hierarchically set default memory protection values for a particular
4407 subtree of the unit hierarchy.
4409 * Memory protection directives can now take a value of zero, allowing
4410 explicit opting out of a default value propagated by an ancestor.
4412 * systemd now defaults to the "unified" cgroup hierarchy setup during
4413 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
4414 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
4415 change reflects the fact that cgroupsv2 support has matured
4416 substantially in both systemd and in the kernel, and is clearly the
4417 way forward. Downstream production distributions might want to
4418 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
4419 their builds as unfortunately the popular container managers have not
4420 caught up with the kernel API changes.
4422 * Man pages are not built by default anymore (html pages were already
4423 disabled by default), to make development builds quicker. When
4424 building systemd for a full installation with documentation, meson
4425 should be called with -Dman=true and/or -Dhtml=true as appropriate.
4426 The default was changed based on the assumption that quick one-off or
4427 repeated development builds are much more common than full optimized
4428 builds for installation, and people need to pass various other
4429 options to when doing "proper" builds anyway, so the gain from making
4430 development builds quicker is bigger than the one time disruption for
4433 Two scripts are created in the *build* directory to generate and
4434 preview man and html pages on demand, e.g.:
4436 build/man/man systemctl
4437 build/man/html systemd.index
4439 * libidn2 is used by default if both libidn2 and libidn are installed.
4440 Please use -Dlibidn=true if libidn is preferred.
4442 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
4443 big-endian machines. Before, bytes were written and read in native
4444 machine order as exposed by the native libc __cpu_mask interface.
4445 Now, little-endian order is always used (CPUs 0–7 are described by
4446 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
4447 This change fixes D-Bus calls that cross endianness boundary.
4449 The presentation format used for CPUAffinity= by "systemctl show" and
4450 "systemd-analyze dump" is changed to present CPU indices instead of
4451 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
4452 shown as CPUAffinity=03000000000000000000000000000… (on
4453 little-endian) or CPUAffinity=00000000000000300000000000000… (on
4454 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
4455 input format. The maximum integer that will be printed in the new
4456 format is 8191 (four digits), while the old format always used a very
4457 long number (with the length varying by architecture), so they can be
4458 unambiguously distinguished.
4460 * /usr/sbin/halt.local is no longer supported. Implementation in
4461 distributions was inconsistent and it seems this functionality was
4464 To replace this functionality, users should:
4465 - either define a new unit and make it a dependency of final.target
4466 (systemctl add-wants final.target my-halt-local.service)
4467 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
4468 and ensure that it accepts "halt", "poweroff", "reboot", and
4469 "kexec" as an argument, see the description in systemd-shutdown(8).
4471 * When a [Match] section in .link or .network file is empty (contains
4472 no match patterns), a warning will be emitted. Please add any "match
4473 all" pattern instead, e.g. OriginalName=* or Name=* in case all
4474 interfaces should really be matched.
4476 * A new setting NUMAPolicy= may be used to set process memory
4477 allocation policy. This setting can be specified in
4478 /etc/systemd/system.conf and hence will set the default policy for
4479 PID1. The default policy can be overridden on a per-service
4480 basis. The related setting NUMAMask= is used to specify NUMA node
4481 mask that should be associated with the selected policy.
4483 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
4484 generates when processes it manages are reaching their memory limits,
4485 and will place their units in a special state, and optionally kill or
4486 stop the whole unit.
4488 * The service manager will now expose bus properties for the IO
4489 resources used by units. This information is also shown in "systemctl
4490 status" now (for services that have IOAccounting=yes set). Moreover,
4491 the IO accounting data is included in the resource log message
4492 generated whenever a unit stops.
4494 * Units may now configure an explicit timeout to wait for when killed
4495 with SIGABRT, for example when a service watchdog is hit. Previously,
4496 the regular TimeoutStopSec= timeout was applied in this case too —
4497 now a separate timeout may be set using TimeoutAbortSec=.
4499 * Services may now send a special WATCHDOG=trigger message with
4500 sd_notify() to trigger an immediate "watchdog missed" event, and thus
4501 trigger service termination. This is useful both for testing watchdog
4502 handling, but also for defining error paths in services, that shall
4503 be handled the same way as watchdog events.
4505 * There are two new per-unit settings IPIngressFilterPath= and
4506 IPEgressFilterPath= which allow configuration of a BPF program
4507 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
4508 to apply to the IP packet ingress/egress path of all processes of a
4509 unit. This is useful to allow running systemd services with BPF
4510 programs set up externally.
4512 * systemctl gained a new "clean" verb for removing the state, cache,
4513 runtime or logs directories of a service while it is terminated. The
4514 new verb may also be used to remove the state maintained on disk for
4515 timer units that have Persistent= configured.
4517 * During the last phase of shutdown systemd will now automatically
4518 increase the log level configured in the "kernel.printk" sysctl so
4519 that any relevant loggable events happening during late shutdown are
4520 made visible. Previously, loggable events happening so late during
4521 shutdown were generally lost if the "kernel.printk" sysctl was set to
4522 high thresholds, as regular logging daemons are terminated at that
4523 time and thus nothing is written to disk.
4525 * If processes terminated during the last phase of shutdown do not exit
4526 quickly systemd will now show their names after a short time, to make
4527 debugging easier. After a longer timeout they are forcibly killed,
4530 * journalctl (and the other tools that display logs) will now highlight
4531 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
4532 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
4533 are now shown in blue color, to separate them visually from regular
4534 logs. References to configuration files are now turned into clickable
4535 links on terminals that support that.
4537 * systemd-journald will now stop logging to /var/log/journal during
4538 shutdown when /var/ is on a separate mount, so that it can be
4539 unmounted safely during shutdown.
4541 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
4543 * systemd-resolved "Cache=" configuration option in resolved.conf has
4544 been extended to also accept the 'no-negative' value. Previously,
4545 only a boolean option was allowed (yes/no), having yes as the
4546 default. If this option is set to 'no-negative', negative answers are
4547 not cached while the old cache heuristics are used positive answers.
4548 The default remains unchanged.
4550 * The predictable naming scheme for network devices now supports
4551 generating predictable names for "netdevsim" devices.
4553 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
4556 Those two changes form a new net.naming-policy-scheme= entry.
4557 Distributions which want to preserve naming stability may want to set
4558 the -Ddefault-net-naming-scheme= configuration option.
4560 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
4561 interfaces natively.
4563 * systemd-networkd's bridge FDB support now allows configuration of a
4564 destination address for each entry (Destination=), as well as the
4565 VXLAN VNI (VNI=), as well as an option to declare what an entry is
4566 associated with (AssociatedWith=).
4568 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
4569 option for configuring the maximum number of DHCP lease requests. It
4570 also learnt a new BlackList= option for deny-listing DHCP servers (a
4571 similar setting has also been added to the IPv6 RA client), as well
4572 as a SendRelease= option for configuring whether to send a DHCP
4573 RELEASE message when terminating.
4575 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
4576 separately in the [DHCPv4] and [DHCPv6] sections.
4578 * systemd-networkd's DHCP support will now optionally create an
4579 implicit host route to the DNS server specified in the DHCP lease, in
4580 addition to the routes listed explicitly in the lease. This should
4581 ensure that in multi-homed systems DNS traffic leaves the systems on
4582 the interface that acquired the DNS server information even if other
4583 routes such as default routes exist. This behaviour may be turned on
4584 with the new RoutesToDNS= option.
4586 * systemd-networkd's VXLAN support gained a new option
4587 GenericProtocolExtension= for enabling VXLAN Generic Protocol
4588 Extension support, as well as IPDoNotFragment= for setting the IP
4589 "Don't fragment" bit on outgoing packets. A similar option has been
4590 added to the GENEVE support.
4592 * In systemd-networkd's [Route] section you may now configure
4593 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
4594 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
4595 propagation. The Type= setting now supports local, broadcast,
4596 anycast, multicast, any, xresolve routes, too.
4598 * systemd-networkd's [Network] section learnt a new option
4599 DefaultRouteOnDevice= for automatically configuring a default route
4600 onto the network device.
4602 * systemd-networkd's bridging support gained two new options ProxyARP=
4603 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
4604 MulticastRouter= for configuring multicast routing behaviour. A new
4605 option MulticastIGMPVersion= may be used to change bridge's multicast
4606 Internet Group Management Protocol (IGMP) version.
4608 * systemd-networkd's FooOverUDP support gained the ability to configure
4609 local and peer IP addresses via Local= and Peer=. A new option
4610 PeerPort= may be used to configure the peer's IP port.
4612 * systemd-networkd's TUN support gained a new setting VnetHeader= for
4613 tweaking Generic Segment Offload support.
4615 * The address family for policy rules may be specified using the new
4616 Family= option in the [RoutingPolicyRule] section.
4618 * networkctl gained a new "delete" command for removing virtual network
4619 devices, as well as a new "--stats" switch for showing device
4622 * networkd.conf gained a new setting SpeedMeter= and
4623 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
4624 measured speed may be shown by 'networkctl status'.
4626 * "networkctl status" now displays MTU and queue lengths, and more
4627 detailed information about VXLAN and bridge devices.
4629 * systemd-networkd's .network and .link files gained a new Property=
4630 setting in the [Match] section, to match against devices with
4631 specific udev properties.
4633 * systemd-networkd's tunnel support gained a new option
4634 AssignToLoopback= for selecting whether to use the loopback device
4635 "lo" as underlying device.
4637 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
4638 been renamed to LinkLayerAddress=, and it now allows configuration of
4641 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
4642 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
4643 IPv6 configuration (static or DHCPv6) was found for a given
4644 interface. It will not touch the sysctl otherwise.
4646 * The order of entries is $PATH used by the user manager instance was
4647 changed to put bin/ entries before the corresponding sbin/ entries.
4648 It is recommended to not rely on this order, and only ever have one
4649 binary with a given name in the system paths under /usr.
4651 * A new tool systemd-network-generator has been added that may generate
4652 .network, .netdev and .link files from IP configuration specified on
4653 the kernel command line in the format used by Dracut.
4655 * The CriticalConnection= setting in .network files is now deprecated,
4656 and replaced by a new KeepConfiguration= setting which allows more
4657 detailed configuration of the IP configuration to keep in place.
4659 * systemd-analyze gained a few new verbs:
4661 - "systemd-analyze timestamp" parses and converts timestamps. This is
4662 similar to the existing "systemd-analyze calendar" command which
4663 does the same for recurring calendar events.
4665 - "systemd-analyze timespan" parses and converts timespans (i.e.
4666 durations as opposed to points in time).
4668 - "systemd-analyze condition" will parse and test ConditionXYZ=
4671 - "systemd-analyze exit-status" will parse and convert exit status
4672 codes to their names and back.
4674 - "systemd-analyze unit-files" will print a list of all unit
4675 file paths and unit aliases.
4677 * SuccessExitStatus=, RestartPreventExitStatus=, and
4678 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
4679 is equivalent to "65"). Those exit status name mappings may be
4680 displayed with the systemd-analyze exit-status verb describe above.
4682 * systemd-logind now exposes a per-session SetBrightness() bus call,
4683 which may be used to securely change the brightness of a kernel
4684 brightness device, if it belongs to the session's seat. By using this
4685 call unprivileged clients can make changes to "backlight" and "leds"
4686 devices securely with strict requirements on session membership.
4687 Desktop environments may use this to generically make brightness
4688 changes to such devices without shipping private SUID binaries or
4689 udev rules for that purpose.
4691 * "udevadm info" gained a --wait-for-initialization switch to wait for
4692 a device to be initialized.
4694 * systemd-hibernate-resume-generator will now look for resumeflags= on
4695 the kernel command line, which is similar to rootflags= and may be
4696 used to configure device timeout for the hibernation device.
4698 * sd-event learnt a new API call sd_event_source_disable_unref() for
4699 disabling and unref'ing an event source in a single function. A
4700 related call sd_event_source_disable_unrefp() has been added for use
4701 with gcc's cleanup extension.
4703 * The sd-id128.h public API gained a new definition
4704 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
4707 * "busctl introspect" gained a new switch --xml-interface for dumping
4708 XML introspection data unmodified.
4710 * PID 1 may now show the unit name instead of the unit description
4711 string in its status output during boot. This may be configured in
4712 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
4713 kernel command line option systemd.status_unit_format=.
4715 * PID 1 now understands a new option KExecWatchdogSec= in
4716 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
4717 Previously watchdog functionality was only available for regular
4718 reboots. The new setting defaults to off, because we don't know in
4719 the general case if the watchdog will be reset after kexec (some
4720 drivers do reset it, but not all), and the new userspace might not be
4721 configured to handle the watchdog.
4723 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
4724 RebootWatchdogSec= to more clearly communicate what it is about. The
4725 old name is still accepted for compatibility.
4727 * The systemd.debug_shell kernel command line option now optionally
4728 takes a tty name to spawn the debug shell on, which allows a
4729 different tty to be selected than the built-in default.
4731 * Service units gained a new ExecCondition= setting which will run
4732 before ExecStartPre= and either continue execution of the unit (for
4733 clean exit codes), stop execution without marking the unit failed
4734 (for exit codes 1 through 254), or stop execution and fail the unit
4735 (for exit code 255 or abnormal termination).
4737 * A new service systemd-pstore.service has been added that pulls data
4738 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
4741 * timedatectl gained new verbs for configuring per-interface NTP
4742 service configuration for systemd-timesyncd.
4744 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
4745 2019. (You can set non-UTF-8 locales though, if you know their name.)
4747 * If variable assignments in sysctl.d/ files are prefixed with "-" any
4748 failures to apply them are now ignored.
4750 * systemd-random-seed.service now optionally credits entropy when
4751 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
4752 true for the service to enable this behaviour, but please consult the
4753 documentation first, since this comes with a couple of caveats.
4755 * systemd-random-seed.service is now a synchronization point for full
4756 initialization of the kernel's entropy pool. Services that require
4757 /dev/urandom to be correctly initialized should be ordered after this
4760 * The systemd-boot boot loader has been updated to optionally maintain
4761 a random seed file in the EFI System Partition (ESP). During the boot
4762 phase, this random seed is read and updated with a new seed
4763 cryptographically derived from it. Another derived seed is passed to
4764 the OS. The latter seed is then credited to the kernel's entropy pool
4765 very early during userspace initialization (from PID 1). This allows
4766 systems to boot up with a fully initialized kernel entropy pool from
4767 earliest boot on, and thus entirely removes all entropy pool
4768 initialization delays from systems using systemd-boot. Special care
4769 is taken to ensure different seeds are derived on system images
4770 replicated to multiple systems. "bootctl status" will show whether
4771 a seed was received from the boot loader.
4773 * bootctl gained two new verbs:
4775 - "bootctl random-seed" will generate the file in ESP and an EFI
4776 variable to allow a random seed to be passed to the OS as described
4779 - "bootctl is-installed" checks whether systemd-boot is currently
4782 * bootctl will warn if it detects that boot entries are misconfigured
4783 (for example if the kernel image was removed without purging the
4786 * A new document has been added describing systemd's use and support
4787 for the kernel's entropy pool subsystem:
4789 https://systemd.io/RANDOM_SEEDS
4791 * When the system is hibernated the swap device to write the
4792 hibernation image to is now automatically picked from all available
4793 swap devices, preferring the swap device with the highest configured
4794 priority over all others, and picking the device with the most free
4795 space if there are multiple devices with the highest priority.
4797 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
4798 option that permits selecting the timeout how long to wait for a
4799 device with an encryption key before asking for the password.
4801 * IOWeight= has learnt to properly set the IO weight when using the
4802 BFQ scheduler officially found in kernels 5.0+.
4804 * A new mailing list has been created for reporting of security issues:
4805 systemd-security@redhat.com. For mode details, see
4806 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
4808 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
4809 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
4810 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
4811 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
4812 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
4813 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
4814 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
4815 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
4816 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
4817 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
4818 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
4819 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
4820 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
4821 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
4822 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
4823 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
4824 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
4825 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
4826 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
4827 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
4828 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
4829 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
4830 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
4831 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
4832 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
4833 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
4834 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
4835 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
4836 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
4837 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
4839 – Camerino, 2019-09-03
4843 * In .link files, MACAddressPolicy=persistent (the default) is changed
4844 to cover more devices. For devices like bridges, tun, tap, bond, and
4845 similar interfaces that do not have other identifying information,
4846 the interface name is used as the basis for persistent seed for MAC
4847 and IPv4LL addresses. The way that devices that were handled
4848 previously is not changed, and this change is about covering more
4849 devices then previously by the "persistent" policy.
4851 MACAddressPolicy=random may be used to force randomized MACs and
4852 IPv4LL addresses for a device if desired.
4854 Hint: the log output from udev (at debug level) was enhanced to
4855 clarify what policy is followed and which attributes are used.
4856 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
4857 may be used to view this.
4859 Hint: if a bridge interface is created without any slaves, and gains
4860 a slave later, then now the bridge does not inherit slave's MAC.
4861 To inherit slave's MAC, for example, create the following file:
4863 # /etc/systemd/network/98-bridge-inherit-mac.link
4868 MACAddressPolicy=none
4871 * The .device units generated by systemd-fstab-generator and other
4872 generators do not automatically pull in the corresponding .mount unit
4873 as a Wants= dependency. This means that simply plugging in the device
4874 will not cause the mount unit to be started automatically. But please
4875 note that the mount unit may be started for other reasons, in
4876 particular if it is part of local-fs.target, and any unit which
4877 (transitively) depends on local-fs.target is started.
4879 * networkctl list/status/lldp now accept globbing wildcards for network
4880 interface names to match against all existing interfaces.
4882 * The $PIDFILE environment variable is set to point the absolute path
4883 configured with PIDFile= for processes of that service.
4885 * The fallback DNS server list was augmented with Cloudflare public DNS
4886 servers. Use `-Ddns-servers=` to set a different fallback.
4888 * A new special target usb-gadget.target will be started automatically
4889 when a USB Device Controller is detected (which means that the system
4890 is a USB peripheral).
4892 * A new unit setting CPUQuotaPeriodSec= assigns the time period
4893 relatively to which the CPU time quota specified by CPUQuota= is
4896 * A new unit setting ProtectHostname= may be used to prevent services
4897 from modifying hostname information (even if they otherwise would
4898 have privileges to do so).
4900 * A new unit setting NetworkNamespacePath= may be used to specify a
4901 namespace for service or socket units through a path referring to a
4902 Linux network namespace pseudo-file.
4904 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
4905 have an effect on .socket units: when used the listening socket is
4906 created within the configured network namespace instead of the host
4909 * ExecStart= command lines in unit files may now be prefixed with ':'
4910 in which case environment variable substitution is
4911 disabled. (Supported for the other ExecXYZ= settings, too.)
4913 * .timer units gained two new boolean settings OnClockChange= and
4914 OnTimezoneChange= which may be used to also trigger a unit when the
4915 system clock is changed or the local timezone is
4916 modified. systemd-run has been updated to make these options easily
4917 accessible from the command line for transient timers.
4919 * Two new conditions for units have been added: ConditionMemory= may be
4920 used to conditionalize a unit based on installed system
4921 RAM. ConditionCPUs= may be used to conditionalize a unit based on
4922 installed CPU cores.
4924 * The @default system call filter group understood by SystemCallFilter=
4925 has been updated to include the new rseq() system call introduced in
4928 * A new time-set.target has been added that indicates that the system
4929 time has been set from a local source (possibly imprecise). The
4930 existing time-sync.target is stronger and indicates that the time has
4931 been synchronized with a precise external source. Services where
4932 approximate time is sufficient should use the new target.
4934 * "systemctl start" (and related commands) learnt a new
4935 --show-transaction option. If specified brief information about all
4936 jobs queued because of the requested operation is shown.
4938 * systemd-networkd recognizes a new operation state 'enslaved', used
4939 (instead of 'degraded' or 'carrier') for interfaces which form a
4940 bridge, bond, or similar, and an new 'degraded-carrier' operational
4941 state used for the bond or bridge master interface when one of the
4942 enslaved devices is not operational.
4944 * .network files learnt the new IgnoreCarrierLoss= option for leaving
4945 networks configured even if the carrier is lost.
4947 * The RequiredForOnline= setting in .network files may now specify a
4948 minimum operational state required for the interface to be considered
4949 "online" by systemd-networkd-wait-online. Related to this
4950 systemd-networkd-wait-online gained a new option --operational-state=
4951 to configure the same, and its --interface= option was updated to
4952 optionally also take an operational state specific for an interface.
4954 * systemd-networkd-wait-online gained a new setting --any for waiting
4955 for only one of the requested interfaces instead of all of them.
4957 * systemd-networkd now implements L2TP tunnels.
4959 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
4960 may be used to cause autonomous and onlink prefixes received in IPv6
4961 Router Advertisements to be ignored.
4963 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
4964 file settings may be used to tweak bridge behaviour.
4966 * The new TripleSampling= option in .network files may be used to
4967 configure CAN triple sampling.
4969 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
4970 used to point to private or preshared key for a WireGuard interface.
4972 * /etc/crypttab now supports the same-cpu-crypt and
4973 submit-from-crypt-cpus options to tweak encryption work scheduling
4976 * systemd-tmpfiles will now take a BSD file lock before operating on a
4977 contents of directory. This may be used to temporarily exclude
4978 directories from aging by taking the same lock (useful for example
4979 when extracting a tarball into /tmp or /var/tmp as a privileged user,
4980 which might create files with really old timestamps, which
4981 nevertheless should not be deleted). For further details, see:
4983 https://systemd.io/TEMPORARY_DIRECTORIES
4985 * systemd-tmpfiles' h line type gained support for the
4986 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
4987 controlling project quota inheritance.
4989 * sd-boot and bootctl now implement support for an Extended Boot Loader
4990 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
4991 addition to the ESP partition mounted to /efi or /boot/efi.
4992 Configuration file fragments, kernels, initrds and other EFI images
4993 to boot will be loaded from both the ESP and XBOOTLDR partitions.
4994 The XBOOTLDR partition was previously described by the Boot Loader
4995 Specification, but implementation was missing in sd-boot. Support for
4996 this concept allows using the sd-boot boot loader in more
4997 conservative scenarios where the boot loader itself is placed in the
4998 ESP but the kernels to boot (and their metadata) in a separate
5001 * A system may now be booted with systemd.volatile=overlay on the
5002 kernel command line, which causes the root file system to be set up
5003 an overlayfs mount combining the root-only root directory with a
5004 writable tmpfs. In this setup, the underlying root device is not
5005 modified, and any changes are lost at reboot.
5007 * Similar, systemd-nspawn can now boot containers with a volatile
5008 overlayfs root with the new --volatile=overlay switch.
5010 * systemd-nspawn can now consume OCI runtime bundles using a new
5011 --oci-bundle= option. This implementation is fully usable, with most
5012 features in the specification implemented, but since this a lot of
5013 new code and functionality, this feature should most likely not
5014 be used in production yet.
5016 * systemd-nspawn now supports various options described by the OCI
5017 runtime specification on the command-line and in .nspawn files:
5018 --inaccessible=/Inaccessible= may be used to mask parts of the file
5019 system tree, --console=/--pipe may be used to configure how standard
5020 input, output, and error are set up.
5022 * busctl learned the `emit` verb to generate D-Bus signals.
5024 * systemd-analyze cat-config may be used to gather and display
5025 configuration spread over multiple files, for example system and user
5026 presets, tmpfiles.d, sysusers.d, udev rules, etc.
5028 * systemd-analyze calendar now takes an optional new parameter
5029 --iterations= which may be used to show a maximum number of iterations
5030 the specified expression will elapse next.
5032 * The sd-bus C API gained support for naming method parameters in the
5035 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
5036 the reboot() system call expects.
5038 * journalctl learnt a new --cursor-file= option that points to a file
5039 from which a cursor should be loaded in the beginning and to which
5040 the updated cursor should be stored at the end.
5042 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
5043 detected by systemd-detect-virt (and may also be used in
5044 ConditionVirtualization=).
5046 * The behaviour of systemd-logind may now be modified with environment
5047 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
5048 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
5049 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
5050 skip the relevant operation completely (when set to false), or to
5051 create a flag file in /run/systemd (when set to true), instead of
5052 actually commencing the real operation when requested. The presence
5053 of /run/systemd/reboot-to-firmware-setup,
5054 /run/systemd/reboot-to-boot-loader-menu, and
5055 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
5056 boot loader implementations to replace some steps logind performs
5057 during reboot with their own operations.
5059 * systemctl can be used to request a reboot into the boot loader menu
5060 or a specific boot loader entry with the new --boot-load-menu= and
5061 --boot-loader-entry= options to a reboot command. (This requires a
5062 boot loader that supports this, for example sd-boot.)
5064 * kernel-install will no longer unconditionally create the output
5065 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
5066 snippets, but will do only if the machine-specific parent directory
5067 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
5068 to create this parent directory during sd-boot installation.
5070 This makes it easier to use kernel-install with plugins which support
5071 a different layout of the bootloader partitions (for example grub2).
5073 * During package installation (with `ninja install`), we would create
5074 symlinks for getty@tty1.service, systemd-networkd.service,
5075 systemd-networkd.socket, systemd-resolved.service,
5076 remote-cryptsetup.target, remote-fs.target,
5077 systemd-networkd-wait-online.service, and systemd-timesyncd.service
5078 in /etc, as if `systemctl enable` was called for those units, to make
5079 the system usable immediately after installation. Now this is not
5080 done anymore, and instead calling `systemctl preset-all` is
5081 recommended after the first installation of systemd.
5083 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
5084 is built on seccomp. When turned on creation of SUID/SGID files is
5087 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
5088 implied if DynamicUser= is turned on for a service. This hardens
5089 these services, so that they neither can benefit from nor create
5090 SUID/SGID executables. This is a minor compatibility breakage, given
5091 that when DynamicUser= was first introduced SUID/SGID behaviour was
5092 unaffected. However, the security benefit of these two options is
5093 substantial, and the setting is still relatively new, hence we opted
5094 to make it mandatory for services with dynamic users.
5096 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
5097 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
5098 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
5099 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
5100 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
5101 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
5102 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
5103 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
5104 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
5105 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
5106 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
5107 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
5108 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
5109 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
5110 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
5111 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
5112 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
5113 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5115 — Warsaw, 2019-04-11
5119 * The default locale can now be configured at compile time. Otherwise,
5120 a suitable default will be selected automatically (one of C.UTF-8,
5121 en_US.UTF-8, and C).
5123 * The version string shown by systemd and other tools now includes the
5124 git commit hash when built from git. An override may be specified
5125 during compilation, which is intended to be used by distributions to
5126 include the package release information.
5128 * systemd-cat can now filter standard input and standard error streams
5129 for different syslog priorities using the new --stderr-priority=
5132 * systemd-journald and systemd-journal-remote reject entries which
5133 contain too many fields (CVE-2018-16865) and set limits on the
5134 process' command line length (CVE-2018-16864).
5136 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
5139 * A new network device NamePolicy "keep" is implemented for link files,
5140 and used by default in 99-default.link (the fallback configuration
5141 provided by systemd). With this policy, if the network device name
5142 was already set by userspace, the device will not be renamed again.
5143 This matches the naming scheme that was implemented before
5144 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
5145 is also enabled by default, even if not specified. Effectively, this
5146 means that if naming-scheme >= 240 is specified, network devices will
5147 be renamed according to the configuration, even if they have been
5148 renamed already, if "keep" is not specified as the naming policy in
5149 the .link file. The 99-default.link file provided by systemd includes
5150 "keep" for backwards compatibility, but it is recommended for user
5151 installed .link files to *not* include it.
5153 The "kernel" policy, which keeps kernel names declared to be
5154 "persistent", now works again as documented.
5156 * kernel-install script now optionally takes the paths to one or more
5157 initrd files, and passes them to all plugins.
5159 * The mincore() system call has been dropped from the @system-service
5160 system call filter group, as it is pretty exotic and may potentially
5161 used for side-channel attacks.
5163 * -fPIE is dropped from compiler and linker options. Please specify
5164 -Db_pie=true option to meson to build position-independent
5165 executables. Note that the meson option is supported since meson-0.49.
5167 * The fs.protected_regular and fs.protected_fifos sysctls, which were
5168 added in Linux 4.19 to make some data spoofing attacks harder, are
5169 now enabled by default. While this will hopefully improve the
5170 security of most installations, it is technically a backwards
5171 incompatible change; to disable these sysctls again, place the
5172 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
5174 fs.protected_regular = 0
5175 fs.protected_fifos = 0
5177 Note that the similar hardlink and symlink protection has been
5178 enabled since v199, and may be disabled likewise.
5180 * The files read from the EnvironmentFile= setting in unit files now
5181 parse backslashes inside quotes literally, matching the behaviour of
5184 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
5185 now automatically become NOPs when run in a chroot() environment.
5187 * The tmpfiles.d/ "C" line type will now copy directory trees not only
5188 when the destination is so far missing, but also if it already exists
5189 as a directory and is empty. This is useful to cater for systems
5190 where directory trees are put together from multiple separate mount
5191 points but otherwise empty.
5193 * A new function sd_bus_close_unref() (and the associated
5194 sd_bus_close_unrefp()) has been added to libsystemd, that combines
5195 sd_bus_close() and sd_bus_unref() in one.
5197 * udevadm control learnt a new option for --ping for testing whether a
5198 systemd-udevd instance is running and reacting.
5200 * udevadm trigger learnt a new option for --wait-daemon for waiting
5201 systemd-udevd daemon to be initialized.
5203 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
5204 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
5205 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
5206 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
5207 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
5208 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
5209 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
5210 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
5211 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
5212 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
5213 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
5214 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
5215 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
5216 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
5217 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
5218 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
5219 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
5221 — Berlin, 2019-02-14
5225 * NoNewPrivileges=yes has been set for all long-running services
5226 implemented by systemd. Previously, this was problematic due to
5227 SELinux (as this would also prohibit the transition from PID1's label
5228 to the service's label). This restriction has since been lifted, but
5229 an SELinux policy update is required.
5230 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
5232 * DynamicUser=yes is dropped from systemd-networkd.service,
5233 systemd-resolved.service and systemd-timesyncd.service, which was
5234 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
5235 and since v236 for systemd-timesyncd.service. The users and groups
5236 systemd-network, systemd-resolve and systemd-timesync are created
5237 by systemd-sysusers again. Distributors or system administrators
5238 may need to create these users and groups if they not exist (or need
5239 to re-enable DynamicUser= for those units) while upgrading systemd.
5240 Also, the clock file for systemd-timesyncd may need to move from
5241 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
5243 * When unit files are loaded from disk, previously systemd would
5244 sometimes (depending on the unit loading order) load units from the
5245 target path of symlinks in .wants/ or .requires/ directories of other
5246 units. This meant that unit could be loaded from different paths
5247 depending on whether the unit was requested explicitly or as a
5248 dependency of another unit, not honouring the priority of directories
5249 in search path. It also meant that it was possible to successfully
5250 load and start units which are not found in the unit search path, as
5251 long as they were requested as a dependency and linked to from
5252 .wants/ or .requires/. The target paths of those symlinks are not
5253 used for loading units anymore and the unit file must be found in
5256 * A new service type has been added: Type=exec. It's very similar to
5257 Type=simple but ensures the service manager will wait for both fork()
5258 and execve() of the main service binary to complete before proceeding
5259 with follow-up units. This is primarily useful so that the manager
5260 propagates any errors in the preparation phase of service execution
5261 back to the job that requested the unit to be started. For example,
5262 consider a service that has ExecStart= set to a file system binary
5263 that doesn't exist. With Type=simple starting the unit would be
5264 considered instantly successful, as only fork() has to complete
5265 successfully and the manager does not wait for execve(), and hence
5266 its failure is seen "too late". With the new Type=exec service type
5267 starting the unit will fail, as the manager will wait for the
5268 execve() and notice its failure, which is then propagated back to the
5271 NOTE: with the next release 241 of systemd we intend to change the
5272 systemd-run tool to default to Type=exec for transient services
5273 started by it. This should be mostly safe, but in specific corner
5274 cases might result in problems, as the systemd-run tool will then
5275 block on NSS calls (such as user name look-ups due to User=) done
5276 between the fork() and execve(), which under specific circumstances
5277 might cause problems. It is recommended to specify "-p Type=simple"
5278 explicitly in the few cases where this applies. For regular,
5279 non-transient services (i.e. those defined with unit files on disk)
5280 we will continue to default to Type=simple.
5282 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
5283 userspace processes is set to 1024 (soft) and 4096
5284 (hard). Previously, systemd passed this on unmodified to all
5285 processes it forked off. With this systemd release the hard limit
5286 systemd passes on is increased to 512K, overriding the kernel's
5287 defaults and substantially increasing the number of simultaneous file
5288 descriptors unprivileged userspace processes can allocate. Note that
5289 the soft limit remains at 1024 for compatibility reasons: the
5290 traditional UNIX select() call cannot deal with file descriptors >=
5291 1024 and increasing the soft limit globally might thus result in
5292 programs unexpectedly allocating a high file descriptor and thus
5293 failing abnormally when attempting to use it with select() (of
5294 course, programs shouldn't use select() anymore, and prefer
5295 poll()/epoll, but the call unfortunately remains undeservedly popular
5296 at this time). This change reflects the fact that file descriptor
5297 handling in the Linux kernel has been optimized in more recent
5298 kernels and allocating large numbers of them should be much cheaper
5299 both in memory and in performance than it used to be. Programs that
5300 want to take benefit of the increased limit have to "opt-in" into
5301 high file descriptors explicitly by raising their soft limit. Of
5302 course, when they do that they must acknowledge that they cannot use
5303 select() anymore (and neither can any shared library they use — or
5304 any shared library used by any shared library they use and so on).
5305 Which default hard limit is most appropriate is of course hard to
5306 decide. However, given reports that ~300K file descriptors are used
5307 in real-life applications we believe 512K is sufficiently high as new
5308 default for now. Note that there are also reports that using very
5309 high hard limits (e.g. 1G) is problematic: some software allocates
5310 large arrays with one element for each potential file descriptor
5311 (Java, …) — a high hard limit thus triggers excessively large memory
5312 allocations in these applications. Hopefully, the new default of 512K
5313 is a good middle ground: higher than what real-life applications
5314 currently need, and low enough for avoid triggering excessively large
5315 allocations in problematic software. (And yes, somebody should fix
5318 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
5319 to the highest possible values, as separate accounting of file
5320 descriptors is no longer necessary, as memcg tracks them correctly as
5321 part of the memory accounting anyway. Thus, from the four limits on
5322 file descriptors currently enforced (fs.file-max, fs.nr_open,
5323 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
5324 and keep only the latter two. A set of build-time options
5325 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
5326 has been added to revert this change in behaviour, which might be
5327 an option for systems that turn off memcg in the kernel.
5329 * When no /etc/locale.conf file exists (and hence no locale settings
5330 are in place), systemd will now use the "C.UTF-8" locale by default,
5331 and set LANG= to it. This locale is supported by various
5332 distributions including Fedora, with clear indications that upstream
5333 glibc is going to make it available too. This locale enables UTF-8
5334 mode by default, which appears appropriate for 2018.
5336 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
5337 default. This effectively switches the RFC3704 Reverse Path filtering
5338 from Strict mode to Loose mode. This is more appropriate for hosts
5339 that have multiple links with routes to the same networks (e.g.
5340 a client with a Wi-Fi and Ethernet both connected to the internet).
5342 Consult the kernel documentation for details on this sysctl:
5343 https://docs.kernel.org/networking/ip-sysctl.html
5345 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
5348 * CPUAccounting=yes no longer enables the CPU controller when using
5349 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
5350 statistics are now provided independently from the CPU controller.
5352 * Support for disabling a particular cgroup controller within a sub-tree
5353 has been added through the DisableControllers= directive.
5355 * cgroup_no_v1=all on the kernel command line now also implies
5356 using the unified cgroup hierarchy, unless one explicitly passes
5357 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
5359 * The new "MemoryMin=" unit file property may now be used to set the
5360 memory usage protection limit of processes invoked by the unit. This
5361 controls the cgroup v2 memory.min attribute. Similarly, the new
5362 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
5363 cgroup v2 io.latency cgroup property for configuring per-service I/O
5366 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
5367 to the cgroup v1 "devices" cgroup controller.
5369 * systemd-escape now is able to combine --unescape with --template. It
5370 also learnt a new option --instance for extracting and unescaping the
5371 instance part of a unit name.
5373 * sd-bus now provides the sd_bus_message_readv() which is similar to
5374 sd_bus_message_read() but takes a va_list object. The pair
5375 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
5376 has been added for configuring the default method call timeout to
5377 use. sd_bus_error_move() may be used to efficiently move the contents
5378 from one sd_bus_error structure to another, invalidating the
5379 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
5380 be used to control whether a bus connection object is automatically
5381 flushed when an sd-event loop is exited.
5383 * When processing classic BSD syslog log messages, journald will now
5384 save the original time-stamp string supplied in the new
5385 SYSLOG_TIMESTAMP= journal field. This permits consumers to
5386 reconstruct the original BSD syslog message more correctly.
5388 * StandardOutput=/StandardError= in service files gained support for
5389 new "append:…" parameters, for connecting STDOUT/STDERR of a service
5390 to a file, and appending to it.
5392 * The signal to use as last step of killing of unit processes is now
5393 configurable. Previously it was hard-coded to SIGKILL, which may now
5394 be overridden with the new KillSignal= setting. Note that this is the
5395 signal used when regular termination (i.e. SIGTERM) does not suffice.
5396 Similarly, the signal used when aborting a program in case of a
5397 watchdog timeout may now be configured too (WatchdogSignal=).
5399 * The XDG_SESSION_DESKTOP environment variable may now be configured in
5400 the pam_systemd argument line, using the new desktop= switch. This is
5401 useful to initialize it properly from a display manager without
5402 having to touch C code.
5404 * Most configuration options that previously accepted percentage values
5405 now also accept permille values with the '‰' suffix (instead of '%').
5407 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
5410 * systemd-resolved's configuration file resolved.conf gained a new
5411 option ReadEtcHosts= which may be used to turn off processing and
5412 honoring /etc/hosts entries.
5414 * The "--wait" switch may now be passed to "systemctl
5415 is-system-running", in which case the tool will synchronously wait
5416 until the system finished start-up.
5418 * hostnamed gained a new bus call to determine the DMI product UUID.
5420 * On x86-64 systemd will now prefer using the RDRAND processor
5421 instruction over /dev/urandom whenever it requires randomness that
5422 neither has to be crypto-grade nor should be reproducible. This
5423 should substantially reduce the amount of entropy systemd requests
5424 from the kernel during initialization on such systems, though not
5425 reduce it to zero. (Why not zero? systemd still needs to allocate
5426 UUIDs and such uniquely, which require high-quality randomness.)
5428 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
5429 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
5430 for forcing the "Other Information" bit in IPv6 RA messages. The
5431 bonding logic gained four new options AdActorSystemPriority=,
5432 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
5433 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
5434 shuffling of flows. The tunnel logic gained a new
5435 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
5436 Deployment. The policy rule logic gained four new options IPProtocol=,
5437 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
5438 support for the MulticastToUnicast= option. networkd also gained
5439 support for configuring static IPv4 ARP or IPv6 neighbor entries.
5441 * .preset files (as read by 'systemctl preset') may now be used to
5442 instantiate services.
5444 * /etc/crypttab now understands the sector-size= option to configure
5445 the sector size for an encrypted partition.
5447 * Key material for encrypted disks may now be placed on a formatted
5448 medium, and referenced from /etc/crypttab by the UUID of the file
5449 system, followed by "=" suffixed by the path to the key file.
5451 * The "collect" udev component has been removed without replacement, as
5452 it is neither used nor maintained.
5454 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
5455 LogsDirectory=, ConfigurationDirectory= settings are used in a
5456 service the executed processes will now receive a set of environment
5457 variables containing the full paths of these directories.
5458 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
5459 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
5460 are used. Note that these options may be used multiple times per
5461 service in which case the resulting paths will be concatenated and
5462 separated by colons.
5464 * Predictable interface naming has been extended to cover InfiniBand
5465 NICs. They will be exposed with an "ib" prefix.
5467 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
5468 which case the respective line failing is ignored.
5470 * .link files may now be used to configure the equivalent to the
5471 "ethtool advertise" commands.
5473 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
5474 alternative to libudev.h. Previously, the latter was just an internal
5475 wrapper around the former, but now these two APIs are exposed
5478 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
5479 which calculates an app-specific boot ID similar to how
5480 sd_id128_get_machine_app_specific() generates an app-specific machine
5483 * A new tool systemd-id128 has been added that can be used to determine
5484 and generate various 128bit IDs.
5486 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
5489 * systemd-hibernate-resume-generator will now honor the "noresume"
5490 kernel command line option, in which case it will bypass resuming
5491 from any hibernated image.
5493 * The systemd-sleep.conf configuration file gained new options
5494 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
5495 AllowHybridSleep= for prohibiting specific sleep modes even if the
5496 kernel exports them.
5498 * portablectl is now officially supported and has thus moved to
5501 * bootctl learnt the two new commands "set-default" and "set-oneshot"
5502 for setting the default boot loader item to boot to (either
5503 persistently or only for the next boot). This is currently only
5504 compatible with sd-boot, but may be implemented on other boot loaders
5505 too, that follow the boot loader interface. The updated interface is
5506 now documented here:
5508 https://systemd.io/BOOT_LOADER_INTERFACE
5510 * A new kernel command line option systemd.early_core_pattern= is now
5511 understood which may be used to influence the core_pattern PID 1
5512 installs during early boot.
5514 * busctl learnt two new options -j and --json= for outputting method
5515 call replies, properties and monitoring output in JSON.
5517 * journalctl's JSON output now supports simple ANSI coloring as well as
5518 a new "json-seq" mode for generating RFC7464 output.
5520 * Unit files now support the %g/%G specifiers that resolve to the UNIX
5521 group/GID of the service manager runs as, similar to the existing
5522 %u/%U specifiers that resolve to the UNIX user/UID.
5524 * systemd-logind learnt a new global configuration option
5525 UserStopDelaySec= that may be set in logind.conf. It specifies how
5526 long the systemd --user instance shall remain started after a user
5527 logs out. This is useful to speed up repetitive re-connections of the
5528 same user, as it means the user's service manager doesn't have to be
5529 stopped/restarted on each iteration, but can be reused between
5530 subsequent options. This setting defaults to 10s. systemd-logind also
5531 exports two new properties on its Manager D-Bus objects indicating
5532 whether the system's lid is currently closed, and whether the system
5535 * systemd gained support for a generic boot counting logic, which
5536 generically permits automatic reverting to older boot loader entries
5537 if newer updated ones don't work. The boot loader side is implemented
5538 in sd-boot, but is kept open for other boot loaders too. For details
5541 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
5543 * The SuccessAction=/FailureAction= unit file settings now learnt two
5544 new parameters: "exit" and "exit-force", which result in immediate
5545 exiting of the service manager, and are only useful in systemd --user
5546 and container environments.
5548 * Unit files gained support for a pair of options
5549 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
5550 exit status to use as service manager exit status when
5551 SuccessAction=/FailureAction= is set to exit or exit-force.
5553 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
5554 options may now be used to configure the log rate limiting applied by
5555 journald per-service.
5557 * systemd-analyze gained a new verb "timespan" for parsing and
5558 normalizing time span values (i.e. strings like "5min 7s 8us").
5560 * systemd-analyze also gained a new verb "security" for analyzing the
5561 security and sand-boxing settings of services in order to determine an
5562 "exposure level" for them, indicating whether a service would benefit
5563 from more sand-boxing options turned on for them.
5565 * "systemd-analyze syscall-filter" will now also show system calls
5566 supported by the local kernel but not included in any of the defined
5569 * .nspawn files now understand the Ephemeral= setting, matching the
5570 --ephemeral command line switch.
5572 * sd-event gained the new APIs sd_event_source_get_floating() and
5573 sd_event_source_set_floating() for controlling whether a specific
5574 event source is "floating", i.e. destroyed along with the even loop
5577 * Unit objects on D-Bus gained a new "Refs" property that lists all
5578 clients that currently have a reference on the unit (to ensure it is
5581 * The JoinControllers= option in system.conf is no longer supported, as
5582 it didn't work correctly, is hard to support properly, is legacy (as
5583 the concept only exists on cgroup v1) and apparently wasn't used.
5585 * Journal messages that are generated whenever a unit enters the failed
5586 state are now tagged with a unique MESSAGE_ID. Similarly, messages
5587 generated whenever a service process exits are now made recognizable,
5588 too. A tagged message is also emitted whenever a unit enters the
5589 "dead" state on success.
5591 * systemd-run gained a new switch --working-directory= for configuring
5592 the working directory of the service to start. A shortcut -d is
5593 equivalent, setting the working directory of the service to the
5594 current working directory of the invoking program. The new --shell
5595 (or just -S) option has been added for invoking the $SHELL of the
5596 caller as a service, and implies --pty --same-dir --wait --collect
5597 --service-type=exec. Or in other words, "systemd-run -S" is now the
5598 quickest way to quickly get an interactive in a fully clean and
5599 well-defined system service context.
5601 * machinectl gained a new verb "import-fs" for importing an OS tree
5602 from a directory. Moreover, when a directory or tarball is imported
5603 and single top-level directory found with the OS itself below the OS
5604 tree is automatically mangled and moved one level up.
5606 * systemd-importd will no longer set up an implicit btrfs loop-back
5607 file system on /var/lib/machines. If one is already set up, it will
5608 continue to be used.
5610 * A new generator "systemd-run-generator" has been added. It will
5611 synthesize a unit from one or more program command lines included in
5612 the kernel command line. This is very useful in container managers
5615 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
5617 This will run "systemd-nspawn" on an image, invoke the specified
5618 command line and immediately shut down the container again, returning
5619 the command line's exit code.
5621 * The block device locking logic is now documented:
5623 https://systemd.io/BLOCK_DEVICE_LOCKING
5625 * loginctl and machinectl now optionally output the various tables in
5626 JSON using the --output= switch. It is our intention to add similar
5627 support to systemctl and all other commands.
5629 * udevadm's query and trigger verb now optionally take a .device unit
5632 * systemd-udevd's network naming logic now understands a new
5633 net.naming-scheme= kernel command line switch, which may be used to
5634 pick a specific version of the naming scheme. This helps stabilizing
5635 interface names even as systemd/udev are updated and the naming logic
5638 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
5639 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
5640 initialize one to all 0xFF.
5642 * After loading the SELinux policy systemd will now recursively relabel
5643 all files and directories listed in
5644 /run/systemd/relabel-extra.d/*.relabel (which should be simple
5645 newline separated lists of paths) in addition to the ones it already
5646 implicitly relabels in /run, /dev and /sys. After the relabelling is
5647 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
5648 removed. This is useful to permit initrds (i.e. code running before
5649 the SELinux policy is in effect) to generate files in the host
5650 filesystem safely and ensure that the correct label is applied during
5651 the transition to the host OS.
5653 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
5654 mknod() handling in user namespaces. Previously mknod() would always
5655 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
5656 but device nodes generated that way cannot be opened, and attempts to
5657 open them result in EPERM. This breaks the "graceful fallback" logic
5658 in systemd's PrivateDevices= sand-boxing option. This option is
5659 implemented defensively, so that when systemd detects it runs in a
5660 restricted environment (such as a user namespace, or an environment
5661 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
5662 where device nodes cannot be created the effect of PrivateDevices= is
5663 bypassed (following the logic that 2nd-level sand-boxing is not
5664 essential if the system systemd runs in is itself already sand-boxed
5665 as a whole). This logic breaks with 4.18 in container managers where
5666 user namespacing is used: suddenly PrivateDevices= succeeds setting
5667 up a private /dev/ file system containing devices nodes — but when
5668 these are opened they don't work.
5670 At this point it is recommended that container managers utilizing
5671 user namespaces that intend to run systemd in the payload explicitly
5672 block mknod() with seccomp or similar, so that the graceful fallback
5675 We are very sorry for the breakage and the requirement to change
5676 container configurations for newer kernels. It's purely caused by an
5677 incompatible kernel change. The relevant kernel developers have been
5678 notified about this userspace breakage quickly, but they chose to
5681 * PermissionsStartOnly= setting is deprecated (but is still supported
5682 for backwards compatibility). The same functionality is provided by
5683 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
5686 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
5687 pam_systemd anymore.
5689 * The naming scheme for network devices was changed to always rename
5690 devices, even if they were already renamed by userspace. The "kernel"
5691 policy was changed to only apply as a fallback, if no other naming
5694 * The requirements to build systemd is bumped to meson-0.46 and
5697 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
5698 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
5699 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
5700 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
5701 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
5702 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
5703 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
5704 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
5705 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
5706 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
5707 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
5708 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
5709 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
5710 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
5711 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
5712 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
5713 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
5714 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
5715 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
5716 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
5717 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
5718 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
5719 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
5720 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
5721 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
5722 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
5723 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
5724 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
5725 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
5726 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
5727 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
5728 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
5729 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
5730 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
5731 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
5732 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
5733 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
5734 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
5735 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
5736 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
5737 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
5738 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
5739 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
5740 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
5741 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
5743 — Warsaw, 2018-12-21
5747 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
5748 builtin will name network interfaces differently than in previous
5749 versions for virtual network interfaces created with SR-IOV and NPAR
5750 and for devices where the PCI network controller device does not have
5751 a slot number associated.
5753 SR-IOV virtual devices are now named based on the name of the parent
5754 interface, with a suffix of "v<N>", where <N> is the virtual device
5755 number. Previously those virtual devices were named as if completely
5758 The ninth and later NPAR virtual devices will be named following the
5759 scheme used for the first eight NPAR partitions. Previously those
5760 devices were not renamed and the kernel default (eth<n>) was used.
5762 "net_id" will also generate names for PCI devices where the PCI
5763 network controller device does not have an associated slot number
5764 itself, but one of its parents does. Previously those devices were
5765 not renamed and the kernel default (eth<n>) was used.
5767 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
5768 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
5769 the unit. So, it is expected that the default behavior of
5770 systemd-logind is not changed. However, if distribution packagers or
5771 administrators disabled or modified IPAddressDeny= setting by a
5772 drop-in config file, then it may be necessary to update the file to
5773 re-enable AF_INET and AF_INET6 to support network user name services,
5776 * When the RestrictNamespaces= unit property is specified multiple
5777 times, then the specified types are merged now. Previously, only the
5778 last assignment was used. So, if distribution packagers or
5779 administrators modified the setting by a drop-in config file, then it
5780 may be necessary to update the file.
5782 * When OnFailure= is used in combination with Restart= on a service
5783 unit, then the specified units will no longer be triggered on
5784 failures that result in restarting. Previously, the specified units
5785 would be activated each time the unit failed, even when the unit was
5786 going to be restarted automatically. This behaviour contradicted the
5787 documentation. With this release the code is adjusted to match the
5790 * systemd-tmpfiles will now print a notice whenever it encounters
5791 tmpfiles.d/ lines referencing the /var/run/ directory. It will
5792 recommend reworking them to use the /run/ directory instead (for
5793 which /var/run/ is simply a symlinked compatibility alias). This way
5794 systemd-tmpfiles can properly detect line conflicts and merge lines
5795 referencing the same file by two paths, without having to access
5798 * systemctl disable/unmask/preset/preset-all cannot be used with
5799 --runtime. Previously this was allowed, but resulted in unintuitive
5800 behaviour that wasn't useful. systemctl disable/unmask will now undo
5801 both runtime and persistent enablement/masking, i.e. it will remove
5802 any relevant symlinks both in /run and /etc.
5804 * Note that all long-running system services shipped with systemd will
5805 now default to a system call allow list (rather than a deny list, as
5806 before). In particular, systemd-udevd will now enforce one too. For
5807 most cases this should be safe, however downstream distributions
5808 which disabled sandboxing of systemd-udevd (specifically the
5809 MountFlags= setting), might want to disable this security feature
5810 too, as the default allow-listing will prohibit all mount, swap,
5811 reboot and clock changing operations from udev rules.
5813 * sd-boot acquired new loader configuration settings to optionally turn
5814 off Windows and MacOS boot partition discovery as well as
5815 reboot-into-firmware menu items. It is also able to pick a better
5816 screen resolution for HiDPI systems, and now provides loader
5817 configuration settings to change the resolution explicitly.
5819 * systemd-resolved now supports DNS-over-TLS. It's still
5820 turned off by default, use DNSOverTLS=opportunistic to turn it on in
5821 resolved.conf. We intend to make this the default as soon as couple
5822 of additional techniques for optimizing the initial latency caused by
5823 establishing a TLS/TCP connection are implemented.
5825 * systemd-resolved.service and systemd-networkd.service now set
5826 DynamicUser=yes. The users systemd-resolve and systemd-network are
5827 not created by systemd-sysusers anymore.
5829 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
5830 that embed a network facing module into any process using getpwuid()
5831 or related call: the dynamic allocation of the user ID for
5832 systemd-resolved.service means the service manager has to check NSS
5833 if the user name is already taken when forking off the service. Since
5834 the user in the common case won't be defined in /etc/passwd the
5835 lookup is likely to trigger nss-ldap which in turn might use NSS to
5836 ask systemd-resolved for hostname lookups. This will hence result in
5837 a deadlock: a user name lookup in order to start
5838 systemd-resolved.service will result in a hostname lookup for which
5839 systemd-resolved.service needs to be started already. There are
5840 multiple ways to work around this problem: pre-allocate the
5841 "systemd-resolve" user on such systems, so that nss-ldap won't be
5842 triggered; or use a different NSS package that doesn't do networking
5843 in-process but provides a local asynchronous name cache; or configure
5844 the NSS package to avoid lookups for UIDs in the range `pkg-config
5845 systemd --variable=dynamicuidmin` … `pkg-config systemd
5846 --variable=dynamicuidmax`, so that it does not consider itself
5847 authoritative for the same UID range systemd allocates dynamic users
5850 * The systemd-resolve tool has been renamed to resolvectl (it also
5851 remains available under the old name, for compatibility), and its
5852 interface is now verb-based, similar in style to the other <xyz>ctl
5853 tools, such as systemctl or loginctl.
5855 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
5856 compatibility. It may be symlinked under the 'resolvconf' name, in
5857 which case it will take arguments and input compatible with the
5858 Debian and FreeBSD resolvconf tool.
5860 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
5861 where the system initially suspends, and after a timeout resumes and
5864 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
5865 set the client will only send a DUID as client identifier.
5867 * The nss-systemd glibc NSS module will now enumerate dynamic users and
5868 groups in effect. Previously, it could resolve UIDs/GIDs to user
5869 names/groups and vice versa, but did not support enumeration.
5871 * journald's Compress= configuration setting now optionally accepts a
5872 byte threshold value. All journal objects larger than this threshold
5873 will be compressed, smaller ones will not. Previously this threshold
5874 was not configurable and set to 512.
5876 * A new system.conf setting NoNewPrivileges= is now available which may
5877 be used to turn off acquisition of new privileges system-wide
5878 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
5879 for all its children). Note that turning this option on means setuid
5880 binaries and file system capabilities lose their special powers.
5881 While turning on this option is a big step towards a more secure
5882 system, doing so is likely to break numerous pre-existing UNIX tools,
5883 in particular su and sudo.
5885 * A new service systemd-time-sync-wait.service has been added. If
5886 enabled it will delay the time-sync.target unit at boot until time
5887 synchronization has been received from the network. This
5888 functionality is useful on systems lacking a local RTC or where it is
5889 acceptable that the boot process shall be delayed by external network
5892 * When hibernating, systemd will now inform the kernel of the image
5893 write offset, on kernels new enough to support this. This means swap
5894 files should work for hibernation now.
5896 * When loading unit files, systemd will now look for drop-in unit files
5897 extensions in additional places. Previously, for a unit file name
5898 "foo-bar-baz.service" it would look for dropin files in
5899 "foo-bar-baz.service.d/*.conf". Now, it will also look in
5900 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
5901 service name truncated after all inner dashes. This scheme allows
5902 writing drop-ins easily that apply to a whole set of unit files at
5903 once. It's particularly useful for mount and slice units (as their
5904 naming is prefix based), but is also useful for service and other
5905 units, for packages that install multiple unit files at once,
5906 following a strict naming regime of beginning the unit file name with
5907 the package's name. Two new specifiers are now supported in unit
5908 files to match this: %j and %J are replaced by the part of the unit
5909 name following the last dash.
5911 * Unit files and other configuration files that support specifier
5912 expansion now understand another three new specifiers: %T and %V will
5913 resolve to /tmp and /var/tmp respectively, or whatever temporary
5914 directory has been set for the calling user. %E will expand to either
5915 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
5917 * The ExecStart= lines of unit files are no longer required to
5918 reference absolute paths. If non-absolute paths are specified the
5919 specified binary name is searched within the service manager's
5920 built-in $PATH, which may be queried with 'systemd-path
5921 search-binaries-default'. It's generally recommended to continue to
5922 use absolute paths for all binaries specified in unit files.
5924 * Units gained a new load state "bad-setting", which is used when a
5925 unit file was loaded, but contained fatal errors which prevent it
5926 from being started (for example, a service unit has been defined
5927 lacking both ExecStart= and ExecStop= lines).
5929 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
5930 support alternative debuggers, for example lldb. The old name
5931 continues to be available however, for compatibility reasons. Use the
5932 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
5933 to pick an alternative debugger instead of the default gdb.
5935 * systemctl and the other tools will now output escape sequences that
5936 generate proper clickable hyperlinks in various terminal emulators
5937 where useful (for example, in the "systemctl status" output you can
5938 now click on the unit file name to quickly open it in the
5939 editor/viewer of your choice). Note that not all terminal emulators
5940 support this functionality yet, but many do. Unfortunately, the
5941 "less" pager doesn't support this yet, hence this functionality is
5942 currently automatically turned off when a pager is started (which
5943 happens quite often due to auto-paging). We hope to remove this
5944 limitation as soon as "less" learns these escape sequences. This new
5945 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
5946 environment variable. For details on these escape sequences see:
5947 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
5949 * networkd's .network files now support a new IPv6MTUBytes= option for
5950 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
5951 option in the [Route] section to configure the MTU to use for
5952 specific routes. It also gained support for configuration of the DHCP
5953 "UserClass" option through the new UserClass= setting. It gained
5954 three new options in the new [CAN] section for configuring CAN
5955 networks. The MULTICAST and ALLMULTI interface flags may now be
5956 controlled explicitly with the new Multicast= and AllMulticast=
5959 * networkd will now automatically make use of the kernel's route
5960 expiration feature, if it is available.
5962 * udevd's .link files now support setting the number of receive and
5963 transmit channels, using the RxChannels=, TxChannels=,
5964 OtherChannels=, CombinedChannels= settings.
5966 * Support for UDPSegmentationOffload= has been removed, given its
5967 limited support in hardware, and waning software support.
5969 * networkd's .netdev files now support creating "netdevsim" interfaces.
5971 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
5972 to query the unit belonging to a specific kernel control group.
5974 * systemd-analyze gained a new verb "cat-config", which may be used to
5975 dump the contents of any configuration file, with all its matching
5976 drop-in files added in, and honouring the usual search and masking
5977 logic applied to systemd configuration files. For example use
5978 "systemd-analyze cat-config systemd/system.conf" to get the complete
5979 system configuration file of systemd how it would be loaded by PID 1
5980 itself. Similar to this, various tools such as systemd-tmpfiles or
5981 systemd-sysusers, gained a new option "--cat-config", which does the
5982 corresponding operation for their own configuration settings. For
5983 example, "systemd-tmpfiles --cat-config" will now output the full
5984 list of tmpfiles.d/ lines in place.
5986 * timedatectl gained three new verbs: "show" shows bus properties of
5987 systemd-timedated, "timesync-status" shows the current NTP
5988 synchronization state of systemd-timesyncd, and "show-timesync"
5989 shows bus properties of systemd-timesyncd.
5991 * systemd-timesyncd gained a bus interface on which it exposes details
5994 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
5995 understood by systemd-timedated. It takes a colon-separated list of
5996 unit names of NTP client services. The list is used by
5997 "timedatectl set-ntp".
5999 * systemd-nspawn gained a new --rlimit= switch for setting initial
6000 resource limits for the container payload. There's a new switch
6001 --hostname= to explicitly override the container's hostname. A new
6002 --no-new-privileges= switch may be used to control the
6003 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
6004 --oom-score-adjust= switch controls the OOM scoring adjustment value
6005 for the payload. The new --cpu-affinity= switch controls the CPU
6006 affinity of the container payload. The new --resolv-conf= switch
6007 allows more detailed control of /etc/resolv.conf handling of the
6008 container. Similarly, the new --timezone= switch allows more detailed
6009 control of /etc/localtime handling of the container.
6011 * systemd-detect-virt gained a new --list switch, which will print a
6012 list of all currently known VM and container environments.
6014 * Support for "Portable Services" has been added, see
6015 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
6016 experimental, but this is expected to change soon. Reflecting this
6017 experimental state, the "portablectl" binary is not installed into
6018 /usr/bin yet. The binary has to be called with the full path
6019 /usr/lib/systemd/portablectl instead.
6021 * journalctl's and systemctl's -o switch now knows a new log output
6022 mode "with-unit". The output it generates is very similar to the
6023 regular "short" mode, but displays the unit name instead of the
6024 syslog tag for each log line. Also, the date is shown with timezone
6025 information. This mode is probably more useful than the classic
6026 "short" output mode for most purposes, except where pixel-perfect
6027 compatibility with classic /var/log/messages formatting is required.
6029 * A new --dump-bus-properties switch has been added to the systemd
6030 binary, which may be used to dump all supported D-Bus properties.
6031 (Options which are still supported, but are deprecated, are *not*
6034 * sd-bus gained a set of new calls:
6035 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
6036 enable/disable the "floating" state of a bus slot object,
6037 i.e. whether the slot object pins the bus it is allocated for into
6038 memory or if the bus slot object gets disconnected when the bus goes
6039 away. sd_bus_open_with_description(),
6040 sd_bus_open_user_with_description(),
6041 sd_bus_open_system_with_description() may be used to allocate bus
6042 objects and set their description string already during allocation.
6044 * sd-event gained support for watching inotify events from the event
6045 loop, in an efficient way, sharing inotify handles between multiple
6046 users. For this a new function sd_event_add_inotify() has been added.
6048 * sd-event and sd-bus gained support for calling special user-supplied
6049 destructor functions for userdata pointers associated with
6050 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
6051 functions sd_bus_slot_set_destroy_callback,
6052 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
6053 sd_bus_track_get_destroy_callback,
6054 sd_event_source_set_destroy_callback,
6055 sd_event_source_get_destroy_callback have been added.
6057 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
6059 * PID 1 will now automatically reschedule .timer units whenever the
6060 local timezone changes. (They previously got rescheduled
6061 automatically when the system clock changed.)
6063 * New documentation has been added to document cgroups delegation,
6064 portable services and the various code quality tools we have set up:
6066 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
6067 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
6068 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
6070 * The Boot Loader Specification has been added to the source tree.
6072 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
6074 While moving it into our source tree we have updated it and further
6075 changes are now accepted through the usual github PR workflow.
6077 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
6078 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
6079 earlier PAM modules. The data in these fields is used to initialize
6080 the session scope's resource properties. Thus external PAM modules
6081 may now configure per-session limits, for example sourced from
6082 external user databases.
6084 * socket units with Accept=yes will now maintain a "refused" counter in
6085 addition to the existing "accepted" counter, counting connections
6086 refused due to the enforced limits.
6088 * The "systemd-path search-binaries-default" command may now be use to
6089 query the default, built-in $PATH PID 1 will pass to the services it
6092 * A new unit file setting PrivateMounts= has been added. It's a boolean
6093 option. If enabled the unit's processes are invoked in their own file
6094 system namespace. Note that this behaviour is also implied if any
6095 other file system namespacing options (such as PrivateTmp=,
6096 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
6097 primarily useful for services that do not use any of the other file
6098 system namespacing options. One such service is systemd-udevd.service
6099 where this is now used by default.
6101 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
6102 when the system is booted in UEFI "secure mode".
6104 * A new unit "system-update-pre.target" is added, which defines an
6105 optional synchronization point for offline system updates, as
6106 implemented by the pre-existing "system-update.target" unit. It
6107 allows ordering services before the service that executes the actual
6108 update process in a generic way.
6110 * Systemd now emits warnings whenever .include syntax is used.
6112 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
6113 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
6114 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
6115 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
6116 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
6117 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
6118 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
6119 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
6120 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
6121 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
6122 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
6123 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
6124 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
6125 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
6126 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
6127 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
6128 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
6129 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
6130 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
6131 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
6132 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
6133 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
6134 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
6135 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
6136 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
6137 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
6138 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
6139 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
6140 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6142 — Berlin, 2018-06-22
6146 * The MemoryAccounting= unit property now defaults to on. After
6147 discussions with the upstream control group maintainers we learnt
6148 that the negative impact of cgroup memory accounting on current
6149 kernels is finally relatively minimal, so that it should be safe to
6150 enable this by default without affecting system performance. Besides
6151 memory accounting only task accounting is turned on by default, all
6152 other forms of resource accounting (CPU, IO, IP) remain off for now,
6153 because it's not clear yet that their impact is small enough to move
6154 from opt-in to opt-out. We recommend downstreams to leave memory
6155 accounting on by default if kernel 4.14 or higher is primarily
6156 used. On very resource constrained systems or when support for old
6157 kernels is a necessity, -Dmemory-accounting-default=false can be used
6158 to revert this change.
6160 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
6161 %udev_rules_update) and the journal catalog (%journal_catalog_update)
6162 from the upgrade scriptlets of individual packages now do nothing.
6163 Transfiletriggers have been added which will perform those updates
6164 once at the end of the transaction.
6166 Similar transfiletriggers have been added to execute any sysctl.d
6167 and binfmt.d rules. Thus, it should be unnecessary to provide any
6168 scriptlets to execute this configuration from package installation
6171 * systemd-sysusers gained a mode where the configuration to execute is
6172 specified on the command line, but this configuration is not executed
6173 directly, but instead it is merged with the configuration on disk,
6174 and the result is executed. This is useful for package installation
6175 scripts which want to create the user before installing any files on
6176 disk (in case some of those files are owned by that user), while
6177 still allowing local admin overrides.
6179 This functionality is exposed to rpm scriptlets through a new
6180 %sysusers_create_package macro. Old %sysusers_create and
6181 %sysusers_create_inline macros are deprecated.
6183 A transfiletrigger for sysusers.d configuration is now installed,
6184 which means that it should be unnecessary to call systemd-sysusers from
6185 package installation scripts, unless the package installs any files
6186 owned by those newly-created users, in which case
6187 %sysusers_create_package should be used.
6189 * Analogous change has been done for systemd-tmpfiles: it gained a mode
6190 where the command-line configuration is merged with the configuration
6191 on disk. This is exposed as the new %tmpfiles_create_package macro,
6192 and %tmpfiles_create is deprecated. A transfiletrigger is installed
6193 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
6194 from package installation scripts.
6196 * sysusers.d configuration for a user may now also specify the group
6197 number, in addition to the user number ("u username 123:456"), or
6198 without the user number ("u username -:456").
6200 * Configution items for systemd-sysusers can now be specified as
6201 positional arguments when the new --inline switch is used.
6203 * The login shell of users created through sysusers.d may now be
6204 specified (previously, it was always /bin/sh for root and
6205 /sbin/nologin for other users).
6207 * systemd-analyze gained a new --global switch to look at global user
6208 configuration. It also gained a unit-paths verb to list the unit load
6209 paths that are compiled into systemd (which can be used with
6210 --systemd, --user, or --global).
6212 * udevadm trigger gained a new --settle/-w option to wait for any
6213 triggered events to finish (but just those, and not any other events
6214 which are triggered meanwhile).
6216 * The action that systemd-logind takes when the lid is closed and the
6217 machine is connected to external power can now be configured using
6218 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
6219 was determined by HandleLidSwitch=, and, for backwards compatibility,
6220 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
6222 * journalctl will periodically call sd_journal_process() to make it
6223 resilient against inotify queue overruns when journal files are
6224 rotated very quickly.
6226 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
6227 sd_bus_get_n_queued_write — may be used to check the number of
6228 pending bus messages.
6230 * systemd gained a new
6231 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
6232 which can be used to migrate foreign processes to scope and service
6233 units. The primary user for this new API is systemd itself: the
6234 systemd --user instance uses this call of the systemd --system
6235 instance to migrate processes if it itself gets the request to
6236 migrate processes and the kernel refuses this due to access
6237 restrictions. Thanks to this "systemd-run --scope --user …" works
6238 again in pure cgroup v2 environments when invoked from the user
6241 * A new TemporaryFileSystem= setting can be used to mask out part of
6242 the real file system tree with tmpfs mounts. This may be combined
6243 with BindPaths= and BindReadOnlyPaths= to hide files or directories
6244 not relevant to the unit, while still allowing some paths lower in
6245 the tree to be accessed.
6247 ProtectHome=tmpfs may now be used to hide user home and runtime
6248 directories from units, in a way that is mostly equivalent to
6249 "TemporaryFileSystem=/home /run/user /root".
6251 * Non-service units are now started with KeyringMode=shared by default.
6252 This means that mount and swapon and other mount tools have access
6253 to keys in the main keyring.
6255 * /sys/fs/bpf is now mounted automatically.
6257 * QNX virtualization is now detected by systemd-detect-virt and may
6258 be used in ConditionVirtualization=.
6260 * IPAccounting= may now be enabled also for slice units.
6262 * A new -Dsplit-bin= build configuration switch may be used to specify
6263 whether bin and sbin directories are merged, or if they should be
6264 included separately in $PATH and various listings of executable
6265 directories. The build configuration scripts will try to autodetect
6266 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
6267 system, but distributions are encouraged to configure this
6270 * A new -Dok-color= build configuration switch may be used to change
6271 the colour of "OK" status messages.
6273 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
6274 PrivateNetwork=yes was buggy in previous versions of systemd. This
6275 means that after the upgrade and daemon-reexec, any such units must
6278 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
6279 will not exclude read-only files owned by root from cleanup.
6281 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
6282 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
6283 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
6284 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
6285 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
6286 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
6287 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
6288 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
6289 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
6290 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
6291 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
6292 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
6293 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
6294 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
6295 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
6296 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
6298 — Warsaw, 2018-03-05
6302 * Some keyboards come with a zoom see-saw or rocker which until now got
6303 mapped to the Linux "zoomin/out" keys in hwdb. However, these
6304 keycodes are not recognized by any major desktop. They now produce
6305 Up/Down key events so that they can be used for scrolling.
6307 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
6308 slightly: previously, if an argument was specified for lines of this
6309 type (i.e. the right-most column was set) this string was appended to
6310 existing files each time systemd-tmpfiles was run. This behaviour was
6311 different from what the documentation said, and not particularly
6312 useful, as repeated systemd-tmpfiles invocations would not be
6313 idempotent and grow such files without bounds. With this release
6314 behaviour has been altered to match what the documentation says:
6315 lines of this type only have an effect if the indicated files don't
6316 exist yet, and only then the argument string is written to the file.
6318 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
6319 systemd-tmpfiles behaviour: previously, read-only files owned by root
6320 were always excluded from the file "aging" algorithm (i.e. the
6321 automatic clean-up of directories like /tmp based on
6322 atime/mtime/ctime). We intend to drop this restriction, and age files
6323 by default even when owned by root and read-only. This behaviour was
6324 inherited from older tools, but there have been requests to remove
6325 it, and it's not obvious why this restriction was made in the first
6326 place. Please speak up now, if you are aware of software that requires
6327 this behaviour, otherwise we'll remove the restriction in v238.
6329 * A new environment variable $SYSTEMD_OFFLINE is now understood by
6330 systemctl. It takes a boolean argument. If on, systemctl assumes it
6331 operates on an "offline" OS tree, and will not attempt to talk to the
6332 service manager. Previously, this mode was implicitly enabled if a
6333 chroot() environment was detected, and this new environment variable
6334 now provides explicit control.
6336 * .path and .socket units may now be created transiently, too.
6337 Previously only service, mount, automount and timer units were
6338 supported as transient units. The systemd-run tool has been updated
6339 to expose this new functionality, you may hence use it now to bind
6340 arbitrary commands to path or socket activation on-the-fly from the
6341 command line. Moreover, almost all properties are now exposed for the
6342 unit types that already supported transient operation.
6344 * The systemd-mount command gained support for a new --owner= parameter
6345 which takes a user name, which is then resolved and included in uid=
6346 and gid= mount options string of the file system to mount.
6348 * A new unit condition ConditionControlGroupController= has been added
6349 that checks whether a specific cgroup controller is available.
6351 * Unit files, udev's .link files, and systemd-networkd's .netdev and
6352 .network files all gained support for a new condition
6353 ConditionKernelVersion= for checking against specific kernel
6356 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
6357 support for configuring device flags in the Flags= setting. In the
6358 same files, the [Tunnel] section gained support for configuring
6359 AllowLocalRemote=. The [Route] section in .network files gained
6360 support for configuring InitialCongestionWindow=,
6361 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
6362 understands RapidCommit=.
6364 * systemd-networkd's DHCPv6 support gained support for Prefix
6367 * sd-bus gained support for a new "watch-bind" feature. When this
6368 feature is enabled, an sd_bus connection may be set up to connect to
6369 an AF_UNIX socket in the file system as soon as it is created. This
6370 functionality is useful for writing early-boot services that
6371 automatically connect to the system bus as soon as it is started,
6372 without ugly time-based polling. systemd-networkd and
6373 systemd-resolved have been updated to make use of this
6374 functionality. busctl exposes this functionality in a new
6375 --watch-bind= command line switch.
6377 * sd-bus will now optionally synthesize a local "Connected" signal as
6378 soon as a D-Bus connection is set up fully. This message mirrors the
6379 already existing "Disconnected" signal which is synthesized when the
6380 connection is terminated. This signal is generally useful but
6381 particularly handy in combination with the "watch-bind" feature
6382 described above. Synthesizing of this message has to be requested
6383 explicitly through the new API call sd_bus_set_connected_signal(). In
6384 addition a new call sd_bus_is_ready() has been added that checks
6385 whether a connection is fully set up (i.e. between the "Connected" and
6386 "Disconnected" signals).
6388 * sd-bus gained two new calls sd_bus_request_name_async() and
6389 sd_bus_release_name_async() for asynchronously registering bus
6390 names. Similar, there is now sd_bus_add_match_async() for installing
6391 a signal match asynchronously. All of systemd's own services have
6392 been updated to make use of these calls. Doing these operations
6393 asynchronously has two benefits: it reduces the risk of deadlocks in
6394 case of cyclic dependencies between bus services, and it speeds up
6395 service initialization since synchronization points for bus
6396 round-trips are removed.
6398 * sd-bus gained two new calls sd_bus_match_signal() and
6399 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
6400 and sd_bus_add_match_async() but instead of taking a D-Bus match
6401 string take match fields as normal function parameters.
6403 * sd-bus gained two new calls sd_bus_set_sender() and
6404 sd_bus_message_set_sender() for setting the sender name of outgoing
6405 messages (either for all outgoing messages or for just one specific
6406 one). These calls are only useful in direct connections as on
6407 brokered connections the broker fills in the sender anyway,
6408 overwriting whatever the client filled in.
6410 * sd-event gained a new pseudo-handle that may be specified on all API
6411 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
6412 used this refers to the default event loop object of the calling
6413 thread. Note however that this does not implicitly allocate one —
6414 which has to be done prior by using sd_event_default(). Similarly
6415 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
6416 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
6417 to the default bus of the specified type of the calling thread. Here
6418 too this does not implicitly allocate bus connection objects, this
6419 has to be done prior with sd_bus_default() and friends.
6421 * sd-event gained a new call pair
6422 sd_event_source_{get|set}_io_fd_own(). This may be used to request
6423 automatic closure of the file descriptor an IO event source watches
6424 when the event source is destroyed.
6426 * systemd-networkd gained support for natively configuring WireGuard
6429 * In previous versions systemd synthesized user records both for the
6430 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
6431 internally. In order to simplify distribution-wide renames of the
6432 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
6433 new transitional flag file has been added: if
6434 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
6435 user and group record within the systemd codebase is disabled.
6437 * systemd-notify gained a new --uid= option for selecting the source
6438 user/UID to use for notification messages sent to the service
6441 * journalctl gained a new --grep= option to list only entries in which
6442 the message matches a certain pattern. By default matching is case
6443 insensitive if the pattern is lowercase, and case sensitive
6444 otherwise. Option --case-sensitive=yes|no can be used to override
6445 this an specify case sensitivity or case insensitivity.
6447 * There's now a "systemd-analyze service-watchdogs" command for printing
6448 the current state of the service runtime watchdog, and optionally
6449 enabling or disabling the per-service watchdogs system-wide if given a
6450 boolean argument (i.e. the concept you configure in WatchdogSec=), for
6451 debugging purposes. There's also a kernel command line option
6452 systemd.service_watchdogs= for controlling the same.
6454 * Two new "log-level" and "log-target" options for systemd-analyze were
6455 added that merge the now deprecated get-log-level, set-log-level and
6456 get-log-target, set-log-target pairs. The deprecated options are still
6457 understood for backwards compatibility. The two new options print the
6458 current value when no arguments are given, and set them when a
6459 level/target is given as an argument.
6461 * sysusers.d's "u" lines now optionally accept both a UID and a GID
6462 specification, separated by a ":" character, in order to create users
6463 where UID and GID do not match.
6465 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
6466 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
6467 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
6468 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
6469 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
6470 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
6471 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
6472 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
6473 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
6474 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
6475 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
6476 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
6477 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
6478 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
6479 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
6480 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
6481 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
6482 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
6483 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
6484 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
6491 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
6492 in v235 has been extended to also set the dummy.ko module option
6493 numdummies=0, preventing the kernel from automatically creating
6494 dummy0. All dummy interfaces must now be explicitly created.
6496 * Unknown '%' specifiers in configuration files are now rejected. This
6497 applies to units and tmpfiles.d configuration. Any percent characters
6498 that are followed by a letter or digit that are not supposed to be
6499 interpreted as the beginning of a specifier should be escaped by
6500 doubling ("%%"). (So "size=5%" is still accepted, as well as
6501 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
6502 valid specifiers today.)
6504 * systemd-resolved now maintains a new dynamic
6505 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
6506 recommended to make /etc/resolv.conf a symlink to it. This file
6507 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
6508 includes dynamically acquired search domains, achieving more correct
6509 DNS resolution by software that bypasses local DNS APIs such as NSS.
6511 * The "uaccess" udev tag has been dropped from /dev/kvm and
6512 /dev/dri/renderD*. These devices now have the 0666 permissions by
6513 default (but this may be changed at build-time). /dev/dri/renderD*
6514 will now be owned by the "render" group along with /dev/kfd.
6516 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
6517 systemd-journal-gatewayd.service and
6518 systemd-journal-upload.service. This means "nss-systemd" must be
6519 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
6520 services are resolved properly.
6522 * In /etc/fstab two new mount options are now understood:
6523 x-systemd.makefs and x-systemd.growfs. The former has the effect that
6524 the configured file system is formatted before it is mounted, the
6525 latter that the file system is resized to the full block device size
6526 after it is mounted (i.e. if the file system is smaller than the
6527 partition it resides on, it's grown). This is similar to the fsck
6528 logic in /etc/fstab, and pulls in systemd-makefs@.service and
6529 systemd-growfs@.service as necessary, similar to
6530 systemd-fsck@.service. Resizing is currently only supported on ext4
6533 * In systemd-networkd, the IPv6 RA logic now optionally may announce
6534 DNS server and domain information.
6536 * Support for the LUKS2 on-disk format for encrypted partitions has
6537 been added. This requires libcryptsetup2 during compilation and
6540 * The systemd --user instance will now signal "readiness" when its
6541 basic.target unit has been reached, instead of when the run queue ran
6542 empty for the first time.
6544 * Tmpfiles.d with user configuration are now also supported.
6545 systemd-tmpfiles gained a new --user switch, and snippets placed in
6546 ~/.config/user-tmpfiles.d/ and corresponding directories will be
6547 executed by systemd-tmpfiles --user running in the new
6548 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
6549 running in the user session.
6551 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
6552 %S resolves to the top-level state directory (/var/lib for the system
6553 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
6554 top-level cache directory (/var/cache for the system instance,
6555 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
6556 logs directory (/var/log for the system instance,
6557 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
6558 existing %t specifier, that resolves to the top-level runtime
6559 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
6562 * journalctl learnt a new parameter --output-fields= for limiting the
6563 set of journal fields to output in verbose and JSON output modes.
6565 * systemd-timesyncd's configuration file gained a new option
6566 RootDistanceMaxSec= for setting the maximum root distance of servers
6567 it'll use, as well as the new options PollIntervalMinSec= and
6568 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
6570 * bootctl gained a new command "list" for listing all available boot
6571 menu items on systems that follow the boot loader specification.
6573 * systemctl gained a new --dry-run switch that shows what would be done
6574 instead of doing it, and is currently supported by the shutdown and
6577 * ConditionSecurity= can now detect the TOMOYO security module.
6579 * Unit file [Install] sections are now also respected in unit drop-in
6580 files. This is intended to be used by drop-ins under /usr/lib/.
6582 * systemd-firstboot may now also set the initial keyboard mapping.
6584 * Udev "changed" events for devices which are exposed as systemd
6585 .device units are now propagated to units specified in
6586 ReloadPropagatedFrom= as reload requests.
6588 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
6589 unit template name (i.e. a name in the form of 'foobar@.service',
6590 without the instance component between the '@' and - the '.'), then
6591 the escaped sysfs path of the device is automatically used as the
6594 * SystemCallFilter= in unit files has been extended so that an "errno"
6595 can be specified individually for each system call. Example:
6596 SystemCallFilter=~uname:EILSEQ.
6598 * The cgroup delegation logic has been substantially updated. Delegate=
6599 now optionally takes a list of controllers (instead of a boolean, as
6600 before), which lists the controllers to delegate at least.
6602 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
6604 * A new LogLevelMax= setting configures the maximum log level any
6605 process of the service may log at (i.e. anything with a lesser
6606 priority than what is specified is automatically dropped). A new
6607 LogExtraFields= setting allows configuration of additional journal
6608 fields to attach to all log records generated by any of the unit's
6611 * New StandardInputData= and StandardInputText= settings along with the
6612 new option StandardInput=data may be used to configure textual or
6613 binary data that shall be passed to the executed service process via
6614 standard input, encoded in-line in the unit file.
6616 * StandardInput=, StandardOutput= and StandardError= may now be used to
6617 connect stdin/stdout/stderr of executed processes directly with a
6618 file or AF_UNIX socket in the file system, using the new "file:" option.
6620 * A new unit file option CollectMode= has been added, that allows
6621 tweaking the garbage collection logic for units. It may be used to
6622 tell systemd to garbage collect units that have failed automatically
6623 (normally it only GCs units that exited successfully). systemd-run
6624 and systemd-mount expose this new functionality with a new -G option.
6626 * "machinectl bind" may now be used to bind mount non-directories
6627 (i.e. regularfiles, devices, fifos, sockets).
6629 * systemd-analyze gained a new verb "calendar" for validating and
6630 testing calendar time specifications to use for OnCalendar= in timer
6631 units. Besides validating the expression it will calculate the next
6632 time the specified expression would elapse.
6634 * In addition to the pre-existing FailureAction= unit file setting
6635 there's now SuccessAction=, for configuring a shutdown action to
6636 execute when a unit completes successfully. This is useful in
6637 particular inside containers that shall terminate after some workload
6638 has been completed. Also, both options are now supported for all unit
6639 types, not just services.
6641 * networkds's IP rule support gained two new options
6642 IncomingInterface= and OutgoingInterface= for configuring the incoming
6643 and outgoing interfaces of configured rules. systemd-networkd also
6644 gained support for "vxcan" network devices.
6646 * networkd gained a new setting RequiredForOnline=, taking a
6647 boolean. If set, systemd-wait-online will take it into consideration
6648 when determining that the system is up, otherwise it will ignore the
6649 interface for this purpose.
6651 * The sd_notify() protocol gained support for a new operation: with
6652 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
6653 store again, ahead of POLLHUP or POLLERR when they are removed
6656 * A new document doc/UIDS-GIDS.md has been added to the source tree,
6657 that documents the UID/GID range and assignment assumptions and
6658 requirements of systemd.
6660 * The watchdog device PID 1 will ping may now be configured through the
6661 WatchdogDevice= configuration file setting, or by setting the
6662 systemd.watchdog_service= kernel commandline option.
6664 * systemd-resolved's gained support for registering DNS-SD services on
6665 the local network using MulticastDNS. Services may either be
6666 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
6667 the same dir below /run, /usr/lib), or through its D-Bus API.
6669 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
6670 extend the effective start, runtime, and stop time. The service must
6671 continue to send EXTEND_TIMEOUT_USEC within the period specified to
6672 prevent the service manager from making the service as timedout.
6674 * systemd-resolved's DNSSEC support gained support for RFC 8080
6675 (Ed25519 keys and signatures).
6677 * The systemd-resolve command line tool gained a new set of options
6678 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
6679 --set-nta= and --revert to configure per-interface DNS configuration
6680 dynamically during runtime. It's useful for pushing DNS information
6681 into systemd-resolved from DNS hook scripts that various interface
6682 managing software supports (such as pppd).
6684 * systemd-nspawn gained a new --network-namespace-path= command line
6685 option, which may be used to make a container join an existing
6686 network namespace, by specifying a path to a "netns" file.
6688 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
6689 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
6690 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
6691 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
6692 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
6693 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
6694 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
6695 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
6696 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
6697 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
6698 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
6699 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
6700 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
6701 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
6702 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
6703 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
6704 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
6705 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
6706 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
6707 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
6708 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
6709 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
6710 Jędrzejewski-Szmek, Zeal Jagannatha
6712 — Berlin, 2017-12-14
6716 * INCOMPATIBILITY: systemd-logind.service and other long-running
6717 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
6718 communication with the outside. This generally improves security of
6719 the system, and is in almost all cases a safe and good choice, as
6720 these services do not and should not provide any network-facing
6721 functionality. However, systemd-logind uses the glibc NSS API to
6722 query the user database. This creates problems on systems where NSS
6723 is set up to directly consult network services for user database
6724 lookups. In particular, this creates incompatibilities with the
6725 "nss-nis" module, which attempts to directly contact the NIS/YP
6726 network servers it is configured for, and will now consistently
6727 fail. In such cases, it is possible to turn off IP sandboxing for
6728 systemd-logind.service (set IPAddressDeny= in its [Service] section
6729 to the empty string, via a .d/ unit file drop-in). Downstream
6730 distributions might want to update their nss-nis packaging to include
6731 such a drop-in snippet, accordingly, to hide this incompatibility
6732 from the user. Another option is to make use of glibc's nscd service
6733 to proxy such network requests through a privilege-separated, minimal
6734 local caching daemon, or to switch to more modern technologies such
6735 sssd, whose NSS hook-ups generally do not involve direct network
6736 access. In general, we think it's definitely time to question the
6737 implementation choices of nss-nis, i.e. whether it's a good idea
6738 today to embed a network-facing loadable module into all local
6739 processes that need to query the user database, including the most
6740 trivial and benign ones, such as "ls". For more details about
6741 IPAddressDeny= see below.
6743 * A new modprobe.d drop-in is now shipped by default that sets the
6744 bonding module option max_bonds=0. This overrides the kernel default,
6745 to avoid conflicts and ambiguity as to whether or not bond0 should be
6746 managed by systemd-networkd or not. This resolves multiple issues
6747 with bond0 properties not being applied, when bond0 is configured
6748 with systemd-networkd. Distributors may choose to not package this,
6749 however in that case users will be prevented from correctly managing
6750 bond0 interface using systemd-networkd.
6752 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
6753 which print the logging level and target of the system manager. They
6754 complement the existing "set-log-level" and "set-log-target" verbs
6755 used to change those values.
6757 * journald.conf gained a new boolean setting ReadKMsg= which defaults
6758 to on. If turned off kernel log messages will not be read by
6759 systemd-journald or included in the logs. It also gained a new
6760 setting LineMax= for configuring the maximum line length in
6761 STDOUT/STDERR log streams. The new default for this value is 48K, up
6762 from the previous hardcoded 2048.
6764 * A new unit setting RuntimeDirectoryPreserve= has been added, which
6765 allows more detailed control of what to do with a runtime directory
6766 configured with RuntimeDirectory= (i.e. a directory below /run or
6767 $XDG_RUNTIME_DIR) after a unit is stopped.
6769 * The RuntimeDirectory= setting for units gained support for creating
6770 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
6771 one top-level directory.
6773 * Units gained new options StateDirectory=, CacheDirectory=,
6774 LogsDirectory= and ConfigurationDirectory= which are closely related
6775 to RuntimeDirectory= but manage per-service directories below
6776 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
6777 possible to write unit files which when activated automatically gain
6778 properly owned service specific directories in these locations, thus
6779 making unit files self-contained and increasing compatibility with
6780 stateless systems and factory reset where /etc or /var are
6781 unpopulated at boot. Matching these new settings there's also
6782 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
6783 ConfigurationDirectoryMode= for configuring the access mode of these
6784 directories. These settings are particularly useful in combination
6785 with DynamicUser=yes as they provide secure, properly-owned,
6786 writable, and stateful locations for storage, excluded from the
6787 sandbox that such services live in otherwise.
6789 * Automake support has been removed from this release. systemd is now
6792 * systemd-journald will now aggressively cache client metadata during
6793 runtime, speeding up log write performance under pressure. This comes
6794 at a small price though: as much of the metadata is read
6795 asynchronously from /proc/ (and isn't implicitly attached to log
6796 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
6797 metadata stored alongside a log entry might be slightly
6798 out-of-date. Previously it could only be slightly newer than the log
6799 message. The time window is small however, and given that the kernel
6800 is unlikely to be improved anytime soon in this regard, this appears
6803 * nss-myhostname/systemd-resolved will now by default synthesize an
6804 A/AAAA resource record for the "_gateway" hostname, pointing to the
6805 current default IP gateway. Previously it did that for the "gateway"
6806 name, hampering adoption, as some distributions wanted to leave that
6807 hostname open for local use. The old behaviour may still be
6808 requested at build time.
6810 * systemd-networkd's [Address] section in .network files gained a new
6811 Scope= setting for configuring the IP address scope. The [Network]
6812 section gained a new boolean setting ConfigureWithoutCarrier= that
6813 tells systemd-networkd to ignore link sensing when configuring the
6814 device. The [DHCP] section gained a new Anonymize= boolean option for
6815 turning on a number of options suggested in RFC 7844. A new
6816 [RoutingPolicyRule] section has been added for configuring the IP
6817 routing policy. The [Route] section has gained support for a new
6818 Type= setting which permits configuring
6819 blackhole/unreachable/prohibit routes.
6821 * The [VRF] section in .netdev files gained a new Table= setting for
6822 configuring the routing table to use. The [Tunnel] section gained a
6823 new Independent= boolean field for configuring tunnels independent of
6824 an underlying network interface. The [Bridge] section gained a new
6825 GroupForwardMask= option for configuration of propagation of link
6826 local frames between bridge ports.
6828 * The WakeOnLan= setting in .link files gained support for a number of
6829 new modes. A new TCP6SegmentationOffload= setting has been added for
6830 configuring TCP/IPv6 hardware segmentation offload.
6832 * The IPv6 RA sender implementation may now optionally send out RDNSS
6833 and RDNSSL records to supply DNS configuration to peers.
6835 * systemd-nspawn gained support for a new --system-call-filter= command
6836 line option for adding and removing entries in the default system
6837 call filter it applies. Moreover systemd-nspawn has been changed to
6838 implement a system call allow list instead of a deny list.
6840 * systemd-run gained support for a new --pipe command line option. If
6841 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
6842 are directly passed on to the activated transient service
6843 executable. This allows invoking arbitrary processes as systemd
6844 services (for example to take benefit of dependency management,
6845 accounting management, resource management or log management that is
6846 done automatically for services) — while still allowing them to be
6847 integrated in a classic UNIX shell pipeline.
6849 * When a service sends RELOAD=1 via sd_notify() and reload propagation
6850 using ReloadPropagationTo= is configured, a reload is now propagated
6851 to configured units. (Previously this was only done on explicitly
6852 requested reloads, using "systemctl reload" or an equivalent
6855 * For each service unit a restart counter is now kept: it is increased
6856 each time the service is restarted due to Restart=, and may be
6857 queried using "systemctl show -p NRestarts …".
6859 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
6860 @signal and @timer have been added, for usage with SystemCallFilter=
6861 in unit files and the new --system-call-filter= command line option
6862 of systemd-nspawn (see above).
6864 * ExecStart= lines in unit files gained two new modifiers: when a
6865 command line is prefixed with "!" the command will be executed as
6866 configured, except for the credentials applied by
6867 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
6868 "+", but does still apply namespacing options unlike "+". There's
6869 also "!!" now, which is mostly identical, but becomes a NOP on
6870 systems that support ambient capabilities. This is useful to write
6871 unit files that work with ambient capabilities where possible but
6872 automatically fall back to traditional privilege dropping mechanisms
6873 on systems where this is not supported.
6875 * ListenNetlink= settings in socket units now support RDMA netlink
6878 * A new unit file setting LockPersonality= has been added which permits
6879 locking down the chosen execution domain ("personality") of a service
6882 * A new special target "getty-pre.target" has been added, which is
6883 ordered before all text logins, and may be used to order services
6884 before textual logins acquire access to the console.
6886 * systemd will now attempt to load the virtio-rng.ko kernel module very
6887 early on if a VM environment supporting this is detected. This should
6888 improve entropy during early boot in virtualized environments.
6890 * A _netdev option is now supported in /etc/crypttab that operates in a
6891 similar way as the same option in /etc/fstab: it permits configuring
6892 encrypted devices that need to be ordered after the network is up.
6893 Following this logic, two new special targets
6894 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
6895 added that are to cryptsetup.target what remote-fs.target and
6896 remote-fs-pre.target are to local-fs.target.
6898 * Service units gained a new UnsetEnvironment= setting which permits
6899 unsetting specific environment variables for services that are
6900 normally passed to it (for example in order to mask out locale
6901 settings for specific services that can't deal with it).
6903 * Units acquired a new boolean option IPAccounting=. When turned on, IP
6904 traffic accounting (packet count as well as byte count) is done for
6905 the service, and shown as part of "systemctl status" or "systemd-run
6908 * Service units acquired two new options IPAddressAllow= and
6909 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
6910 for configuring a simple IP access control list for all sockets of
6911 the unit. These options are available also on .slice and .socket
6912 units, permitting flexible access list configuration for individual
6913 services as well as groups of services (as defined by a slice unit),
6914 including system-wide. Note that IP ACLs configured this way are
6915 enforced on every single IPv4 and IPv6 socket created by any process
6916 of the service unit, and apply to ingress as well as egress traffic.
6918 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
6919 structured log message is generated each time the unit is stopped,
6920 containing information about the consumed resources of this
6923 * A new setting KeyringMode= has been added to unit files, which may be
6924 used to control how the kernel keyring is set up for executed
6927 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
6928 "systemctl kexec" and "systemctl exit" are now always asynchronous in
6929 behaviour (that is: these commands return immediately after the
6930 operation was enqueued instead of waiting for the operation to
6931 complete). Previously, "systemctl poweroff" and "systemctl reboot"
6932 were asynchronous on systems using systemd-logind (i.e. almost
6933 always, and like they were on sysvinit), and the other three commands
6934 were unconditionally synchronous. With this release this is cleaned
6935 up, and callers will see the same asynchronous behaviour on all
6936 systems for all five operations.
6938 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
6941 * .timer units now accept calendar specifications in other timezones
6942 than UTC or the local timezone.
6944 * The tmpfiles snippet var.conf has been changed to create
6945 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
6946 the "utmp" group already, and it appears to be generally understood
6947 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
6948 databases. Previously this was implemented correctly for all these
6949 databases excepts btmp, which has been opened up like this now
6950 too. Note that while the other databases are world-readable
6951 (i.e. 0644), btmp is not and remains more restrictive.
6953 * The systemd-resolve tool gained a new --reset-server-features
6954 switch. When invoked like this systemd-resolved will forget
6955 everything it learnt about the features supported by the configured
6956 upstream DNS servers, and restarts the feature probing logic on the
6957 next resolver look-up for them at the highest feature level
6960 * The status dump systemd-resolved sends to the logs upon receiving
6961 SIGUSR1 now also includes information about all DNS servers it is
6962 configured to use, and the features levels it probed for them.
6964 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
6965 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
6966 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
6967 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
6968 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
6969 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
6970 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
6971 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
6972 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
6973 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
6974 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
6975 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
6976 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
6977 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
6978 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
6979 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
6980 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
6981 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
6982 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
6983 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6985 — Berlin, 2017-10-06
6989 * Meson is now supported as build system in addition to Automake. It is
6990 our plan to remove Automake in one of our next releases, so that
6991 Meson becomes our exclusive build system. Hence, please start using
6992 the Meson build system in your downstream packaging. There's plenty
6993 of documentation around how to use Meson, the extremely brief
6996 ./autogen.sh && ./configure && make && sudo make install
7000 meson build && ninja -C build && sudo ninja -C build install
7002 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
7003 which permits configuring a timeout on the time a job is
7004 running. This is particularly useful for setting timeouts on jobs for
7007 * Unit files gained two new options ConditionUser= and ConditionGroup=
7008 for conditionalizing units based on the identity of the user/group
7009 running a systemd user instance.
7011 * systemd-networkd now understands a new FlowLabel= setting in the
7012 [VXLAN] section of .network files, as well as a Priority= in
7013 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
7014 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
7015 gained support for configuration of GENEVE links, and IPv6 address
7016 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
7018 * .link files now understand a new Port= setting.
7020 * systemd-networkd's DHCP support gained support for DHCP option 119
7021 (domain search list).
7023 * systemd-networkd gained support for serving IPv6 address ranges using
7024 the Router Advertisement protocol. The new .network configuration
7025 section [IPv6Prefix] may be used to configure the ranges to
7026 serve. This is implemented based on a new, minimal, native server
7027 implementation of RA.
7029 * journalctl's --output= switch gained support for a new parameter
7030 "short-iso-precise" for a mode where timestamps are shown as precise
7033 * systemd-udevd's "net_id" builtin may now generate stable network
7034 interface names from IBM PowerVM VIO devices as well as ACPI platform
7037 * MulticastDNS support in systemd-resolved may now be explicitly
7038 enabled/disabled using the new MulticastDNS= configuration file
7041 * systemd-resolved may now optionally use libidn2 instead of the libidn
7042 for processing internationalized domain names. Support for libidn2
7043 should be considered experimental and should not be enabled by
7046 * "machinectl pull-tar" and related call may now do verification of
7047 downloaded images using SUSE-style .sha256 checksum files in addition
7048 to the already existing support for validating using Ubuntu-style
7051 * sd-bus gained support for a new sd_bus_message_appendv() call which
7052 is va_list equivalent of sd_bus_message_append().
7054 * sd-boot gained support for validating images using SHIM/MOK.
7056 * The SMACK code learnt support for "onlycap".
7058 * systemd-mount --umount is now much smarter in figuring out how to
7059 properly unmount a device given its mount or device path.
7061 * The code to call libnss_dns as a fallback from libnss_resolve when
7062 the communication with systemd-resolved fails was removed. This
7063 fallback was redundant and interfered with the [!UNAVAIL=return]
7064 suffix. See nss-resolve(8) for the recommended configuration.
7066 * systemd-logind may now be restarted without losing state. It stores
7067 the file descriptors for devices it manages in the system manager
7068 using the FDSTORE= mechanism. Please note that further changes in
7069 other components may be required to make use of this (for example
7070 Xorg has code to listen for stops of systemd-logind and terminate
7071 itself when logind is stopped or restarted, in order to avoid using
7072 stale file descriptors for graphical devices, which is now
7073 counterproductive and must be reverted in order for restarts of
7074 systemd-logind to be safe. See
7075 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
7077 * All kernel-install plugins are called with the environment variable
7078 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
7079 /etc/machine-id. If the machine ID could not be determined,
7080 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
7081 anything in the entry directory (passed as the second argument) if
7082 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
7083 temporary directory is passed as the entry directory and removed
7084 after all the plugins exit.
7086 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
7087 will now use its value as the machine ID instead of the machine ID
7088 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
7089 /etc/machine-info and no machine ID is set in /etc/machine-id,
7090 kernel-install will try to store the current machine ID there as
7091 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
7092 will generate a new UUID, store it in /etc/machine-info as
7093 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
7095 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
7096 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
7097 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
7098 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
7099 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
7100 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
7101 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
7102 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
7103 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
7104 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
7105 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
7106 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
7107 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
7108 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
7109 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
7110 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
7111 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
7112 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
7113 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
7114 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
7115 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
7116 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
7117 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
7118 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
7119 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
7120 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
7121 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
7122 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
7123 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
7126 — Berlin, 2017-07-12
7130 * The "hybrid" control group mode has been modified to improve
7131 compatibility with "legacy" cgroups-v1 setups. Specifically, the
7132 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
7133 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
7134 cgroups-v1 hierarchy), the only externally visible change being that
7135 the cgroups-v2 hierarchy is also mounted, to
7136 /sys/fs/cgroup/unified. This should provide a large degree of
7137 compatibility with "legacy" cgroups-v1, while taking benefit of the
7138 better management capabilities of cgroups-v2.
7140 * The default control group setup mode may be selected both a boot-time
7141 via a set of kernel command line parameters (specifically:
7142 systemd.unified_cgroup_hierarchy= and
7143 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
7144 default selected on the configure command line
7145 (--with-default-hierarchy=). The upstream default is "hybrid"
7146 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
7147 this will change in a future systemd version to be "unified" (pure
7148 cgroups-v2 mode). The third option for the compile time option is
7149 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
7150 distributions to default to "hybrid" mode for release distributions,
7151 starting with v233. We recommend "unified" for development
7152 distributions (specifically: distributions such as Fedora's rawhide)
7153 as that's where things are headed in the long run. Use "legacy" for
7154 greatest stability and compatibility only.
7156 * Note one current limitation of "unified" and "hybrid" control group
7157 setup modes: the kernel currently does not permit the systemd --user
7158 instance (i.e. unprivileged code) to migrate processes between two
7159 disconnected cgroup subtrees, even if both are managed and owned by
7160 the user. This effectively means "systemd-run --user --scope" doesn't
7161 work when invoked from outside of any "systemd --user" service or
7162 scope. Specifically, it is not supported from session scopes. We are
7163 working on fixing this in a future systemd version. (See #3388 for
7164 further details about this.)
7166 * DBus policy files are now installed into /usr rather than /etc. Make
7167 sure your system has dbus >= 1.9.18 running before upgrading to this
7168 version, or override the install path with --with-dbuspolicydir= .
7170 * All python scripts shipped with systemd (specifically: the various
7171 tests written in Python) now require Python 3.
7173 * systemd unit tests can now run standalone (without the source or
7174 build directories), and can be installed into /usr/lib/systemd/tests/
7175 with 'make install-tests'.
7177 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
7178 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
7181 * Support for the %c, %r, %R specifiers in unit files has been
7182 removed. Specifiers are not supposed to be dependent on configuration
7183 in the unit file itself (so that they resolve the same regardless
7184 where used in the unit files), but these specifiers were influenced
7185 by the Slice= option.
7187 * The shell invoked by debug-shell.service now defaults to /bin/sh in
7188 all cases. If distributions want to use a different shell for this
7189 purpose (for example Fedora's /sbin/sushell) they need to specify
7190 this explicitly at configure time using --with-debug-shell=.
7192 * The confirmation spawn prompt has been reworked to offer the
7195 (c)ontinue, proceed without asking anymore
7196 (D)ump, show the state of the unit
7197 (f)ail, don't execute the command and pretend it failed
7199 (i)nfo, show a short summary of the unit
7200 (j)obs, show jobs that are in progress
7201 (s)kip, don't execute the command and pretend it succeeded
7202 (y)es, execute the command
7204 The 'n' choice for the confirmation spawn prompt has been removed,
7205 because its meaning was confusing.
7207 The prompt may now also be redirected to an alternative console by
7208 specifying the console as parameter to systemd.confirm_spawn=.
7210 * Services of Type=notify require a READY=1 notification to be sent
7211 during startup. If no such message is sent, the service now fails,
7212 even if the main process exited with a successful exit code.
7214 * Services that fail to start up correctly now always have their
7215 ExecStopPost= commands executed. Previously, they'd enter "failed"
7216 state directly, without executing these commands.
7218 * The option MulticastDNS= of network configuration files has acquired
7219 an actual implementation. With MulticastDNS=yes a host can resolve
7220 names of remote hosts and reply to mDNS A and AAAA requests.
7222 * When units are about to be started an additional check is now done to
7223 ensure that all dependencies of type BindsTo= (when used in
7224 combination with After=) have been started.
7226 * systemd-analyze gained a new verb "syscall-filter" which shows which
7227 system call groups are defined for the SystemCallFilter= unit file
7228 setting, and which system calls they contain.
7230 * A new system call filter group "@filesystem" has been added,
7231 consisting of various file system related system calls. Group
7232 "@reboot" has been added, covering reboot, kexec and shutdown related
7233 calls. Finally, group "@swap" has been added covering swap
7234 configuration related calls.
7236 * A new unit file option RestrictNamespaces= has been added that may be
7237 used to restrict access to the various process namespace types the
7238 Linux kernel provides. Specifically, it may be used to take away the
7239 right for a service unit to create additional file system, network,
7240 user, and other namespaces. This sandboxing option is particularly
7241 relevant due to the high amount of recently discovered namespacing
7242 related vulnerabilities in the kernel.
7244 * systemd-udev's .link files gained support for a new AutoNegotiation=
7245 setting for configuring Ethernet auto-negotiation.
7247 * systemd-networkd's .network files gained support for a new
7248 ListenPort= setting in the [DHCP] section to explicitly configure the
7249 UDP client port the DHCP client shall listen on.
7251 * .network files gained a new Unmanaged= boolean setting for explicitly
7252 excluding one or more interfaces from management by systemd-networkd.
7254 * The systemd-networkd ProxyARP= option has been renamed to
7255 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
7256 renamed to ReduceARPProxy=. The old names continue to be available
7259 * systemd-networkd gained support for configuring IPv6 Proxy NDP
7260 addresses via the new IPv6ProxyNDPAddress= .network file setting.
7262 * systemd-networkd's bonding device support gained support for two new
7263 configuration options ActiveSlave= and PrimarySlave=.
7265 * The various options in the [Match] section of .network files gained
7266 support for negative matching.
7268 * New systemd-specific mount options are now understood in /etc/fstab:
7270 x-systemd.mount-timeout= may be used to configure the maximum
7271 permitted runtime of the mount command.
7273 x-systemd.device-bound may be set to bind a mount point to its
7274 backing device unit, in order to automatically remove a mount point
7275 if its backing device is unplugged. This option may also be
7276 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
7277 on the block device, which is now automatically set for all CDROM
7278 drives, so that mounted CDs are automatically unmounted when they are
7279 removed from the drive.
7281 x-systemd.after= and x-systemd.before= may be used to explicitly
7282 order a mount after or before another unit or mount point.
7284 * Enqueued start jobs for device units are now automatically garbage
7285 collected if there are no jobs waiting for them anymore.
7287 * systemctl list-jobs gained two new switches: with --after, for every
7288 queued job the jobs it's waiting for are shown; with --before the
7289 jobs which it's blocking are shown.
7291 * systemd-nspawn gained support for ephemeral boots from disk images
7292 (or in other words: --ephemeral and --image= may now be
7293 combined). Moreover, ephemeral boots are now supported for normal
7294 directories, even if the backing file system is not btrfs. Of course,
7295 if the file system does not support file system snapshots or
7296 reflinks, the initial copy operation will be relatively expensive, but
7297 this should still be suitable for many use cases.
7299 * Calendar time specifications in .timer units now support
7300 specifications relative to the end of a month by using "~" instead of
7301 "-" as separator between month and day. For example, "*-02~03" means
7302 "the third last day in February". In addition a new syntax for
7303 repeated events has been added using the "/" character. For example,
7304 "9..17/2:00" means "every two hours from 9am to 5pm".
7306 * systemd-socket-proxyd gained a new parameter --connections-max= for
7307 configuring the maximum number of concurrent connections.
7309 * sd-id128 gained a new API for generating unique IDs for the host in a
7310 way that does not leak the machine ID. Specifically,
7311 sd_id128_get_machine_app_specific() derives an ID based on the
7312 machine ID in a well-defined, non-reversible, stable way. This is
7313 useful whenever an identifier for the host is needed but where the
7314 identifier shall not be useful to identify the system beyond the
7315 scope of the application itself. (Internally this uses HMAC-SHA256 as
7316 keyed hash function using the machine ID as input.)
7318 * NotifyAccess= gained a new supported value "exec". When set
7319 notifications are accepted from all processes systemd itself invoked,
7320 including all control processes.
7322 * .nspawn files gained support for defining overlay mounts using the
7323 Overlay= and OverlayReadOnly= options. Previously this functionality
7324 was only available on the systemd-nspawn command line.
7326 * systemd-nspawn's --bind= and --overlay= options gained support for
7327 bind/overlay mounts whose source lies within the container tree by
7328 prefixing the source path with "+".
7330 * systemd-nspawn's --bind= and --overlay= options gained support for
7331 automatically allocating a temporary source directory in /var/tmp
7332 that is removed when the container dies. Specifically, if the source
7333 directory is specified as empty string this mechanism is selected. An
7334 example usage is --overlay=+/var::/var, which creates an overlay
7335 mount based on the original /var contained in the image, overlaid
7336 with a temporary directory in the host's /var/tmp. This way changes
7337 to /var are automatically flushed when the container shuts down.
7339 * systemd-nspawn --image= option does now permit raw file system block
7340 devices (in addition to images containing partition tables, as
7343 * The disk image dissection logic in systemd-nspawn gained support for
7344 automatically setting up LUKS encrypted as well as Verity protected
7345 partitions. When a container is booted from an encrypted image the
7346 passphrase is queried at start-up time. When a container with Verity
7347 data is started, the root hash is search in a ".roothash" file
7348 accompanying the disk image (alternatively, pass the root hash via
7349 the new --root-hash= command line option).
7351 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
7352 be used to dissect disk images the same way as systemd-nspawn does
7353 it, following the Bootable Partition Specification. It may even be
7354 used to mount disk images with complex partition setups (including
7355 LUKS and Verity partitions) to a local host directory, in order to
7356 inspect them. This tool is not considered public API (yet), and is
7357 thus not installed into /usr/bin. Please do not rely on its
7358 existence, since it might go away or be changed in later systemd
7361 * A new generator "systemd-verity-generator" has been added, similar in
7362 style to "systemd-cryptsetup-generator", permitting automatic setup of
7363 Verity root partitions when systemd boots up. In order to make use of
7364 this your partition setup should follow the Discoverable Partitions
7365 Specification, and the GPT partition ID of the root file system
7366 partition should be identical to the upper 128bit of the Verity root
7367 hash. The GPT partition ID of the Verity partition protecting it
7368 should be the lower 128bit of the Verity root hash. If the partition
7369 image follows this model it is sufficient to specify a single
7370 "roothash=" kernel command line argument to both configure which root
7371 image and verity partition to use as well as the root hash for
7372 it. Note that systemd-nspawn's Verity support follows the same
7373 semantics, meaning that disk images with proper Verity data in place
7374 may be booted in containers with systemd-nspawn as well as on
7375 physical systems via the verity generator. Also note that the "mkosi"
7376 tool available at https://github.com/systemd/mkosi has been updated
7377 to generate Verity protected disk images following this scheme. In
7378 fact, it has been updated to generate disk images that optionally
7379 implement a complete UEFI SecureBoot trust chain, involving a signed
7380 kernel and initrd image that incorporates such a root hash as well as
7381 a Verity-enabled root partition.
7383 * The hardware database (hwdb) udev supports has been updated to carry
7384 accelerometer quirks.
7386 * All system services are now run with a fresh kernel keyring set up
7387 for them. The invocation ID is stored by default in it, thus
7388 providing a safe, non-overridable way to determine the invocation
7391 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
7392 options for bind mounting arbitrary paths in a service-specific
7393 way. When these options are used, arbitrary host or service files and
7394 directories may be mounted to arbitrary locations in the service's
7397 * Documentation has been added that lists all of systemd's low-level
7398 environment variables:
7400 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
7402 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
7403 whether a specific socket file descriptor matches a specified socket
7406 * systemd-firstboot has been updated to check for the
7407 systemd.firstboot= kernel command line option. It accepts a boolean
7408 and when set to false the first boot questions are skipped.
7410 * systemd-fstab-generator has been updated to check for the
7411 systemd.volatile= kernel command line option, which either takes an
7412 optional boolean parameter or the special value "state". If used the
7413 system may be booted in a "volatile" boot mode. Specifically,
7414 "systemd.volatile" is used, the root directory will be mounted as
7415 tmpfs, and only /usr is mounted from the actual root file system. If
7416 "systemd.volatile=state" is used, the root directory will be mounted
7417 as usual, but /var is mounted as tmpfs. This concept provides similar
7418 functionality as systemd-nspawn's --volatile= option, but provides it
7419 on physical boots. Use this option for implementing stateless
7420 systems, or testing systems with all state and/or configuration reset
7421 to the defaults. (Note though that many distributions are not
7422 prepared to boot up without a populated /etc or /var, though.)
7424 * systemd-gpt-auto-generator gained support for LUKS encrypted root
7425 partitions. Previously it only supported LUKS encrypted partitions
7426 for all other uses, except for the root partition itself.
7428 * Socket units gained support for listening on AF_VSOCK sockets for
7429 communication in virtualized QEMU environments.
7431 * The "configure" script gained a new option --with-fallback-hostname=
7432 for specifying the fallback hostname to use if none is configured in
7433 /etc/hostname. For example, by specifying
7434 --with-fallback-hostname=fedora it is possible to default to a
7435 hostname of "fedora" on pristine installations.
7437 * systemd-cgls gained support for a new --unit= switch for listing only
7438 the control groups of a specific unit. Similar --user-unit= has been
7439 added for listing only the control groups of a specific user unit.
7441 * systemd-mount gained a new --umount switch for unmounting a mount or
7442 automount point (and all mount/automount points below it).
7444 * systemd will now refuse full configuration reloads (via systemctl
7445 daemon-reload and related calls) unless at least 16MiB of free space
7446 are available in /run. This is a safety precaution in order to ensure
7447 that generators can safely operate after the reload completed.
7449 * A new unit file option RootImage= has been added, which has a similar
7450 effect as RootDirectory= but mounts the service's root directory from
7451 a disk image instead of plain directory. This logic reuses the same
7452 image dissection and mount logic that systemd-nspawn already uses,
7453 and hence supports any disk images systemd-nspawn supports, including
7454 those following the Discoverable Partition Specification, as well as
7455 Verity enabled images. This option enables systemd to run system
7456 services directly off disk images acting as resource bundles,
7457 possibly even including full integrity data.
7459 * A new MountAPIVFS= unit file option has been added, taking a boolean
7460 argument. If enabled /proc, /sys and /dev (collectively called the
7461 "API VFS") will be mounted for the service. This is only relevant if
7462 RootDirectory= or RootImage= is used for the service, as these mounts
7463 are of course in place in the host mount namespace anyway.
7465 * systemd-nspawn gained support for a new --pivot-root= switch. If
7466 specified the root directory within the container image is pivoted to
7467 the specified mount point, while the original root disk is moved to a
7468 different place. This option enables booting of ostree images
7469 directly with systemd-nspawn.
7471 * The systemd build scripts will no longer complain if the NTP server
7472 addresses are not changed from the defaults. Google now supports
7473 these NTP servers officially. We still recommend downstreams to
7474 properly register an NTP pool with the NTP pool project though.
7476 * coredumpctl gained a new "--reverse" option for printing the list
7477 of coredumps in reverse order.
7479 * coredumpctl will now show additional information about truncated and
7480 inaccessible coredumps, as well as coredumps that are still being
7481 processed. It also gained a new --quiet switch for suppressing
7482 additional informational message in its output.
7484 * coredumpctl gained support for only showing coredumps newer and/or
7485 older than specific timestamps, using the new --since= and --until=
7486 options, reminiscent of journalctl's options by the same name.
7488 * The systemd-coredump logic has been improved so that it may be reused
7489 to collect backtraces in non-compiled languages, for example in
7490 scripting languages such as Python.
7492 * machinectl will now show the UID shift of local containers, if user
7493 namespacing is enabled for them.
7495 * systemd will now optionally run "environment generator" binaries at
7496 configuration load time. They may be used to add environment
7497 variables to the environment block passed to services invoked. One
7498 user environment generator is shipped by default that sets up
7499 environment variables based on files dropped into /etc/environment.d
7500 and ~/.config/environment.d/.
7502 * systemd-resolved now includes the new, recently published 2017 DNSSEC
7505 * hostnamed has been updated to report a new chassis type of
7506 "convertible" to cover "foldable" laptops that can both act as a
7507 tablet and as a laptop, such as various Lenovo Yoga devices.
7509 Contributions from: Adrián López, Alexander Galanin, Alexander
7510 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
7511 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
7512 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
7513 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
7514 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
7515 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
7516 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
7517 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
7518 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
7519 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
7520 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
7521 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
7522 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
7523 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
7524 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
7525 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
7526 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
7527 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
7528 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
7529 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
7530 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
7531 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
7532 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
7533 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
7534 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
7535 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
7536 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
7539 — Berlin, 2017-03-01
7543 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
7544 RestrictAddressFamilies= enabled. These sandboxing options should
7545 generally be compatible with the various external udev call-out
7546 binaries we are aware of, however there may be exceptions, in
7547 particular when exotic languages for these call-outs are used. In
7548 this case, consider turning off these settings locally.
7550 * The new RemoveIPC= option can be used to remove IPC objects owned by
7551 the user or group of a service when that service exits.
7553 * The new ProtectKernelModules= option can be used to disable explicit
7554 load and unload operations of kernel modules by a service. In
7555 addition access to /usr/lib/modules is removed if this option is set.
7557 * ProtectSystem= option gained a new value "strict", which causes the
7558 whole file system tree with the exception of /dev, /proc, and /sys,
7559 to be remounted read-only for a service.
7561 * The new ProtectKernelTunables= option can be used to disable
7562 modification of configuration files in /sys and /proc by a service.
7563 Various directories and files are remounted read-only, so access is
7564 restricted even if the file permissions would allow it.
7566 * The new ProtectControlGroups= option can be used to disable write
7567 access by a service to /sys/fs/cgroup.
7569 * Various systemd services have been hardened with
7570 ProtectKernelTunables=yes, ProtectControlGroups=yes,
7571 RestrictAddressFamilies=.
7573 * Support for dynamically creating users for the lifetime of a service
7574 has been added. If DynamicUser=yes is specified, user and group IDs
7575 will be allocated from the range 61184…65519 for the lifetime of the
7576 service. They can be resolved using the new nss-systemd.so NSS
7577 module. The module must be enabled in /etc/nsswitch.conf. Services
7578 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
7579 any resources allocated by the service will be cleaned up when the
7580 service exits. They also have ProtectHome=read-only and
7581 ProtectSystem=strict enabled, so they are not able to make any
7582 permanent modifications to the system.
7584 * The nss-systemd module also always resolves root and nobody, making
7585 it possible to have no /etc/passwd or /etc/group files in minimal
7586 container or chroot environments.
7588 * Services may be started with their own user namespace using the new
7589 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
7590 under which the service is running are mapped. All other users are
7593 * Support for the cgroup namespace has been added to systemd-nspawn. If
7594 supported by kernel, the container system started by systemd-nspawn
7595 will have its own view of the cgroup hierarchy. This new behaviour
7596 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
7598 * The new MemorySwapMax= option can be used to limit the maximum swap
7599 usage under the unified cgroup hierarchy.
7601 * Support for the CPU controller in the unified cgroup hierarchy has
7602 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
7603 options. This controller requires out-of-tree patches for the kernel
7604 and the support is provisional.
7606 * Mount and automount units may now be created transiently
7607 (i.e. dynamically at runtime via the bus API, instead of requiring
7608 unit files in the file system).
7610 * systemd-mount is a new tool which may mount file systems – much like
7611 mount(8), optionally pulling in additional dependencies through
7612 transient .mount and .automount units. For example, this tool
7613 automatically runs fsck on a backing block device before mounting,
7614 and allows the automount logic to be used dynamically from the
7615 command line for establishing mount points. This tool is particularly
7616 useful when dealing with removable media, as it will ensure fsck is
7617 run – if necessary – before the first access and that the file system
7618 is quickly unmounted after each access by utilizing the automount
7619 logic. This maximizes the chance that the file system on the
7620 removable media stays in a clean state, and if it isn't in a clean
7621 state is fixed automatically.
7623 * LazyUnmount=yes option for mount units has been added to expose the
7624 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
7627 * /efi will be used as the mount point of the EFI boot partition, if
7628 the directory is present, and the mount point was not configured
7629 through other means (e.g. fstab). If /efi directory does not exist,
7630 /boot will be used as before. This makes it easier to automatically
7631 mount the EFI partition on systems where /boot is used for something
7634 * When operating on GPT disk images for containers, systemd-nspawn will
7635 now mount the ESP to /boot or /efi according to the same rules as PID
7636 1 running on a host. This allows tools like "bootctl" to operate
7637 correctly within such containers, in order to make container images
7638 bootable on physical systems.
7640 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
7642 * Two new user session targets have been added to support running
7643 graphical sessions under the systemd --user instance:
7644 graphical-session.target and graphical-session-pre.target. See
7645 systemd.special(7) for a description of how those targets should be
7648 * The vconsole initialization code has been significantly reworked to
7649 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
7650 support unicode keymaps. Font and keymap configuration will now be
7651 copied to all allocated virtual consoles.
7653 * FreeBSD's bhyve virtualization is now detected.
7655 * Information recorded in the journal for core dumps now includes the
7656 contents of /proc/mountinfo and the command line of the process at
7657 the top of the process hierarchy (which is usually the init process
7660 * systemd-journal-gatewayd learned the --directory= option to serve
7661 files from the specified location.
7663 * journalctl --root=… can be used to peruse the journal in the
7664 /var/log/ directories inside of a container tree. This is similar to
7665 the existing --machine= option, but does not require the container to
7668 * The hardware database has been extended to support
7669 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
7672 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
7673 specify the click rate for mice which include a horizontal wheel with
7674 a click rate that is different than the one for the vertical wheel.
7676 * systemd-run gained a new --wait option that makes service execution
7677 synchronous. (Specifically, the command will not return until the
7678 specified service binary exited.)
7680 * systemctl gained a new --wait option that causes the start command to
7681 wait until the units being started have terminated again.
7683 * A new journal output mode "short-full" has been added which displays
7684 timestamps with abbreviated English day names and adds a timezone
7685 suffix. Those timestamps include more information than the default
7686 "short" output mode, and can be passed directly to journalctl's
7687 --since= and --until= options.
7689 * /etc/resolv.conf will be bind-mounted into containers started by
7690 systemd-nspawn, if possible, so any changes to resolv.conf contents
7691 are automatically propagated to the container.
7693 * The number of instances for socket-activated services originating
7694 from a single IP address can be limited with
7695 MaxConnectionsPerSource=, extending the existing setting of
7698 * systemd-networkd gained support for vcan ("Virtual CAN") interface
7701 * .netdev and .network configuration can now be extended through
7704 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
7705 Segmentation Offload, Generic Receive Offload, Large Receive Offload
7706 can be enabled and disabled using the new UDPSegmentationOffload=,
7707 TCPSegmentationOffload=, GenericSegmentationOffload=,
7708 GenericReceiveOffload=, LargeReceiveOffload= options in the
7709 [Link] section of .link files.
7711 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
7712 Port VLAN ID can be configured for bridge devices using the new STP=,
7713 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
7714 section of .netdev files.
7716 * The route table to which routes received over DHCP or RA should be
7717 added can be configured with the new RouteTable= option in the [DHCP]
7718 and [IPv6AcceptRA] sections of .network files.
7720 * The Address Resolution Protocol can be disabled on links managed by
7721 systemd-networkd using the ARP=no setting in the [Link] section of
7724 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
7725 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
7726 encode information about the result and exit codes of the current
7727 service runtime cycle.
7729 * systemd-sysctl will now configure kernel parameters in the order
7730 they occur in the configuration files. This matches what sysctl
7731 has been traditionally doing.
7733 * kernel-install "plugins" that are executed to perform various
7734 tasks after a new kernel is added and before an old one is removed
7735 can now return a special value to terminate the procedure and
7736 prevent any later plugins from running.
7738 * Journald's SplitMode=login setting has been deprecated. It has been
7739 removed from documentation, and its use is discouraged. In a future
7740 release it will be completely removed, and made equivalent to current
7741 default of SplitMode=uid.
7743 * Storage=both option setting in /etc/systemd/coredump.conf has been
7744 removed. With fast LZ4 compression storing the core dump twice is not
7747 * The --share-system systemd-nspawn option has been replaced with an
7748 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
7749 this functionality is discouraged. In addition the variables
7750 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
7751 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
7752 individual namespaces.
7754 * "machinectl list" now shows the IP address of running containers in
7755 the output, as well as OS release information.
7757 * "loginctl list" now shows the TTY of each session in the output.
7759 * sd-bus gained new API calls sd_bus_track_set_recursive(),
7760 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
7761 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
7762 tracking objects in a "recursive" mode, where a single client can be
7763 counted multiple times, if it takes multiple references.
7765 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
7766 sd_bus_get_exit_on_disconnect(). They may be used to make a
7767 process using sd-bus automatically exit if the bus connection is
7770 * Bus clients of the service manager may now "pin" loaded units into
7771 memory, by taking an explicit reference on them. This is useful to
7772 ensure the client can retrieve runtime data about the service even
7773 after the service completed execution. Taking such a reference is
7774 available only for privileged clients and should be helpful to watch
7775 running services in a race-free manner, and in particular collect
7776 information about exit statuses and results.
7778 * The nss-resolve module has been changed to strictly return UNAVAIL
7779 when communication via D-Bus with resolved failed, and NOTFOUND when
7780 a lookup completed but was negative. This means it is now possible to
7781 neatly configure fallbacks using nsswitch.conf result checking
7782 expressions. Taking benefit of this, the new recommended
7783 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
7785 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
7787 * A new setting CtrlAltDelBurstAction= has been added to
7788 /etc/systemd/system.conf which may be used to configure the precise
7789 behaviour if the user on the console presses Ctrl-Alt-Del more often
7790 than 7 times in 2s. Previously this would unconditionally result in
7791 an expedited, immediate reboot. With this new setting the precise
7792 operation may be configured in more detail, and also turned off
7795 * In .netdev files two new settings RemoteChecksumTx= and
7796 RemoteChecksumRx= are now understood that permit configuring the
7797 remote checksumming logic for VXLAN networks.
7799 * The service manager learnt a new "invocation ID" concept for invoked
7800 services. Each runtime cycle of a service will get a new invocation
7801 ID (a 128bit random UUID) assigned that identifies the current
7802 run of the service uniquely and globally. A new invocation ID
7803 is generated each time a service starts up. The journal will store
7804 the invocation ID of a service along with any logged messages, thus
7805 making the invocation ID useful for matching the online runtime of a
7806 service with the offline log data it generated in a safe way without
7807 relying on synchronized timestamps. In many ways this new service
7808 invocation ID concept is similar to the kernel's boot ID concept that
7809 uniquely and globally identifies the runtime of each boot. The
7810 invocation ID of a service is passed to the service itself via an
7811 environment variable ($INVOCATION_ID). A new bus call
7812 GetUnitByInvocationID() has been added that is similar to GetUnit()
7813 but instead of retrieving the bus path for a unit by its name
7814 retrieves it by its invocation ID. The returned path is valid only as
7815 long as the passed invocation ID is current.
7817 * systemd-resolved gained a new "DNSStubListener" setting in
7818 resolved.conf. It either takes a boolean value or the special values
7819 "udp" and "tcp", and configures whether to enable the stub DNS
7820 listener on 127.0.0.53:53.
7822 * IP addresses configured via networkd may now carry additional
7823 configuration settings supported by the kernel. New options include:
7824 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
7825 PrefixRoute=, AutoJoin=.
7827 * The PAM configuration fragment file for "user@.service" shipped with
7828 systemd (i.e. the --user instance of systemd) has been stripped to
7829 the minimum necessary to make the system boot. Previously, it
7830 contained Fedora-specific stanzas that did not apply to other
7831 distributions. It is expected that downstream distributions add
7832 additional configuration lines, matching their needs to this file,
7833 using it only as rough template of what systemd itself needs. Note
7834 that this reduced fragment does not even include an invocation of
7835 pam_limits which most distributions probably want to add, even though
7836 systemd itself does not need it. (There's also the new build time
7837 option --with-pamconfdir=no to disable installation of the PAM
7840 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
7841 capability is now also dropped from its set (in addition to
7842 CAP_SYS_MKNOD as before).
7844 * In service unit files it is now possible to connect a specific named
7845 file descriptor with stdin/stdout/stdout of an executed service. The
7846 name may be specified in matching .socket units using the
7847 FileDescriptorName= setting.
7849 * A number of journal settings may now be configured on the kernel
7850 command line. Specifically, the following options are now understood:
7851 systemd.journald.max_level_console=,
7852 systemd.journald.max_level_store=,
7853 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
7854 systemd.journald.max_level_wall=.
7856 * "systemctl is-enabled --full" will now show by which symlinks a unit
7857 file is enabled in the unit dependency tree.
7859 * Support for VeraCrypt encrypted partitions has been added to the
7860 "cryptsetup" logic and /etc/crypttab.
7862 * systemd-detect-virt gained support for a new --private-users switch
7863 that checks whether the invoking processes are running inside a user
7864 namespace. Similar, a new special value "private-users" for the
7865 existing ConditionVirtualization= setting has been added, permitting
7866 skipping of specific units in user namespace environments.
7868 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
7869 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
7870 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
7871 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
7872 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
7873 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
7874 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
7875 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
7876 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
7877 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
7878 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
7879 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
7880 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
7881 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
7882 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
7883 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
7884 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
7885 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
7886 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
7887 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
7888 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
7889 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
7890 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
7891 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
7892 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
7893 Jędrzejewski-Szmek, Zeal Jagannatha
7895 — Santa Fe, 2016-11-03
7899 * In service units the various ExecXYZ= settings have been extended
7900 with an additional special character as first argument of the
7901 assigned value: if the character '+' is used the specified command
7902 line it will be run with full privileges, regardless of User=,
7903 Group=, CapabilityBoundingSet= and similar options. The effect is
7904 similar to the existing PermissionsStartOnly= option, but allows
7905 configuration of this concept for each executed command line
7908 * Services may now alter the service watchdog timeout at runtime by
7909 sending a WATCHDOG_USEC= message via sd_notify().
7911 * MemoryLimit= and related unit settings now optionally take percentage
7912 specifications. The percentage is taken relative to the amount of
7913 physical memory in the system (or in case of containers, the assigned
7914 amount of memory). This allows scaling service resources neatly with
7915 the amount of RAM available on the system. Similarly, systemd-logind's
7916 RuntimeDirectorySize= option now also optionally takes percentage
7919 * In similar fashion TasksMax= takes percentage values now, too. The
7920 value is taken relative to the configured maximum number of processes
7921 on the system. The per-service task maximum has been changed to 15%
7922 using this functionality. (Effectively this is an increase of 512 →
7923 4915 for service units, given the kernel's default pid_max setting.)
7925 * Calendar time specifications in .timer units now understand a ".."
7926 syntax for time ranges. Example: "4..7:10" may now be used for
7927 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
7930 * The InaccessableDirectories=, ReadOnlyDirectories= and
7931 ReadWriteDirectories= unit file settings have been renamed to
7932 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
7933 applied to all kinds of file nodes, and not just directories, with
7934 the exception of symlinks. Specifically these settings may now be
7935 used on block and character device nodes, UNIX sockets and FIFOS as
7936 well as regular files. The old names of these settings remain
7937 available for compatibility.
7939 * systemd will now log about all service processes it kills forcibly
7940 (using SIGKILL) because they remained after the clean shutdown phase
7941 of the service completed. This should help identifying services that
7942 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
7943 systemd-logind's configuration a similar log message is generated for
7944 processes killed at the end of each session due to this setting.
7946 * systemd will now set the $JOURNAL_STREAM environment variable for all
7947 services whose stdout/stderr are connected to the Journal (which
7948 effectively means by default: all services). The variable contains
7949 the device and inode number of the file descriptor used for
7950 stdout/stderr. This may be used by invoked programs to detect whether
7951 their stdout/stderr is connected to the Journal, in which case they
7952 can switch over to direct Journal communication, thus being able to
7953 pass extended, structured metadata along with their log messages. As
7954 one example, this is now used by glib's logging primitives.
7956 * When using systemd's default tmp.mount unit for /tmp, the mount point
7957 will now be established with the "nosuid" and "nodev" options. This
7958 avoids privilege escalation attacks that put traps and exploits into
7959 /tmp. However, this might cause problems if you e. g. put container
7960 images or overlays into /tmp; if you need this, override tmp.mount's
7961 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
7964 * systemd now supports the "memory" cgroup controller also on
7967 * The systemd-cgtop tool now optionally takes a control group path as
7968 command line argument. If specified, the control group list shown is
7969 limited to subgroups of that group.
7971 * The SystemCallFilter= unit file setting gained support for
7972 pre-defined, named system call filter sets. For example
7973 SystemCallFilter=@clock is now an effective way to make all clock
7974 changing-related system calls unavailable to a service. A number of
7975 similar pre-defined groups are defined. Writing system call filters
7976 for system services is simplified substantially with this new
7977 concept. Accordingly, all of systemd's own, long-running services now
7978 enable system call filtering based on this, by default.
7980 * A new service setting MemoryDenyWriteExecute= has been added, taking
7981 a boolean value. If turned on, a service may no longer create memory
7982 mappings that are writable and executable at the same time. This
7983 enhances security for services where this is enabled as it becomes
7984 harder to dynamically write and then execute memory in exploited
7985 service processes. This option has been enabled for all of systemd's
7986 own long-running services.
7988 * A new RestrictRealtime= service setting has been added, taking a
7989 boolean argument. If set the service's processes may no longer
7990 acquire realtime scheduling. This improves security as realtime
7991 scheduling may otherwise be used to easily freeze the system.
7993 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
7994 value. This may be used for requesting that the system manager inside
7995 of the container reports start-up completion to nspawn which then
7996 propagates this notification further to the service manager
7997 supervising nspawn itself. A related option NotifyReady= in .nspawn
7998 files has been added too. This functionality allows ordering of the
7999 start-up of multiple containers using the usual systemd ordering
8002 * machinectl gained a new command "stop" that is an alias for
8005 * systemd-resolved gained support for contacting DNS servers on
8006 link-local IPv6 addresses.
8008 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
8009 its caches. A method call for requesting the same operation has been
8010 added to the bus API too, and is made available via "systemd-resolve
8013 * systemd-resolve gained a new --status switch. If passed a brief
8014 summary of the used DNS configuration with per-interface information
8017 * resolved.conf gained a new Cache= boolean option, defaulting to
8018 on. If turned off local DNS caching is disabled. This comes with a
8019 performance penalty in particular when DNSSEC is enabled. Note that
8020 resolved disables its internal caching implicitly anyway, when the
8021 configured DNS server is on a host-local IP address such as ::1 or
8022 127.0.0.1, thus automatically avoiding double local caching.
8024 * systemd-resolved now listens on the local IP address 127.0.0.53:53
8025 for DNS requests. This improves compatibility with local programs
8026 that do not use the libc NSS or systemd-resolved's bus APIs for name
8027 resolution. This minimal DNS service is only available to local
8028 programs and does not implement the full DNS protocol, but enough to
8029 cover local DNS clients. A new, static resolv.conf file, listing just
8030 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
8031 now recommended to make /etc/resolv.conf a symlink to this file in
8032 order to route all DNS lookups to systemd-resolved, regardless if
8033 done via NSS, the bus API or raw DNS packets. Note that this local
8034 DNS service is not as fully featured as the libc NSS or
8035 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
8036 used to deliver link-local address information (as this implies
8037 sending a local interface index along), LLMNR/mDNS support via this
8038 interface is severely restricted. It is thus strongly recommended for
8039 all applications to use the libc NSS API or native systemd-resolved
8042 * systemd-networkd's bridge support learned a new setting
8043 VLANFiltering= for controlling VLAN filtering. Moreover a new section
8044 in .network files has been added for configuring VLAN bridging in
8045 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
8047 * systemd-networkd's IPv6 Router Advertisement code now makes use of
8048 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
8049 now be acquired without relying on DHCPv6. Two new options
8050 UseDomains= and UseDNS= have been added to configure this behaviour.
8052 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
8053 renamed IPv6AcceptRA=, without altering its behaviour. The old
8054 setting name remains available for compatibility reasons.
8056 * The systemd-networkd VTI/VTI6 tunneling support gained new options
8057 Key=, InputKey= and OutputKey=.
8059 * systemd-networkd gained support for VRF ("Virtual Routing Function")
8060 interface configuration.
8062 * "systemctl edit" may now be used to create new unit files by
8063 specifying the --force switch.
8065 * sd-event gained a new function sd_event_get_iteration() for
8066 requesting the current iteration counter of the event loop. It starts
8067 at zero and is increased by one with each event loop iteration.
8069 * A new rpm macro %systemd_ordering is provided by the macros.systemd
8070 file. It can be used in lieu of %systemd_requires in packages which
8071 don't use any systemd functionality and are intended to be installed
8072 in minimal containers without systemd present. This macro provides
8073 ordering dependencies to ensure that if the package is installed in
8074 the same rpm transaction as systemd, systemd will be installed before
8075 the scriptlets for the package are executed, allowing unit presets
8078 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
8079 been added to simplify packaging of generators.
8081 * The os-release file gained VERSION_CODENAME field for the
8082 distribution nickname (e.g. VERSION_CODENAME=woody).
8084 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
8085 can be set to disable parsing of metadata and the creation
8086 of persistent symlinks for that device.
8088 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
8089 to make them available to logged-in users has been reverted.
8091 * Much of the common code of the various systemd components is now
8092 built into an internal shared library libsystemd-shared-231.so
8093 (incorporating the systemd version number in the name, to be updated
8094 with future releases) that the components link to. This should
8095 decrease systemd footprint both in memory during runtime and on
8096 disk. Note that the shared library is not for public use, and is
8097 neither API nor ABI stable, but is likely to change with every new
8098 released update. Packagers need to make sure that binaries
8099 linking to libsystemd-shared.so are updated in step with the
8102 * Configuration for "mkosi" is now part of the systemd
8103 repository. mkosi is a tool to easily build legacy-free OS images,
8104 and is available on github: https://github.com/systemd/mkosi. If
8105 "mkosi" is invoked in the build tree a new raw OS image is generated
8106 incorporating the systemd sources currently being worked on and a
8107 clean, fresh distribution installation. The generated OS image may be
8108 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
8109 UEFI PC. This functionality is particularly useful to easily test
8110 local changes made to systemd in a pristine, defined environment. See
8111 doc/HACKING for details.
8113 * configure learned the --with-support-url= option to specify the
8114 distribution's bugtracker.
8116 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
8117 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
8118 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
8119 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
8120 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
8121 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
8122 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
8123 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
8124 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
8125 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
8126 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
8127 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
8128 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
8129 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
8130 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
8131 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
8132 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
8133 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
8134 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
8136 — Berlin, 2016-07-25
8140 * DNSSEC is now turned on by default in systemd-resolved (in
8141 "allow-downgrade" mode), but may be turned off during compile time by
8142 passing "--with-default-dnssec=no" to "configure" (and of course,
8143 during runtime with DNSSEC= in resolved.conf). We recommend
8144 downstreams to leave this on at least during development cycles and
8145 report any issues with the DNSSEC logic upstream. We are very
8146 interested in collecting feedback about the DNSSEC validator and its
8147 limitations in the wild. Note however, that DNSSEC support is
8148 probably nothing downstreams should turn on in stable distros just
8149 yet, as it might create incompatibilities with a few DNS servers and
8150 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
8151 automatically whenever we detect such incompatible setups, but there
8152 might be systems we do not cover yet. Hence: please help us testing
8153 the DNSSEC code, leave this on where you can, report back, but then
8154 again don't consider turning this on in your stable, LTS or
8155 production release just yet. (Note that you have to enable
8156 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
8157 and its DNSSEC mode for hostname resolution from local
8160 * systemd-resolve conveniently resolves DANE records with the --tlsa
8161 option and OPENPGPKEY records with the --openpgp option. It also
8162 supports dumping raw DNS record data via the new --raw= switch.
8164 * systemd-logind will now by default terminate user processes that are
8165 part of the user session scope unit (session-XX.scope) when the user
8166 logs out. This behavior is controlled by the KillUserProcesses=
8167 setting in logind.conf, and the previous default of "no" is now
8168 changed to "yes". This means that user sessions will be properly
8169 cleaned up after, but additional steps are necessary to allow
8170 intentionally long-running processes to survive logout.
8172 While the user is logged in at least once, user@.service is running,
8173 and any service that should survive the end of any individual login
8174 session can be started at a user service or scope using systemd-run.
8175 systemd-run(1) man page has been extended with an example which shows
8176 how to run screen in a scope unit underneath user@.service. The same
8177 command works for tmux.
8179 After the user logs out of all sessions, user@.service will be
8180 terminated too, by default, unless the user has "lingering" enabled.
8181 To effectively allow users to run long-term tasks even if they are
8182 logged out, lingering must be enabled for them. See loginctl(1) for
8183 details. The default polkit policy was modified to allow users to
8184 set lingering for themselves without authentication.
8186 Previous defaults can be restored at compile time by the
8187 --without-kill-user-processes option to "configure".
8189 * systemd-logind gained new configuration settings SessionsMax= and
8190 InhibitorsMax=, both with a default of 8192. It will not register new
8191 user sessions or inhibitors above this limit.
8193 * systemd-logind will now reload configuration on SIGHUP.
8195 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
8196 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
8197 enable. Also, support for the "io" cgroup controller in the unified
8198 hierarchy has been added, so that the "memory", "pids" and "io" are
8199 now the controllers that are supported on the unified hierarchy.
8201 WARNING: it is not possible to use previous systemd versions with
8202 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
8203 is necessary to also update systemd in the initramfs if using the
8204 unified hierarchy. An updated SELinux policy is also required.
8206 * LLDP support has been extended, and both passive (receive-only) and
8207 active (sender) modes are supported. Passive mode ("routers-only") is
8208 enabled by default in systemd-networkd. Active LLDP mode is enabled
8209 by default for containers on the internal network. The "networkctl
8210 lldp" command may be used to list information gathered. "networkctl
8211 status" will also show basic LLDP information on connected peers now.
8213 * The IAID and DUID unique identifier sent in DHCP requests may now be
8214 configured for the system and each .network file managed by
8215 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
8217 * systemd-networkd gained support for configuring proxy ARP support for
8218 each interface, via the ProxyArp= setting in .network files. It also
8219 gained support for configuring the multicast querier feature of
8220 bridge devices, via the new MulticastQuerier= setting in .netdev
8221 files. Similarly, snooping on the IGMP traffic can be controlled
8222 via the new setting MulticastSnooping=.
8224 A new setting PreferredLifetime= has been added for addresses
8225 configured in .network file to configure the lifetime intended for an
8228 The systemd-networkd DHCP server gained the option EmitRouter=, which
8229 defaults to yes, to configure whether the DHCP Option 3 (Router)
8232 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
8233 systemd-socket-activate and installed into /usr/bin. It is now fully
8236 * systemd-journald now uses separate threads to flush changes to disk
8237 when closing journal files, thus reducing impact of slow disk I/O on
8238 logging performance.
8240 * The sd-journal API gained two new calls
8241 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
8242 can be used to open journal files using file descriptors instead of
8243 file or directory paths. sd_journal_open_container() has been
8244 deprecated, sd_journal_open_directory_fd() should be used instead
8245 with the flag SD_JOURNAL_OS_ROOT.
8247 * journalctl learned a new output mode "-o short-unix" that outputs log
8248 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
8249 UTC). It also gained support for a new --no-hostname setting to
8250 suppress the hostname column in the family of "short" output modes.
8252 * systemd-ask-password now optionally skips printing of the password to
8253 stdout with --no-output which can be useful in scripts.
8255 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
8256 (devices tagged with ID_MAKER_TOOL) are now tagged with
8257 "uaccess" and are available to logged in users.
8259 * The DeviceAllow= unit setting now supports specifiers (with "%").
8261 * "systemctl show" gained a new --value switch, which allows print a
8262 only the contents of a specific unit property, without also printing
8263 the property's name. Similar support was added to "show*" verbs
8264 of loginctl and machinectl that output "key=value" lists.
8266 * A new unit type "generated" was added for files dynamically generated
8267 by generator tools. Similarly, a new unit type "transient" is used
8268 for unit files created using the runtime API. "systemctl enable" will
8269 refuse to operate on such files.
8271 * A new command "systemctl revert" has been added that may be used to
8272 revert to the vendor version of a unit file, in case local changes
8273 have been made by adding drop-ins or overriding the unit file.
8275 * "machinectl clean" gained a new verb to automatically remove all or
8276 just hidden container images.
8278 * systemd-tmpfiles gained support for a new line type "e" for emptying
8279 directories, if they exist, without creating them if they don't.
8281 * systemd-nspawn gained support for automatically patching the UID/GIDs
8282 of the owners and the ACLs of all files and directories in a
8283 container tree to match the UID/GID user namespacing range selected
8284 for the container invocation. This mode is enabled via the new
8285 --private-users-chown switch. It also gained support for
8286 automatically choosing a free, previously unused UID/GID range when
8287 starting a container, via the new --private-users=pick setting (which
8288 implies --private-users-chown). Together, these options for the first
8289 time make user namespacing for nspawn containers fully automatic and
8290 thus deployable. The systemd-nspawn@.service template unit file has
8291 been changed to use this functionality by default.
8293 * systemd-nspawn gained a new --network-zone= switch, that allows
8294 creating ad-hoc virtual Ethernet links between multiple containers,
8295 that only exist as long as at least one container referencing them is
8296 running. This allows easy connecting of multiple containers with a
8297 common link that implements an Ethernet broadcast domain. Each of
8298 these network "zones" may be named relatively freely by the user, and
8299 may be referenced by any number of containers, but each container may
8300 only reference one of these "zones". On the lower level, this is
8301 implemented by an automatically managed bridge network interface for
8302 each zone, that is created when the first container referencing its
8303 zone is created and removed when the last one referencing its zone
8306 * The default start timeout may now be configured on the kernel command
8307 line via systemd.default_timeout_start_sec=. It was already
8308 configurable via the DefaultTimeoutStartSec= option in
8309 /etc/systemd/system.conf.
8311 * Socket units gained a new TriggerLimitIntervalSec= and
8312 TriggerLimitBurst= setting to configure a limit on the activation
8313 rate of the socket unit.
8315 * The LimitNICE= setting now optionally takes normal UNIX nice values
8316 in addition to the raw integer limit value. If the specified
8317 parameter is prefixed with "+" or "-" and is in the range -20…19 the
8318 value is understood as UNIX nice value. If not prefixed like this it
8319 is understood as raw RLIMIT_NICE limit.
8321 * Note that the effect of the PrivateDevices= unit file setting changed
8322 slightly with this release: the per-device /dev file system will be
8323 mounted read-only from this version on, and will have "noexec"
8324 set. This (minor) change of behavior might cause some (exceptional)
8325 legacy software to break, when PrivateDevices=yes is set for its
8326 service. Please leave PrivateDevices= off if you run into problems
8329 * systemd-bootchart has been split out to a separate repository:
8330 https://github.com/systemd/systemd-bootchart
8332 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
8333 merged into the kernel in its current form.
8335 * The compatibility libraries libsystemd-daemon.so,
8336 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
8337 which have been deprecated since systemd-209 have been removed along
8338 with the corresponding pkg-config files. All symbols provided by
8339 those libraries are provided by libsystemd.so.
8341 * The Capabilities= unit file setting has been removed (it is ignored
8342 for backwards compatibility). AmbientCapabilities= and
8343 CapabilityBoundingSet= should be used instead.
8345 * A new special target has been added, initrd-root-device.target,
8346 which creates a synchronization point for dependencies of the root
8347 device in early userspace. Initramfs builders must ensure that this
8348 target is now included in early userspace.
8350 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
8351 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
8352 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
8353 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
8354 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
8355 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
8356 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
8357 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
8358 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
8359 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
8360 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
8361 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
8362 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
8363 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
8364 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
8365 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
8366 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
8367 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
8368 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
8369 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
8370 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
8371 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
8372 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
8373 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
8374 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
8377 — Fairfax, 2016-05-21
8381 * The systemd-resolved DNS resolver service has gained a substantial
8382 set of new features, most prominently it may now act as a DNSSEC
8383 validating stub resolver. DNSSEC mode is currently turned off by
8384 default, but is expected to be turned on by default in one of the
8385 next releases. For now, we invite everybody to test the DNSSEC logic
8386 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
8387 service also gained a full set of D-Bus interfaces, including calls
8388 to configure DNS and DNSSEC settings per link (for use by external
8389 network management software). systemd-resolved and systemd-networkd
8390 now distinguish between "search" and "routing" domains. The former
8391 are used to qualify single-label names, the latter are used purely
8392 for routing lookups within certain domains to specific links.
8393 resolved now also synthesizes RRs for all entries from /etc/hosts.
8395 * The systemd-resolve tool (which is a client utility for
8396 systemd-resolved) has been improved considerably and is now fully
8397 supported and documented. Hence it has moved from /usr/lib/systemd to
8400 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
8403 * The coredump collection logic has been reworked: when a coredump is
8404 collected it is now written to disk, compressed and processed
8405 (including stacktrace extraction) from a new instantiated service
8406 systemd-coredump@.service, instead of directly from the
8407 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
8408 processing large coredumps can take up a substantial amount of
8409 resources and time, and this previously happened entirely outside of
8410 systemd's service supervision. With the new logic the core_pattern
8411 hook only does minimal metadata collection before passing off control
8412 to the new instantiated service, which is configured with a time
8413 limit, a nice level and other settings to minimize negative impact on
8414 the rest of the system. Also note that the new logic will honour the
8415 RLIMIT_CORE setting of the crashed process, which now allows users
8416 and processes to turn off coredumping for their processes by setting
8419 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
8420 and all forked processes by default. Previously, PID 1 would leave
8421 the setting at "0" for all processes, as set by the kernel. Note that
8422 the resource limit traditionally has no effect on the generated
8423 coredumps on the system if the /proc/sys/kernel/core_pattern hook
8424 logic is used. Since the limit is now honoured (see above) its
8425 default has been changed so that the coredumping logic is enabled by
8426 default for all processes, while allowing specific opt-out.
8428 * When the stacktrace is extracted from processes of system users, this
8429 is now done as "systemd-coredump" user, in order to sandbox this
8430 potentially security sensitive parsing operation. (Note that when
8431 processing coredumps of normal users this is done under the user ID
8432 of process that crashed, as before.) Packagers should take notice
8433 that it is now necessary to create the "systemd-coredump" system user
8434 and group at package installation time.
8436 * The systemd-activate socket activation testing tool gained support
8437 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
8438 and --seqpacket switches. It also has been extended to support both
8439 new-style and inetd-style file descriptor passing. Use the new
8440 --inetd switch to request inetd-style file descriptor passing.
8442 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
8443 variable, which takes a boolean value. If set to false, ANSI color
8444 output is disabled in the tools even when run on a terminal that
8447 * The VXLAN support in networkd now supports two new settings
8448 DestinationPort= and PortRange=.
8450 * A new systemd.machine_id= kernel command line switch has been added,
8451 that may be used to set the machine ID in /etc/machine-id if it is
8452 not initialized yet. This command line option has no effect if the
8453 file is already initialized.
8455 * systemd-nspawn gained a new --as-pid2 switch that invokes any
8456 specified command line as PID 2 rather than PID 1 in the
8457 container. In this mode PID 1 is a minimal stub init process that
8458 implements the special POSIX and Linux semantics of PID 1 regarding
8459 signal and child process management. Note that this stub init process
8460 is implemented in nspawn itself and requires no support from the
8461 container image. This new logic is useful to support running
8462 arbitrary commands in the container, as normal processes are
8463 generally not prepared to run as PID 1.
8465 * systemd-nspawn gained a new --chdir= switch for setting the current
8466 working directory for the process started in the container.
8468 * "journalctl /dev/sda" will now output all kernel log messages for
8469 specified device from the current boot, in addition to all devices
8470 that are parents of it. This should make log output about devices
8471 pretty useful, as long as kernel drivers attach enough metadata to
8472 the log messages. (The usual SATA drivers do.)
8474 * The sd-journal API gained two new calls
8475 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
8476 that report whether log data from /run or /var has been found.
8478 * journalctl gained a new switch "--fields" that prints all journal
8479 record field names currently in use in the journal. This is backed
8480 by two new sd-journal API calls sd_journal_enumerate_fields() and
8481 sd_journal_restart_fields().
8483 * Most configurable timeouts in systemd now expect an argument of
8484 "infinity" to turn them off, instead of "0" as before. The semantics
8485 from now on is that a timeout of "0" means "now", and "infinity"
8486 means "never". To maintain backwards compatibility, "0" continues to
8487 turn off previously existing timeout settings.
8489 * "systemctl reload-or-try-restart" has been renamed to "systemctl
8490 try-reload-or-restart" to clarify what it actually does: the "try"
8491 logic applies to both reloading and restarting, not just restarting.
8492 The old name continues to be accepted for compatibility.
8494 * On boot-up, when PID 1 detects that the system clock is behind the
8495 release date of the systemd version in use, the clock is now set
8496 to the latter. Previously, this was already done in timesyncd, in order
8497 to avoid running with clocks set to the various clock epochs such as
8498 1902, 1938 or 1970. With this change the logic is now done in PID 1
8499 in addition to timesyncd during early boot-up, so that it is enforced
8500 before the first process is spawned by systemd. Note that the logic
8501 in timesyncd remains, as it is more comprehensive and ensures
8502 clock monotonicity by maintaining a persistent timestamp file in
8503 /var. Since /var is generally not available in earliest boot or the
8504 initrd, this part of the logic remains in timesyncd, and is not done
8507 * Support for tweaking details in net_cls.class_id through the
8508 NetClass= configuration directive has been removed, as the kernel
8509 people have decided to deprecate that controller in cgroup v2.
8510 Userspace tools such as nftables are moving over to setting rules
8511 that are specific to the full cgroup path of a task, which obsoletes
8512 these controllers anyway. The NetClass= directive is kept around for
8513 legacy compatibility reasons. For a more in-depth description of the
8514 kernel change, please refer to the respective upstream commit:
8516 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
8518 * A new service setting RuntimeMaxSec= has been added that may be used
8519 to specify a maximum runtime for a service. If the timeout is hit, the
8520 service is terminated and put into a failure state.
8522 * A new service setting AmbientCapabilities= has been added. It allows
8523 configuration of additional Linux process capabilities that are
8524 passed to the activated processes. This is only available on very
8527 * The process resource limit settings in service units may now be used
8528 to configure hard and soft limits individually.
8530 * The various libsystemd APIs such as sd-bus or sd-event now publicly
8531 expose support for gcc's __attribute__((cleanup())) C extension.
8532 Specifically, for many object destructor functions alternative
8533 versions have been added that have names suffixed with "p" and take a
8534 pointer to a pointer to the object to destroy, instead of just a
8535 pointer to the object itself. This is useful because these destructor
8536 functions may be used directly as parameters to the cleanup
8537 construct. Internally, systemd has been a heavy user of this GCC
8538 extension for a long time, and with this change similar support is
8539 now available to consumers of the library outside of systemd. Note
8540 that by using this extension in your sources compatibility with old
8541 and strictly ANSI compatible C compilers is lost. However, all gcc or
8542 LLVM versions of recent years support this extension.
8544 * Timer units gained support for a new setting RandomizedDelaySec= that
8545 allows configuring some additional randomized delay to the configured
8546 time. This is useful to spread out timer events to avoid load peaks in
8547 clusters or larger setups.
8549 * Calendar time specifications now support sub-second accuracy.
8551 * Socket units now support listening on SCTP and UDP-lite protocol
8554 * The sd-event API now comes with a full set of man pages.
8556 * Older versions of systemd contained experimental support for
8557 compressing journal files and coredumps with the LZ4 compressor that
8558 was not compatible with the lz4 binary (due to API limitations of the
8559 lz4 library). This support has been removed; only support for files
8560 compatible with the lz4 binary remains. This LZ4 logic is now
8561 officially supported and no longer considered experimental.
8563 * The dkr image import logic has been removed again from importd. dkr's
8564 micro-services focus doesn't fit into the machine image focus of
8565 importd, and quickly got out of date with the upstream dkr API.
8567 * Creation of the /run/lock/lockdev/ directory was dropped from
8568 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
8569 been available for many years. If you still need this, you need to
8570 create your own tmpfiles.d config file with:
8572 d /run/lock/lockdev 0775 root lock -
8574 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
8575 and RebootArgument= have been moved from the [Service] section of
8576 unit files to [Unit], and they are now supported on all unit types,
8577 not just service units. Of course, systemd will continue to
8578 understand these settings also at the old location, in order to
8579 maintain compatibility.
8581 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
8582 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
8583 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
8584 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
8585 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
8586 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
8587 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
8588 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
8589 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
8590 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
8591 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
8592 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
8593 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
8594 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
8595 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
8596 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
8597 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
8598 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
8599 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8601 — Berlin, 2016-02-11
8605 * A number of properties previously only settable in unit
8606 files are now also available as properties to set when
8607 creating transient units programmatically via the bus, as it
8608 is exposed with systemd-run's --property=
8609 setting. Specifically, these are: SyslogIdentifier=,
8610 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
8611 EnvironmentFile=, ReadWriteDirectories=,
8612 ReadOnlyDirectories=, InaccessibleDirectories=,
8613 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
8615 * When creating transient services via the bus API it is now
8616 possible to pass in a set of file descriptors to use as
8617 STDIN/STDOUT/STDERR for the invoked process.
8619 * Slice units may now be created transiently via the bus APIs,
8620 similar to the way service and scope units may already be
8621 created transiently.
8623 * Wherever systemd expects a calendar timestamp specification
8624 (like in journalctl's --since= and --until= switches) UTC
8625 timestamps are now supported. Timestamps suffixed with "UTC"
8626 are now considered to be in Universal Time Coordinated
8627 instead of the local timezone. Also, timestamps may now
8628 optionally be specified with sub-second accuracy. Both of
8629 these additions also apply to recurring calendar event
8630 specification, such as OnCalendar= in timer units.
8632 * journalctl gained a new "--sync" switch that asks the
8633 journal daemon to write all so far unwritten log messages to
8634 disk and sync the files, before returning.
8636 * systemd-tmpfiles learned two new line types "q" and "Q" that
8637 operate like "v", but also set up a basic btrfs quota
8638 hierarchy when used on a btrfs file system with quota
8641 * tmpfiles' "v", "q" and "Q" will now create a plain directory
8642 instead of a subvolume (even on a btrfs file system) if the
8643 root directory is a plain directory, and not a
8644 subvolume. This should simplify things with certain chroot()
8645 environments which are not aware of the concept of btrfs
8648 * systemd-detect-virt gained a new --chroot switch to detect
8649 whether execution takes place in a chroot() environment.
8651 * CPUAffinity= now takes CPU index ranges in addition to
8654 * The various memory-related resource limit settings (such as
8655 LimitAS=) now understand the usual K, M, G, … suffixes to
8656 the base of 1024 (IEC). Similar, the time-related resource
8657 limit settings understand the usual min, h, day, … suffixes
8660 * There's a new system.conf setting DefaultTasksMax= to
8661 control the default TasksMax= setting for services and
8662 scopes running on the system. (TasksMax= is the primary
8663 setting that exposes the "pids" cgroup controller on systemd
8664 and was introduced in the previous systemd release.) The
8665 setting now defaults to 512, which means services that are
8666 not explicitly configured otherwise will only be able to
8667 create 512 processes or threads at maximum, from this
8668 version on. Note that this means that thread- or
8669 process-heavy services might need to be reconfigured to set
8670 TasksMax= to a higher value. It is sufficient to set
8671 TasksMax= in these specific unit files to a higher value, or
8672 even "infinity". Similar, there's now a logind.conf setting
8673 UserTasksMax= that defaults to 4096 and limits the total
8674 number of processes or tasks each user may own
8675 concurrently. nspawn containers also have the TasksMax=
8676 value set by default now, to 8192. Note that all of this
8677 only has an effect if the "pids" cgroup controller is
8678 enabled in the kernel. The general benefit of these changes
8679 should be a more robust and safer system, that provides a
8680 certain amount of per-service fork() bomb protection.
8682 * systemd-nspawn gained the new --network-veth-extra= switch
8683 to define additional and arbitrarily-named virtual Ethernet
8684 links between the host and the container.
8686 * A new service execution setting PassEnvironment= has been
8687 added that allows importing select environment variables
8688 from PID1's environment block into the environment block of
8691 * Timer units gained support for a new RemainAfterElapse=
8692 setting which takes a boolean argument. It defaults to on,
8693 exposing behaviour unchanged to previous releases. If set to
8694 off, timer units are unloaded after they elapsed if they
8695 cannot elapse again. This is particularly useful for
8696 transient timer units, which shall not stay around longer
8697 than until they first elapse.
8699 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
8700 default now (the kernel default is 16). This is beneficial
8701 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
8702 allows substantially larger numbers of queued
8703 datagrams. This should increase the capability of systemd to
8704 parallelize boot-up, as logging and sd_notify() are unlikely
8705 to stall execution anymore. If you need to change the value
8706 from the new defaults, use the usual sysctl.d/ snippets.
8708 * The compression framing format used by the journal or
8709 coredump processing has changed to be in line with what the
8710 official LZ4 tools generate. LZ4 compression support in
8711 systemd was considered unsupported previously, as the format
8712 was not compatible with the normal tools. With this release
8713 this has changed now, and it is hence safe for downstream
8714 distributions to turn it on. While not compressing as well
8715 as the XZ, LZ4 is substantially faster, which makes
8716 it a good default choice for the compression logic in the
8717 journal and in coredump handling.
8719 * Any reference to /etc/mtab has been dropped from
8720 systemd. The file has been obsolete since a while, but
8721 systemd refused to work on systems where it was incorrectly
8722 set up (it should be a symlink or non-existent). Please make
8723 sure to update to util-linux 2.27.1 or newer in conjunction
8724 with this systemd release, which also drops any reference to
8725 /etc/mtab. If you maintain a distribution make sure that no
8726 software you package still references it, as this is a
8727 likely source of bugs. There's also a glibc bug pending,
8728 asking for removal of any reference to this obsolete file:
8730 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
8732 Note that only util-linux versions built with
8733 --enable-libmount-force-mountinfo are supported.
8735 * Support for the ".snapshot" unit type has been removed. This
8736 feature turned out to be little useful and little used, and
8737 has now been removed from the core and from systemctl.
8739 * The dependency types RequiresOverridable= and
8740 RequisiteOverridable= have been removed from systemd. They
8741 have been used only very sparingly to our knowledge and
8742 other options that provide a similar effect (such as
8743 systemctl --mode=ignore-dependencies) are much more useful
8744 and commonly used. Moreover, they were only half-way
8745 implemented as the option to control behaviour regarding
8746 these dependencies was never added to systemctl. By removing
8747 these dependency types the execution engine becomes a bit
8748 simpler. Unit files that use these dependencies should be
8749 changed to use the non-Overridable dependency types
8750 instead. In fact, when parsing unit files with these
8751 options, that's what systemd will automatically convert them
8752 too, but it will also warn, asking users to fix the unit
8753 files accordingly. Removal of these dependency types should
8754 only affect a negligible number of unit files in the wild.
8756 * Behaviour of networkd's IPForward= option changed
8757 (again). It will no longer maintain a per-interface setting,
8758 but propagate one way from interfaces where this is enabled
8759 to the global kernel setting. The global setting will be
8760 enabled when requested by a network that is set up, but
8761 never be disabled again. This change was made to make sure
8762 IPv4 and IPv6 behaviour regarding packet forwarding is
8763 similar (as the Linux IPv6 stack does not support
8764 per-interface control of this setting) and to minimize
8767 * In unit files the behaviour of %u, %U, %h, %s has
8768 changed. These specifiers will now unconditionally resolve
8769 to the various user database fields of the user that the
8770 systemd instance is running as, instead of the user
8771 configured in the specific unit via User=. Note that this
8772 effectively doesn't change much, as resolving of these
8773 specifiers was already turned off in the --system instance
8774 of systemd, as we cannot do NSS lookups from PID 1. In the
8775 --user instance of systemd these specifiers where correctly
8776 resolved, but hardly made any sense, since the user instance
8777 lacks privileges to do user switches anyway, and User= is
8778 hence useless. Moreover, even in the --user instance of
8779 systemd behaviour was awkward as it would only take settings
8780 from User= assignment placed before the specifier into
8781 account. In order to unify and simplify the logic around
8782 this the specifiers will now always resolve to the
8783 credentials of the user invoking the manager (which in case
8784 of PID 1 is the root user).
8786 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
8787 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
8788 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
8789 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
8790 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
8791 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
8792 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
8793 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
8794 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
8795 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
8796 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
8797 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
8798 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
8799 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
8802 — Berlin, 2015-11-18
8806 * systemd now depends on util-linux v2.27. More specifically,
8807 the newly added mount monitor feature in libmount now
8808 replaces systemd's former own implementation.
8810 * libmount mandates /etc/mtab not to be regular file, and
8811 systemd now enforces this condition at early boot.
8812 /etc/mtab has been deprecated and warned about for a very
8813 long time, so systems running systemd should already have
8814 stopped having this file around as anything else than a
8815 symlink to /proc/self/mounts.
8817 * Support for the "pids" cgroup controller has been added. It
8818 allows accounting the number of tasks in a cgroup and
8819 enforcing limits on it. This adds two new setting
8820 TasksAccounting= and TasksMax= to each unit, as well as a
8821 global option DefaultTasksAccounting=.
8823 * Support for the "net_cls" cgroup controller has been added.
8824 It allows assigning a net class ID to each task in the
8825 cgroup, which can then be used in firewall rules and traffic
8826 shaping configurations. Note that the kernel netfilter net
8827 class code does not currently work reliably for ingress
8828 packets on unestablished sockets.
8830 This adds a new config directive called NetClass= to CGroup
8831 enabled units. Allowed values are positive numbers for fixed
8832 assignments and "auto" for picking a free value
8835 * 'systemctl is-system-running' now returns 'offline' if the
8836 system is not booted with systemd. This command can now be
8837 used as a substitute for 'systemd-notify --booted'.
8839 * Watchdog timeouts have been increased to 3 minutes for all
8840 in-tree service files. Apparently, disk IO issues are more
8841 frequent than we hoped, and user reported >1 minute waiting
8844 * 'machine-id-commit' functionality has been merged into
8845 'machine-id-setup --commit'. The separate binary has been
8848 * The WorkingDirectory= directive in unit files may now be set
8849 to the special value '~'. In this case, the working
8850 directory is set to the home directory of the user
8851 configured in User=.
8853 * "machinectl shell" will now open the shell in the home
8854 directory of the selected user by default.
8856 * The CrashChVT= configuration file setting is renamed to
8857 CrashChangeVT=, following our usual logic of not
8858 abbreviating unnecessarily. The old directive is still
8859 supported for compat reasons. Also, this directive now takes
8860 an integer value between 1 and 63, or a boolean value. The
8861 formerly supported '-1' value for disabling stays around for
8864 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
8865 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
8866 RootDirectory= properties can now be set for transient
8869 * The systemd-analyze tool gained a new "set-log-target" verb
8870 to change the logging target the system manager logs to
8871 dynamically during runtime. This is similar to how
8872 "systemd-analyze set-log-level" already changes the log
8875 * In nspawn /sys is now mounted as tmpfs, with only a selected
8876 set of subdirectories mounted in from the real sysfs. This
8877 enhances security slightly, and is useful for ensuring user
8878 namespaces work correctly.
8880 * Support for USB FunctionFS activation has been added. This
8881 allows implementation of USB gadget services that are
8882 activated as soon as they are requested, so that they don't
8883 have to run continuously, similar to classic socket
8886 * The "systemctl exit" command now optionally takes an
8887 additional parameter that sets the exit code to return from
8888 the systemd manager when exiting. This is only relevant when
8889 running the systemd user instance, or when running the
8890 system instance in a container.
8892 * sd-bus gained the new API calls sd_bus_path_encode_many()
8893 and sd_bus_path_decode_many() that allow easy encoding and
8894 decoding of multiple identifier strings inside a D-Bus
8895 object path. Another new call sd_bus_default_flush_close()
8896 has been added to flush and close per-thread default
8899 * systemd-cgtop gained support for a -M/--machine= switch to
8900 show the control groups within a certain container only.
8902 * "systemctl kill" gained support for an optional --fail
8903 switch. If specified the requested operation will fail of no
8904 processes have been killed, because the unit had no
8905 processes attached, or similar.
8907 * A new systemd.crash_reboot=1 kernel command line option has
8908 been added that triggers a reboot after crashing. This can
8909 also be set through CrashReboot= in systemd.conf.
8911 * The RuntimeDirectory= setting now understands unit
8912 specifiers like %i or %f.
8914 * A new (still internal) library API sd-ipv4acd has been added,
8915 that implements address conflict detection for IPv4. It's
8916 based on code from sd-ipv4ll, and will be useful for
8917 detecting DHCP address conflicts.
8919 * File descriptors passed during socket activation may now be
8920 named. A new API sd_listen_fds_with_names() is added to
8921 access the names. The default names may be overridden,
8922 either in the .socket file using the FileDescriptorName=
8923 parameter, or by passing FDNAME= when storing the file
8924 descriptors using sd_notify().
8926 * systemd-networkd gained support for:
8928 - Setting the IPv6 Router Advertisement settings via
8929 IPv6AcceptRouterAdvertisements= in .network files.
8931 - Configuring the HelloTimeSec=, MaxAgeSec= and
8932 ForwardDelaySec= bridge parameters in .netdev files.
8934 - Configuring PreferredSource= for static routes in
8937 * The "ask-password" framework used to query for LUKS harddisk
8938 passwords or SSL passwords during boot gained support for
8939 caching passwords in the kernel keyring, if it is
8940 available. This makes sure that the user only has to type in
8941 a passphrase once if there are multiple objects to unlock
8942 with the same one. Previously, such password caching was
8943 available only when Plymouth was used; this moves the
8944 caching logic into the systemd codebase itself. The
8945 "systemd-ask-password" utility gained a new --keyname=
8946 switch to control which kernel keyring key to use for
8947 caching a password in. This functionality is also useful for
8948 enabling display managers such as gdm to automatically
8949 unlock the user's GNOME keyring if its passphrase, the
8950 user's password and the harddisk password are the same, if
8951 gdm-autologin is used.
8953 * When downloading tar or raw images using "machinectl
8954 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
8955 file is now also downloaded, if it is available and stored
8956 next to the image file.
8958 * Units of type ".socket" gained a new boolean setting
8959 Writable= which is only useful in conjunction with
8960 ListenSpecial=. If true, enables opening the specified
8961 special file in O_RDWR mode rather than O_RDONLY mode.
8963 * systemd-rfkill has been reworked to become a singleton
8964 service that is activated through /dev/rfkill on each rfkill
8965 state change and saves the settings to disk. This way,
8966 systemd-rfkill is now compatible with devices that exist
8967 only intermittendly, and even restores state if the previous
8968 system shutdown was abrupt rather than clean.
8970 * The journal daemon gained support for vacuuming old journal
8971 files controlled by the number of files that shall remain,
8972 in addition to the already existing control by size and by
8973 date. This is useful as journal interleaving performance
8974 degrades with too many separate journal files, and allows
8975 putting an effective limit on them. The new setting defaults
8976 to 100, but this may be changed by setting SystemMaxFiles=
8977 and RuntimeMaxFiles= in journald.conf. Also, the
8978 "journalctl" tool gained the new --vacuum-files= switch to
8979 manually vacuum journal files to leave only the specified
8980 number of files in place.
8982 * udev will now create /dev/disk/by-path links for ATA devices
8983 on kernels where that is supported.
8985 * Galician, Serbian, Turkish and Korean translations were added.
8987 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
8988 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
8989 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
8990 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
8991 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
8992 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
8993 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
8994 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
8995 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
8996 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
8997 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
8998 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
8999 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
9000 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
9001 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
9002 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
9003 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
9004 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
9006 — Berlin, 2015-10-07
9010 * The DHCP implementation of systemd-networkd gained a set of
9013 - The DHCP server now supports emitting DNS and NTP
9014 information. It may be enabled and configured via
9015 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
9016 and NTP information is enabled, but no servers are
9017 configured, the corresponding uplink information (if there
9018 is any) is propagated.
9020 - Server and client now support transmission and reception
9021 of timezone information. It can be configured via the
9022 newly introduced network options UseTimezone=,
9023 EmitTimezone=, and Timezone=. Transmission of timezone
9024 information is enabled between host and containers by
9025 default now: the container will change its local timezone
9026 to what the host has set.
9028 - Lease timeouts can now be configured via
9029 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
9031 - The DHCP server improved on the stability of
9032 leases. Clients are more likely to get the same lease
9033 information back, even if the server loses state.
9035 - The DHCP server supports two new configuration options to
9036 control the lease address pool metrics, PoolOffset= and
9039 * The encapsulation limit of tunnels in systemd-networkd may
9040 now be configured via 'EncapsulationLimit='. It allows
9041 modifying the maximum additional levels of encapsulation
9042 that are permitted to be prepended to a packet.
9044 * systemd now supports the concept of user buses replacing
9045 session buses, if used with dbus-1.10 (and enabled via dbus
9046 --enable-user-session). It previously only supported this on
9047 kdbus-enabled systems, and this release expands this to
9048 'dbus-daemon' systems.
9050 * systemd-networkd now supports predictable interface names
9053 * systemd now optionally supports the new Linux kernel
9054 "unified" control group hierarchy. If enabled via the kernel
9055 command-line option 'systemd.unified_cgroup_hierarchy=1',
9056 systemd will try to mount the unified cgroup hierarchy
9057 directly on /sys/fs/cgroup. If not enabled, or not
9058 available, systemd will fall back to the legacy cgroup
9059 hierarchy setup, as before. Host system and containers can
9060 mix and match legacy and unified hierarchies as they
9061 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
9062 environment variable to individually select the hierarchy to
9063 use for executed containers. By default, nspawn will use the
9064 unified hierarchy for the containers if the host uses the
9065 unified hierarchy, and the legacy hierarchy otherwise.
9066 Please note that at this point the unified hierarchy is an
9067 experimental kernel feature and is likely to change in one
9068 of the next kernel releases. Therefore, it should not be
9069 enabled by default in downstream distributions yet. The
9070 minimum required kernel version for the unified hierarchy to
9071 work is 4.2. Note that when the unified hierarchy is used
9072 for the first time delegated access to controllers is
9073 safe. Because of this systemd-nspawn containers will get
9074 access to controllers now, as will systemd user
9075 sessions. This means containers and user sessions may now
9076 manage their own resources, partitioning up what the system
9079 * A new special scope unit "init.scope" has been introduced
9080 that encapsulates PID 1 of the system. It may be used to
9081 determine resource usage and enforce resource limits on PID
9082 1 itself. PID 1 hence moved out of the root of the control
9085 * The cgtop tool gained support for filtering out kernel
9086 threads when counting tasks in a control group. Also, the
9087 count of processes is now recursively summed up by
9088 default. Two options -k and --recursive= have been added to
9089 revert to old behaviour. The tool has also been updated to
9090 work correctly in containers now.
9092 * systemd-nspawn's --bind= and --bind-ro= options have been
9093 extended to allow creation of non-recursive bind mounts.
9095 * libsystemd gained two new calls sd_pid_get_cgroup() and
9096 sd_peer_get_cgroup() which return the control group path of
9097 a process or peer of a connected AF_UNIX socket. This
9098 function call is particularly useful when implementing
9099 delegated subtrees support in the control group hierarchy.
9101 * The "sd-event" event loop API of libsystemd now supports
9102 correct dequeuing of real-time signals, without losing
9105 * When systemd requests a polkit decision when managing units it
9106 will now add additional fields to the request, including unit
9107 name and desired operation. This enables more powerful polkit
9108 policies, that make decisions depending on these parameters.
9110 * nspawn learnt support for .nspawn settings files, that may
9111 accompany the image files or directories of containers, and
9112 may contain additional settings for the container. This is
9113 an alternative to configuring container parameters via the
9114 nspawn command line.
9116 Contributions from: Cristian Rodríguez, Daniel Mack, David
9117 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
9118 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
9119 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
9120 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
9121 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
9122 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
9123 Andersen, Tom Gundersen, Torstein Husebø
9125 — Berlin, 2015-09-08
9129 * machinectl gained a new verb 'shell' which opens a fresh
9130 shell on the target container or the host. It is similar to
9131 the existing 'login' command of machinectl, but spawns the
9132 shell directly without prompting for username or
9133 password. The pseudo machine '.host' now refers to the local
9134 host and is used by default. Hence, 'machinectl shell' can
9135 be used as replacement for 'su -' which spawns a session as
9136 a fresh systemd unit in a way that is fully isolated from
9137 the originating session.
9139 * systemd-networkd learned to cope with private-zone DHCP
9140 options and allows other programs to query the values.
9142 * SELinux access control when enabling/disabling units is no
9143 longer enforced with this release. The previous implementation
9144 was incorrect, and a new corrected implementation is not yet
9145 available. As unit file operations are still protected via
9146 polkit and D-Bus policy this is not a security problem. Yet,
9147 distributions which care about optimal SELinux support should
9148 probably not stabilize on this release.
9150 * sd-bus gained support for matches of type "arg0has=", that
9151 test for membership of strings in string arrays sent in bus
9154 * systemd-resolved now dumps the contents of its DNS and LLMNR
9155 caches to the logs on reception of the SIGUSR1 signal. This
9156 is useful to debug DNS behaviour.
9158 * The coredumpctl tool gained a new --directory= option to
9159 operate on journal files in a specific directory.
9161 * "systemctl reboot" and related commands gained a new
9162 "--message=" option which may be used to set a free-text
9163 wall message when shutting down or rebooting the
9164 system. This message is also logged, which is useful for
9165 figuring out the reason for a reboot or shutdown a
9168 * The "systemd-resolve-host" tool's -i switch now takes
9169 network interface numbers as alternative to interface names.
9171 * A new unit file setting for services has been introduced:
9172 UtmpMode= allows configuration of how precisely systemd
9173 handles utmp and wtmp entries for the service if this is
9174 enabled. This allows writing services that appear similar to
9175 user sessions in the output of the "w", "who", "last" and
9178 * systemd-resolved will now locally synthesize DNS resource
9179 records for the "localhost" and "gateway" domains as well as
9180 the local hostname. This should ensure that clients querying
9181 RRs via resolved will get similar results as those going via
9182 NSS, if nss-myhostname is enabled.
9184 Contributions from: Alastair Hughes, Alex Crawford, Daniel
9185 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
9186 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
9187 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
9188 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
9189 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
9190 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
9191 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
9192 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
9193 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
9194 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
9195 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
9197 — Berlin, 2015-08-27
9201 * The systemd-efi-boot-generator functionality was merged into
9202 systemd-gpt-auto-generator.
9204 * systemd-networkd now supports Group Policy for vxlan
9205 devices. It can be enabled via the new boolean configuration
9206 option called 'GroupPolicyExtension='.
9208 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
9209 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
9210 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
9212 — Berlin, 2015-07-31
9216 * The python-systemd code has been removed from the systemd repository.
9217 A new repository has been created which accommodates the code from
9218 now on, and we kindly ask distributions to create a separate package
9219 for this: https://github.com/systemd/python-systemd
9221 * The systemd daemon will now reload its main configuration
9222 (/etc/systemd/system.conf) on daemon-reload.
9224 * sd-dhcp now exposes vendor specific extensions via
9225 sd_dhcp_lease_get_vendor_specific().
9227 * systemd-networkd gained a number of new configuration options.
9229 - A new boolean configuration option for TAP devices called
9230 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
9231 device, thus allowing to send and receive GSO packets.
9233 - A new tunnel configuration option called 'CopyDSCP='.
9234 If enabled, the DSCP field of ip6 tunnels is copied into the
9235 decapsulated packet.
9237 - A set of boolean bridge configuration options were added.
9238 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
9239 and 'UnicastFlood=' are now parsed by networkd and applied to the
9240 respective bridge link device via the respective IFLA_BRPORT_*
9243 - A new string configuration option to override the hostname sent
9244 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
9245 is true, networkd will use the configured hostname instead of the
9246 system hostname when sending DHCP requests.
9248 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
9249 networkd will configure the IPv6 flow-label of the tunnel device
9250 according to RFC2460.
9252 - The 'macvtap' virtual network devices are now supported, similar to
9253 the already supported 'macvlan' devices.
9255 * systemd-resolved now implements RFC5452 to improve resilience against
9256 cache poisoning. Additionally, source port randomization is enabled
9257 by default to further protect against DNS spoofing attacks.
9259 * nss-mymachines now supports translating UIDs and GIDs of running
9260 containers with user-namespaces enabled. If a container 'foo'
9261 translates a host uid 'UID' to the container uid 'TUID', then
9262 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
9263 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
9264 mapped as 'vg-foo-TGID'.
9266 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
9267 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
9268 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
9269 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
9270 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
9271 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
9272 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
9273 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
9274 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
9275 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
9277 — Berlin, 2015-07-29
9281 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
9282 There are no known issues with current sysfs, and udev does not need
9283 or should be used to work around such bugs.
9285 * udev does no longer enable USB HID power management. Several reports
9286 indicate, that some devices cannot handle that setting.
9288 * The udev accelerometer helper was removed. The functionality
9289 is now fully included in iio-sensor-proxy. But this means,
9290 older iio-sensor-proxy versions will no longer provide
9291 accelerometer/orientation data with this systemd version.
9292 Please upgrade iio-sensor-proxy to version 1.0.
9294 * networkd gained a new configuration option IPv6PrivacyExtensions=
9295 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
9296 for Stateless Address") on selected networks.
9298 * For the sake of fewer build-time dependencies and less code in the
9299 main repository, the python bindings are about to be removed in the
9300 next release. A new repository has been created which accommodates
9301 the code from now on, and we kindly ask distributions to create a
9302 separate package for this. The removal will take place in v223.
9304 https://github.com/systemd/python-systemd
9306 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
9307 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
9308 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
9309 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
9310 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
9311 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
9312 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
9313 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
9314 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
9315 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
9317 — Berlin, 2015-07-07
9321 * The sd-bus.h and sd-event.h APIs have now been declared
9322 stable and have been added to the official interface of
9323 libsystemd.so. sd-bus implements an alternative D-Bus client
9324 library, that is relatively easy to use, very efficient and
9325 supports both classic D-Bus as well as kdbus as transport
9326 backend. sd-event is a generic event loop abstraction that
9327 is built around Linux epoll, but adds features such as event
9328 prioritization or efficient timer handling. Both APIs are good
9329 choices for C programs looking for a bus and/or event loop
9330 implementation that is minimal and does not have to be
9331 portable to other kernels.
9333 * kdbus support is no longer compile-time optional. It is now
9334 always built-in. However, it can still be disabled at
9335 runtime using the kdbus=0 kernel command line setting, and
9336 that setting may be changed to default to off, by specifying
9337 --disable-kdbus at build-time. Note though that the kernel
9338 command line setting has no effect if the kdbus.ko kernel
9339 module is not installed, in which case kdbus is (obviously)
9340 also disabled. We encourage all downstream distributions to
9341 begin testing kdbus by adding it to the kernel images in the
9342 development distributions, and leaving kdbus support in
9345 * The minimal required util-linux version has been bumped to
9348 * Support for chkconfig (--enable-chkconfig) was removed in
9349 favor of calling an abstraction tool
9350 /lib/systemd/systemd-sysv-install. This needs to be
9351 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
9352 in README for details.
9354 * If there's a systemd unit and a SysV init script for the
9355 same service name, and the user executes "systemctl enable"
9356 for it (or a related call), then this will now enable both
9357 (or execute the related operation on both), not just the
9360 * The libudev API documentation has been converted from gtkdoc
9363 * gudev has been removed from the systemd tree, it is now an
9366 * The systemd-cgtop tool learnt a new --raw switch to generate
9367 "raw" (machine parsable) output.
9369 * networkd's IPForwarding= .network file setting learnt the
9370 new setting "kernel", which ensures that networkd does not
9371 change the IP forwarding sysctl from the default kernel
9374 * The systemd-logind bus API now exposes a new boolean
9375 property "Docked" that reports whether logind considers the
9376 system "docked", i.e. connected to a docking station or not.
9378 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
9379 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
9380 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
9381 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
9382 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
9383 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
9384 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
9385 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
9386 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
9387 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
9388 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
9389 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
9390 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
9391 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
9392 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
9393 Fink, Zbigniew Jędrzejewski-Szmek
9395 — Berlin, 2015-06-19
9399 * The gudev library has been extracted into a separate repository
9400 available at: https://git.gnome.org/browse/libgudev/
9401 It is now managed as part of the Gnome project. Distributions
9402 are recommended to pass --disable-gudev to systemd and use
9403 gudev from the Gnome project instead. gudev is still included
9404 in systemd, for now. It will be removed soon, though. Please
9405 also see the announcement-thread on systemd-devel:
9406 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
9408 * systemd now exposes a CPUUsageNSec= property for each
9409 service unit on the bus, that contains the overall consumed
9410 CPU time of a service (the sum of what each process of the
9411 service consumed). This value is only available if
9412 CPUAccounting= is turned on for a service, and is then shown
9413 in the "systemctl status" output.
9415 * Support for configuring alternative mappings of the old SysV
9416 runlevels to systemd targets has been removed. They are now
9417 hardcoded in a way that runlevels 2, 3, 4 all map to
9418 multi-user.target and 5 to graphical.target (which
9419 previously was already the default behaviour).
9421 * The auto-mounter logic gained support for mount point
9422 expiry, using a new TimeoutIdleSec= setting in .automount
9423 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
9425 * The EFI System Partition (ESP) as mounted to /boot by
9426 systemd-efi-boot-generator will now be unmounted
9427 automatically after 2 minutes of not being used. This should
9428 minimize the risk of ESP corruptions.
9430 * New /etc/fstab options x-systemd.requires= and
9431 x-systemd.requires-mounts-for= are now supported to express
9432 additional dependencies for mounts. This is useful for
9433 journaling file systems that support external journal
9434 devices or overlay file systems that require underlying file
9435 systems to be mounted.
9437 * systemd does not support direct live-upgrades (via systemctl
9438 daemon-reexec) from versions older than v44 anymore. As no
9439 distribution we are aware of shipped such old versions in a
9440 stable release this should not be problematic.
9442 * When systemd forks off a new per-connection service instance
9443 it will now set the $REMOTE_ADDR environment variable to the
9444 remote IP address, and $REMOTE_PORT environment variable to
9445 the remote IP port. This behaviour is similar to the
9446 corresponding environment variables defined by CGI.
9448 * systemd-networkd gained support for uplink failure
9449 detection. The BindCarrier= option allows binding interface
9450 configuration dynamically to the link sense of other
9451 interfaces. This is useful to achieve behaviour like in
9454 * systemd-networkd gained support for configuring the DHCP
9455 client identifier to use when requesting leases.
9457 * systemd-networkd now has a per-network UseNTP= option to
9458 configure whether NTP server information acquired via DHCP
9459 is passed on to services like systemd-timesyncd.
9461 * systemd-networkd gained support for vti6 tunnels.
9463 * Note that systemd-networkd manages the sysctl variable
9464 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
9465 it is configured for since v219. The variable controls IP
9466 forwarding, and is a per-interface alternative to the global
9467 /proc/sys/net/ipv[46]/ip_forward. This setting is
9468 configurable in the IPForward= option, which defaults to
9469 "no". This means if networkd is used for an interface it is
9470 no longer sufficient to set the global sysctl option to turn
9471 on IP forwarding! Instead, the .network file option
9472 IPForward= needs to be turned on! Note that the
9473 implementation of this behaviour was broken in v219 and has
9476 * Many bonding and vxlan options are now configurable in
9479 * systemd-nspawn gained a new --property= setting to set unit
9480 properties for the container scope. This is useful for
9481 setting resource parameters (e.g. "CPUShares=500") on
9482 containers started from the command line.
9484 * systemd-nspawn gained a new --private-users= switch to make
9485 use of user namespacing available on recent Linux kernels.
9487 * systemd-nspawn may now be called as part of a shell pipeline
9488 in which case the pipes used for stdin and stdout are passed
9489 directly to the process invoked in the container, without
9490 indirection via a pseudo tty.
9492 * systemd-nspawn gained a new switch to control the UNIX
9493 signal to use when killing the init process of the container
9496 * systemd-nspawn gained a new --overlay= switch for mounting
9497 overlay file systems into the container using the new kernel
9500 * When a container image is imported via systemd-importd and
9501 the host file system is not btrfs, a loopback block device
9502 file is created in /var/lib/machines.raw with a btrfs file
9503 system inside. It is then mounted to /var/lib/machines to
9504 enable btrfs features for container management. The loopback
9505 file and btrfs file system is grown as needed when container
9506 images are imported via systemd-importd.
9508 * systemd-machined/systemd-importd gained support for btrfs
9509 quota, to enforce container disk space limits on disk. This
9510 is exposed in "machinectl set-limit".
9512 * systemd-importd now can import containers from local .tar,
9513 .raw and .qcow2 images, and export them to .tar and .raw. It
9514 can also import dkr v2 images now from the network (on top
9517 * systemd-importd gained support for verifying downloaded
9518 images with gpg2 (previously only gpg1 was supported).
9520 * systemd-machined, systemd-logind, systemd: most bus calls are
9521 now accessible to unprivileged processes via polkit. Also,
9522 systemd-logind will now allow users to kill their own sessions
9523 without further privileges or authorization.
9525 * systemd-shutdownd has been removed. This service was
9526 previously responsible for implementing scheduled shutdowns
9527 as exposed in /usr/bin/shutdown's time parameter. This
9528 functionality has now been moved into systemd-logind and is
9529 accessible via a bus interface.
9531 * "systemctl reboot" gained a new switch --firmware-setup that
9532 can be used to reboot into the EFI firmware setup, if that
9533 is available. systemd-logind now exposes an API on the bus
9534 to trigger such reboots, in case graphical desktop UIs want
9535 to cover this functionality.
9537 * "systemctl enable", "systemctl disable" and "systemctl mask"
9538 now support a new "--now" switch. If specified the units
9539 that are enabled will also be started, and the ones
9540 disabled/masked also stopped.
9542 * The Gummiboot EFI boot loader tool has been merged into
9543 systemd, and renamed to "systemd-boot". The bootctl tool has been
9544 updated to support systemd-boot.
9546 * An EFI kernel stub has been added that may be used to create
9547 kernel EFI binaries that contain not only the actual kernel,
9548 but also an initrd, boot splash, command line and OS release
9549 information. This combined binary can then be signed as a
9550 single image, so that the firmware can verify it all in one
9551 step. systemd-boot has special support for EFI binaries created
9552 like this and can extract OS release information from them
9553 and show them in the boot menu. This functionality is useful
9554 to implement cryptographically verified boot schemes.
9556 * Optional support has been added to systemd-fsck to pass
9557 fsck's progress report to an AF_UNIX socket in the file
9560 * udev will no longer create device symlinks for all block devices by
9561 default. A deny list for excluding special block devices from this
9562 logic has been turned into an allow list that requires picking block
9563 devices explicitly that require device symlinks.
9565 * A new (currently still internal) API sd-device.h has been
9566 added to libsystemd. This modernized API is supposed to
9567 replace libudev eventually. In fact, already much of libudev
9568 is now just a wrapper around sd-device.h.
9570 * A new hwdb database for storing metadata about pointing
9571 stick devices has been added.
9573 * systemd-tmpfiles gained support for setting file attributes
9574 similar to the "chattr" tool with new 'h' and 'H' lines.
9576 * systemd-journald will no longer unconditionally set the
9577 btrfs NOCOW flag on new journal files. This is instead done
9578 with tmpfiles snippet using the new 'h' line type. This
9579 allows easy disabling of this logic, by masking the
9580 journal-nocow.conf tmpfiles file.
9582 * systemd-journald will now translate audit message types to
9583 human readable identifiers when writing them to the
9584 journal. This should improve readability of audit messages.
9586 * The LUKS logic gained support for the offset= and skip=
9587 options in /etc/crypttab, as previously implemented by
9590 * /usr/lib/os-release gained a new optional field VARIANT= for
9591 distributions that support multiple variants (such as a
9592 desktop edition, a server edition, …)
9594 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
9595 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
9596 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
9597 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
9598 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
9599 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
9600 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
9601 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
9602 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
9603 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
9604 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
9605 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
9606 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
9607 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
9608 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
9609 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
9610 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
9611 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
9612 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
9613 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
9614 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
9615 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
9616 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
9617 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
9618 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
9619 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
9620 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
9622 — Berlin, 2015-05-22
9626 * Introduce a new API "sd-hwdb.h" for querying the hardware
9627 metadata database. With this minimal interface one can query
9628 and enumerate the udev hwdb, decoupled from the old libudev
9629 library. libudev's interface for this is now only a wrapper
9630 around sd-hwdb. A new tool systemd-hwdb has been added to
9631 interface with and update the database.
9633 * When any of systemd's tools copies files (for example due to
9634 tmpfiles' C lines) a btrfs reflink will attempted first,
9635 before bytewise copying is done.
9637 * systemd-nspawn gained a new --ephemeral switch. When
9638 specified a btrfs snapshot is taken of the container's root
9639 directory, and immediately removed when the container
9640 terminates again. Thus, a container can be started whose
9641 changes never alter the container's root directory, and are
9642 lost on container termination. This switch can also be used
9643 for starting a container off the root file system of the
9644 host without affecting the host OS. This switch is only
9645 available on btrfs file systems.
9647 * systemd-nspawn gained a new --template= switch. It takes the
9648 path to a container tree to use as template for the tree
9649 specified via --directory=, should that directory be
9650 missing. This allows instantiating containers dynamically,
9651 on first run. This switch is only available on btrfs file
9654 * When a .mount unit refers to a mount point on which multiple
9655 mounts are stacked, and the .mount unit is stopped all of
9656 the stacked mount points will now be unmounted until no
9657 mount point remains.
9659 * systemd now has an explicit notion of supported and
9660 unsupported unit types. Jobs enqueued for unsupported unit
9661 types will now fail with an "unsupported" error code. More
9662 specifically .swap, .automount and .device units are not
9663 supported in containers, .busname units are not supported on
9664 non-kdbus systems. .swap and .automount are also not
9665 supported if their respective kernel compile time options
9668 * machinectl gained support for two new "copy-from" and
9669 "copy-to" commands for copying files from a running
9670 container to the host or vice versa.
9672 * machinectl gained support for a new "bind" command to bind
9673 mount host directories into local containers. This is
9674 currently only supported for nspawn containers.
9676 * networkd gained support for configuring bridge forwarding
9677 database entries (fdb) from .network files.
9679 * A new tiny daemon "systemd-importd" has been added that can
9680 download container images in tar, raw, qcow2 or dkr formats,
9681 and make them available locally in /var/lib/machines, so
9682 that they can run as nspawn containers. The daemon can GPG
9683 verify the downloads (not supported for dkr, since it has no
9684 provisions for verifying downloads). It will transparently
9685 decompress bz2, xz, gzip compressed downloads if necessary,
9686 and restore sparse files on disk. The daemon uses privilege
9687 separation to ensure the actual download logic runs with
9688 fewer privileges than the daemon itself. machinectl has
9689 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
9690 make the functionality of importd available to the
9691 user. With this in place the Fedora and Ubuntu "Cloud"
9692 images can be downloaded and booted as containers unmodified
9693 (the Fedora images lack the appropriate GPG signature files
9694 currently, so they cannot be verified, but this will change
9695 soon, hopefully). Note that downloading images is currently
9696 only fully supported on btrfs.
9698 * machinectl is now able to list container images found in
9699 /var/lib/machines, along with some metadata about sizes of
9700 disk and similar. If the directory is located on btrfs and
9701 quota is enabled, this includes quota display. A new command
9702 "image-status" has been added that shows additional
9703 information about images.
9705 * machinectl is now able to clone container images
9706 efficiently, if the underlying file system (btrfs) supports
9707 it, with the new "machinectl clone" command. It also
9708 gained commands for renaming and removing images, as well as
9709 marking them read-only or read-write (supported also on
9710 legacy file systems).
9712 * networkd gained support for collecting LLDP network
9713 announcements, from hardware that supports this. This is
9714 shown in networkctl output.
9716 * systemd-run gained support for a new -t (--pty) switch for
9717 invoking a binary on a pty whose input and output is
9718 connected to the invoking terminal. This allows executing
9719 processes as system services while interactively
9720 communicating with them via the terminal. Most interestingly
9721 this is supported across container boundaries. Invoking
9722 "systemd-run -t /bin/bash" is an alternative to running a
9723 full login session, the difference being that the former
9724 will not register a session, nor go through the PAM session
9727 * tmpfiles gained support for a new "v" line type for creating
9728 btrfs subvolumes. If the underlying file system is a legacy
9729 file system, this automatically degrades to creating a
9730 normal directory. Among others /var/lib/machines is now
9731 created like this at boot, should it be missing.
9733 * The directory /var/lib/containers/ has been deprecated and
9734 been replaced by /var/lib/machines. The term "machines" has
9735 been used in the systemd context as generic term for both
9736 VMs and containers, and hence appears more appropriate for
9737 this, as the directory can also contain raw images bootable
9740 * systemd-nspawn when invoked with -M but without --directory=
9741 or --image= is now capable of searching for the container
9742 root directory, subvolume or disk image automatically, in
9743 /var/lib/machines. systemd-nspawn@.service has been updated
9744 to make use of this, thus allowing it to be used for raw
9747 * A new machines.target unit has been introduced that is
9748 supposed to group all containers/VMs invoked as services on
9749 the system. systemd-nspawn@.service has been updated to
9750 integrate with that.
9752 * machinectl gained a new "start" command, for invoking a
9753 container as a service. "machinectl start foo" is mostly
9754 equivalent to "systemctl start systemd-nspawn@foo.service",
9755 but handles escaping in a nicer way.
9757 * systemd-nspawn will now mount most of the cgroupfs tree
9758 read-only into each container, with the exception of the
9759 container's own subtree in the name=systemd hierarchy.
9761 * journald now sets the special FS_NOCOW file flag for its
9762 journal files. This should improve performance on btrfs, by
9763 avoiding heavy fragmentation when journald's write-pattern
9764 is used on COW file systems. It degrades btrfs' data
9765 integrity guarantees for the files to the same levels as for
9766 ext3/ext4 however. This should be OK though as journald does
9767 its own data integrity checks and all its objects are
9768 checksummed on disk. Also, journald should handle btrfs disk
9769 full events a lot more gracefully now, by processing SIGBUS
9770 errors, and not relying on fallocate() anymore.
9772 * When journald detects that journal files it is writing to
9773 have been deleted it will immediately start new journal
9776 * systemd now provides a way to store file descriptors
9777 per-service in PID 1. This is useful for daemons to ensure
9778 that fds they require are not lost during a daemon
9779 restart. The fds are passed to the daemon on the next
9780 invocation in the same way socket activation fds are
9781 passed. This is now used by journald to ensure that the
9782 various sockets connected to all the system's stdout/stderr
9783 are not lost when journald is restarted. File descriptors
9784 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
9785 an extension to sd_notify(). Note that a limit is enforced
9786 on the number of fds a service can store in PID 1, and it
9787 defaults to 0, so that no fds may be stored, unless this is
9788 explicitly turned on.
9790 * The default TERM variable to use for units connected to a
9791 terminal, when no other value is explicitly is set is now
9792 vt220 rather than vt102. This should be fairly safe still,
9793 but allows PgUp/PgDn work.
9795 * The /etc/crypttab option header= as known from Debian is now
9798 * "loginctl user-status" and "loginctl session-status" will
9799 now show the last 10 lines of log messages of the
9800 user/session following the status output. Similar,
9801 "machinectl status" will show the last 10 log lines
9802 associated with a virtual machine or container
9803 service. (Note that this is usually not the log messages
9804 done in the VM/container itself, but simply what the
9805 container manager logs. For nspawn this includes all console
9808 * "loginctl session-status" without further argument will now
9809 show the status of the session of the caller. Similar,
9810 "lock-session", "unlock-session", "activate",
9811 "enable-linger", "disable-linger" may now be called without
9812 session/user parameter in which case they apply to the
9813 caller's session/user.
9815 * An X11 session scriptlet is now shipped that uploads
9816 $DISPLAY and $XAUTHORITY into the environment of the systemd
9817 --user daemon if a session begins. This should improve
9818 compatibility with X11 enabled applications run as systemd
9821 * Generators are now subject to masking via /etc and /run, the
9822 same way as unit files.
9824 * networkd .network files gained support for configuring
9825 per-link IPv4/IPv6 packet forwarding as well as IPv4
9826 masquerading. This is by default turned on for veth links to
9827 containers, as registered by systemd-nspawn. This means that
9828 nspawn containers run with --network-veth will now get
9829 automatic routed access to the host's networks without any
9830 further configuration or setup, as long as networkd runs on
9833 * systemd-nspawn gained the --port= (-p) switch to expose TCP
9834 or UDP posts of a container on the host. With this in place
9835 it is possible to run containers with private veth links
9836 (--network-veth), and have their functionality exposed on
9837 the host as if their services were running directly on the
9840 * systemd-nspawn's --network-veth switch now gained a short
9841 version "-n", since with the changes above it is now truly
9842 useful out-of-the-box. The systemd-nspawn@.service has been
9843 updated to make use of it too by default.
9845 * systemd-nspawn will now maintain a per-image R/W lock, to
9846 ensure that the same image is not started more than once
9847 writable. (It's OK to run an image multiple times
9848 simultaneously in read-only mode.)
9850 * systemd-nspawn's --image= option is now capable of
9851 dissecting and booting MBR and GPT disk images that contain
9852 only a single active Linux partition. Previously it
9853 supported only GPT disk images with proper GPT type
9854 IDs. This allows running cloud images from major
9855 distributions directly with systemd-nspawn, without
9858 * In addition to collecting mouse dpi data in the udev
9859 hardware database, there's now support for collecting angle
9860 information for mouse scroll wheels. The database is
9861 supposed to guarantee similar scrolling behavior on mice
9862 that it knows about. There's also support for collecting
9863 information about Touchpad types.
9865 * udev's input_id built-in will now also collect touch screen
9866 dimension data and attach it to probed devices.
9868 * /etc/os-release gained support for a Distribution Privacy
9871 * networkd gained support for creating "ipvlan", "gretap",
9872 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
9874 * systemd-tmpfiles gained support for "a" lines for setting
9877 * systemd-nspawn will now mount /tmp in the container to
9878 tmpfs, automatically.
9880 * systemd now exposes the memory.usage_in_bytes cgroup
9881 attribute and shows it for each service in the "systemctl
9882 status" output, if available.
9884 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
9885 immediate reboot is triggered. This useful if shutdown is
9886 hung and is unable to complete, to expedite the
9887 operation. Note that this kind of reboot will still unmount
9888 all file systems, and hence should not result in fsck being
9891 * A .device unit for an optical block device will now be
9892 considered active only when a medium is in the drive. Also,
9893 mount units are now bound to their backing devices thus
9894 triggering automatic unmounting when devices become
9895 unavailable. With this in place systemd will now
9896 automatically unmount left-over mounts when a CD-ROM is
9897 ejected or an USB stick is yanked from the system.
9899 * networkd-wait-online now has support for waiting for
9900 specific interfaces only (with globbing), and for giving up
9901 after a configurable timeout.
9903 * networkd now exits when idle. It will be automatically
9904 restarted as soon as interfaces show up, are removed or
9905 change state. networkd will stay around as long as there is
9906 at least one DHCP state machine or similar around, that keep
9909 * networkd may now configure IPv6 link-local addressing in
9910 addition to IPv4 link-local addressing.
9912 * The IPv6 "token" for use in SLAAC may now be configured for
9913 each .network interface in networkd.
9915 * Routes configured with networkd may now be assigned a scope
9918 * networkd's [Match] sections now support globbing and lists
9919 of multiple space-separated matches per item.
9921 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
9922 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
9923 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
9924 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
9925 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
9926 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
9927 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
9928 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
9929 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
9930 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
9931 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
9932 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
9933 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
9934 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
9935 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
9936 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
9937 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
9938 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
9939 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
9940 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
9941 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
9942 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
9943 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
9944 Hoffmann, Zbigniew Jędrzejewski-Szmek
9946 — Berlin, 2015-02-16
9950 * When querying unit file enablement status (for example via
9951 "systemctl is-enabled"), a new state "indirect" is now known
9952 which indicates that a unit might not be enabled itself, but
9953 another unit listed in its Also= setting might be.
9955 * Similar to the various existing ConditionXYZ= settings for
9956 units, there are now matching AssertXYZ= settings. While
9957 failing conditions cause a unit to be skipped, but its job
9958 to succeed, failing assertions declared like this will cause
9959 a unit start operation and its job to fail.
9961 * hostnamed now knows a new chassis type "embedded".
9963 * systemctl gained a new "edit" command. When used on a unit
9964 file, this allows extending unit files with .d/ drop-in
9965 configuration snippets or editing the full file (after
9966 copying it from /usr/lib to /etc). This will invoke the
9967 user's editor (as configured with $EDITOR), and reload the
9968 modified configuration after editing.
9970 * "systemctl status" now shows the suggested enablement state
9971 for a unit, as declared in the (usually vendor-supplied)
9972 system preset files.
9974 * nss-myhostname will now resolve the single-label hostname
9975 "gateway" to the locally configured default IP routing
9976 gateways, ordered by their metrics. This assigns a stable
9977 name to the used gateways, regardless which ones are
9978 currently configured. Note that the name will only be
9979 resolved after all other name sources (if nss-myhostname is
9980 configured properly) and should hence not negatively impact
9981 systems that use the single-label hostname "gateway" in
9984 * systemd-inhibit now allows filtering by mode when listing
9987 * Scope and service units gained a new "Delegate" boolean
9988 property, which, when set, allows processes running inside the
9989 unit to further partition resources. This is primarily
9990 useful for systemd user instances as well as container
9993 * journald will now pick up audit messages directly from
9994 the kernel, and log them like any other log message. The
9995 audit fields are split up and fully indexed. This means that
9996 journalctl in many ways is now a (nicer!) alternative to
9997 ausearch, the traditional audit client. Note that this
9998 implements only a minimal audit client. If you want the
9999 special audit modes like reboot-on-log-overflow, please use
10000 the traditional auditd instead, which can be used in
10001 parallel to journald.
10003 * The ConditionSecurity= unit file option now understands the
10004 special string "audit" to check whether auditing is
10007 * journalctl gained two new commands --vacuum-size= and
10008 --vacuum-time= to delete old journal files until the
10009 remaining ones take up no more than the specified size on disk,
10010 or are not older than the specified time.
10012 * A new, native PPPoE library has been added to sd-network,
10013 systemd's library of light-weight networking protocols. This
10014 library will be used in a future version of networkd to
10015 enable PPPoE communication without an external pppd daemon.
10017 * The busctl tool now understands a new "capture" verb that
10018 works similar to "monitor", but writes a packet capture
10019 trace to STDOUT that can be redirected to a file which is
10020 compatible with libcap's capture file format. This can then
10021 be loaded in Wireshark and similar tools to inspect bus
10024 * The busctl tool now understands a new "tree" verb that shows
10025 the object trees of a specific service on the bus, or of all
10028 * The busctl tool now understands a new "introspect" verb that
10029 shows all interfaces and members of objects on the bus,
10030 including their signature and values. This is particularly
10031 useful to get more information about bus objects shown by
10032 the new "busctl tree" command.
10034 * The busctl tool now understands new verbs "call",
10035 "set-property" and "get-property" for invoking bus method
10036 calls, setting and getting bus object properties in a
10039 * busctl gained a new --augment-creds= argument that controls
10040 whether the tool shall augment credential information it
10041 gets from the bus with data from /proc, in a possibly
10044 * nspawn's --link-journal= switch gained two new values
10045 "try-guest" and "try-host" that work like "guest" and
10046 "host", but do not fail if the host has no persistent
10047 journaling enabled. -j is now equivalent to
10048 --link-journal=try-guest.
10050 * macvlan network devices created by nspawn will now have
10051 stable MAC addresses.
10053 * A new SmackProcessLabel= unit setting has been added, which
10054 controls the SMACK security label processes forked off by
10055 the respective unit shall use.
10057 * If compiled with --enable-xkbcommon, systemd-localed will
10058 verify x11 keymap settings by compiling the given keymap. It
10059 will spew out warnings if the compilation fails. This
10060 requires libxkbcommon to be installed.
10062 * When a coredump is collected, a larger number of metadata
10063 fields is now collected and included in the journal records
10064 created for it. More specifically, control group membership,
10065 environment variables, memory maps, working directory,
10066 chroot directory, /proc/$PID/status, and a list of open file
10067 descriptors is now stored in the log entry.
10069 * The udev hwdb now contains DPI information for mice. For
10072 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
10074 * All systemd programs that read standalone configuration
10075 files in /etc now also support a corresponding series of
10076 .conf.d configuration directories in /etc/, /run/,
10077 /usr/local/lib/, /usr/lib/, and (if configured with
10078 --enable-split-usr) /lib/. In particular, the following
10079 configuration files now have corresponding configuration
10080 directories: system.conf user.conf, logind.conf,
10081 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
10082 resolved.conf, timesyncd.conf, journal-remote.conf, and
10083 journal-upload.conf. Note that distributions should use the
10084 configuration directories in /usr/lib/; the directories in
10085 /etc/ are reserved for the system administrator.
10087 * systemd-rfkill will no longer take the rfkill device name
10088 into account when storing rfkill state on disk, as the name
10089 might be dynamically assigned and not stable. Instead, the
10090 ID_PATH udev variable combined with the rfkill type (wlan,
10091 bluetooth, …) is used.
10093 * A new service systemd-machine-id-commit.service has been
10094 added. When used on systems where /etc is read-only during
10095 boot, and /etc/machine-id is not initialized (but an empty
10096 file), this service will copy the temporary machine ID
10097 created as replacement into /etc after the system is fully
10098 booted up. This is useful for systems that are freshly
10099 installed with a non-initialized machine ID, but should get
10100 a fixed machine ID for subsequent boots.
10102 * networkd's .netdev files now provide a large set of
10103 configuration parameters for VXLAN devices. Similarly, the
10104 bridge port cost parameter is now configurable in .network
10105 files. There's also new support for configuring IP source
10106 routing. networkd .link files gained support for a new
10107 OriginalName= match that is useful to match against the
10108 original interface name the kernel assigned. .network files
10109 may include MTU= and MACAddress= fields for altering the MTU
10110 and MAC address while being connected to a specific network
10113 * The LUKS logic gained supported for configuring
10114 UUID-specific key files. There's also new support for naming
10115 LUKS device from the kernel command line, using the new
10116 luks.name= argument.
10118 * Timer units may now be transiently created via the bus API
10119 (this was previously already available for scope and service
10120 units). In addition it is now possible to create multiple
10121 transient units at the same time with a single bus call. The
10122 "systemd-run" tool has been updated to make use of this for
10123 running commands on a specified time, in at(1)-style.
10125 * tmpfiles gained support for "t" lines, for assigning
10126 extended attributes to files. Among other uses this may be
10127 used to assign SMACK labels to files.
10129 Contributions from: Alin Rauta, Alison Chaiken, Andrej
10130 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
10131 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
10132 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
10133 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
10134 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
10135 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
10136 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
10137 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
10138 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
10139 Michael Biebl, Michael Chapman, Michael Marineau, Michal
10140 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
10141 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
10142 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
10143 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
10144 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10145 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
10146 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
10148 — Berlin, 2014-12-10
10152 * journalctl gained the new options -t/--identifier= to match
10153 on the syslog identifier (aka "tag"), as well as --utc to
10154 show log timestamps in the UTC timezone. journalctl now also
10155 accepts -n/--lines=all to disable line capping in a pager.
10157 * journalctl gained a new switch, --flush, that synchronously
10158 flushes logs from /run/log/journal to /var/log/journal if
10159 persistent storage is enabled. systemd-journal-flush.service
10160 now waits until the operation is complete.
10162 * Services can notify the manager before they start a reload
10163 (by sending RELOADING=1) or shutdown (by sending
10164 STOPPING=1). This allows the manager to track and show the
10165 internal state of daemons and closes a race condition when
10166 the process is still running but has closed its D-Bus
10169 * Services with Type=oneshot do not have to have any ExecStart
10172 * User units are now loaded also from
10173 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
10174 /run/systemd/user directory that was already previously
10175 supported, but is under the control of the user.
10177 * Job timeouts (i.e. timeouts on the time a job that is
10178 queued stays in the run queue) can now optionally result in
10179 immediate reboot or power-off actions (JobTimeoutAction= and
10180 JobTimeoutRebootArgument=). This is useful on ".target"
10181 units, to limit the maximum time a target remains
10182 undispatched in the run queue, and to trigger an emergency
10183 operation in such a case. This is now used by default to
10184 turn off the system if boot-up (as defined by everything in
10185 basic.target) hangs and does not complete for at least
10186 15min. Also, if power-off or reboot hang for at least 30min
10187 an immediate power-off/reboot operation is triggered. This
10188 functionality is particularly useful to increase reliability
10189 on embedded devices, but also on laptops which might
10190 accidentally get powered on when carried in a backpack and
10191 whose boot stays stuck in a hard disk encryption passphrase
10194 * systemd-logind can be configured to also handle lid switch
10195 events even when the machine is docked or multiple displays
10196 are attached (HandleLidSwitchDocked= option).
10198 * A helper binary and a service have been added which can be
10199 used to resume from hibernation in the initramfs. A
10200 generator will parse the resume= option on the kernel
10201 command line to trigger resume.
10203 * A user console daemon systemd-consoled has been
10204 added. Currently, it is a preview, and will so far open a
10205 single terminal on each session of the user marked as
10206 Desktop=systemd-console.
10208 * Route metrics can be specified for DHCP routes added by
10211 * The SELinux context of socket-activated services can be set
10212 from the information provided by the networking stack
10213 (SELinuxContextFromNet= option).
10215 * Userspace firmware loading support has been removed and
10216 the minimum supported kernel version is thus bumped to 3.7.
10218 * Timeout for udev workers has been increased from 1 to 3
10219 minutes, but a warning will be printed after 1 minute to
10220 help diagnose kernel modules that take a long time to load.
10222 * Udev rules can now remove tags on devices with TAG-="foobar".
10224 * systemd's readahead implementation has been removed. In many
10225 circumstances it didn't give expected benefits even for
10226 rotational disk drives and was becoming less relevant in the
10227 age of SSDs. As none of the developers has been using
10228 rotating media anymore, and nobody stepped up to actively
10229 maintain this component of systemd it has now been removed.
10231 * Swap units can use Options= to specify discard options.
10232 Discard options specified for swaps in /etc/fstab are now
10235 * Docker containers are now detected as a separate type of
10238 * The Password Agent protocol gained support for queries where
10239 the user input is shown, useful e.g. for user names.
10240 systemd-ask-password gained a new --echo option to turn that
10243 * The default sysctl.d/ snippets will now set:
10245 net.core.default_qdisc = fq_codel
10247 This selects Fair Queuing Controlled Delay as the default
10248 queuing discipline for network interfaces. fq_codel helps
10249 fight the network bufferbloat problem. It is believed to be
10250 a good default with no tuning required for most workloads.
10251 Downstream distributions may override this choice. On 10Gbit
10252 servers that do not do forwarding, "fq" may perform better.
10253 Systems without a good clocksource should use "pfifo_fast".
10255 * If kdbus is enabled during build a new option BusPolicy= is
10256 available for service units, that allows locking all service
10257 processes into a stricter bus policy, in order to limit
10258 access to various bus services, or even hide most of them
10259 from the service's view entirely.
10261 * networkctl will now show the .network and .link file
10262 networkd has applied to a specific interface.
10264 * sd-login gained a new API call sd_session_get_desktop() to
10265 query which desktop environment has been selected for a
10268 * UNIX utmp support is now compile-time optional to support
10269 legacy-free systems.
10271 * systemctl gained two new commands "add-wants" and
10272 "add-requires" for pulling in units from specific targets
10275 * If the word "rescue" is specified on the kernel command line
10276 the system will now boot into rescue mode (aka
10277 rescue.target), which was previously available only by
10278 specifying "1" or "systemd.unit=rescue.target" on the kernel
10279 command line. This new kernel command line option nicely
10280 mirrors the already existing "emergency" kernel command line
10283 * New kernel command line options mount.usr=, mount.usrflags=,
10284 mount.usrfstype= have been added that match root=, rootflags=,
10285 rootfstype= but allow mounting a specific file system to
10288 * The $NOTIFY_SOCKET is now also passed to control processes of
10289 services, not only the main process.
10291 * This version reenables support for fsck's -l switch. This
10292 means at least version v2.25 of util-linux is required for
10293 operation, otherwise dead-locks on device nodes may
10294 occur. Again: you need to update util-linux to at least
10295 v2.25 when updating systemd to v217.
10297 * The "multi-seat-x" tool has been removed from systemd, as
10298 its functionality has been integrated into X servers 1.16,
10299 and the tool is hence redundant. It is recommended to update
10300 display managers invoking this tool to simply invoke X
10301 directly from now on, again.
10303 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
10304 message flag has been added for all of systemd's polkit
10305 authenticated method calls has been added. In particular this
10306 now allows optional interactive authorization via polkit for
10307 many of PID1's privileged operations such as unit file
10308 enabling and disabling.
10310 * "udevadm hwdb --update" learnt a new switch "--usr" for
10311 placing the rebuilt hardware database in /usr instead of
10312 /etc. When used only hardware database entries stored in
10313 /usr will be used, and any user database entries in /etc are
10314 ignored. This functionality is useful for vendors to ship a
10315 pre-built database on systems where local configuration is
10316 unnecessary or unlikely.
10318 * Calendar time specifications in .timer units now also
10319 understand the strings "semi-annually", "quarterly" and
10320 "minutely" as shortcuts (in addition to the preexisting
10321 "annually", "hourly", …).
10323 * systemd-tmpfiles will now correctly create files in /dev
10324 at boot which are marked for creation only at boot. It is
10325 recommended to always create static device nodes with 'c!'
10326 and 'b!', so that they are created only at boot and not
10327 overwritten at runtime.
10329 * When the watchdog logic is used for a service (WatchdogSec=)
10330 and the watchdog timeout is hit the service will now be
10331 terminated with SIGABRT (instead of just SIGTERM), in order
10332 to make sure a proper coredump and backtrace is
10333 generated. This ensures that hanging services will result in
10334 similar coredump/backtrace behaviour as services that hit a
10335 segmentation fault.
10337 Contributions from: Andreas Henriksson, Andrei Borzenkov,
10338 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
10339 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
10340 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
10341 Herrmann, David Sommerseth, David Strauss, Emil Renner
10342 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
10343 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
10344 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
10345 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
10346 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
10347 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
10348 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
10349 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
10350 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
10351 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
10352 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
10353 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
10354 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
10355 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
10356 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
10357 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
10360 — Berlin, 2014-10-28
10364 * timedated no longer reads NTP implementation unit names from
10365 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
10366 implementations should add a
10368 Conflicts=systemd-timesyncd.service
10370 to their unit files to take over and replace systemd's NTP
10371 default functionality.
10373 * systemd-sysusers gained a new line type "r" for configuring
10374 which UID/GID ranges to allocate system users/groups
10375 from. Lines of type "u" may now add an additional column
10376 that specifies the home directory for the system user to be
10377 created. Also, systemd-sysusers may now optionally read user
10378 information from STDIN instead of a file. This is useful for
10379 invoking it from RPM preinst scriptlets that need to create
10380 users before the first RPM file is installed since these
10381 files might need to be owned by them. A new
10382 %sysusers_create_inline RPM macro has been introduced to do
10383 just that. systemd-sysusers now updates the shadow files as
10384 well as the user/group databases, which should enhance
10385 compatibility with certain tools like grpck.
10387 * A number of bus APIs of PID 1 now optionally consult polkit to
10388 permit access for otherwise unprivileged clients under certain
10389 conditions. Note that this currently doesn't support
10390 interactive authentication yet, but this is expected to be
10391 added eventually, too.
10393 * /etc/machine-info now has new fields for configuring the
10394 deployment environment of the machine, as well as the
10395 location of the machine. hostnamectl has been updated with
10396 new command to update these fields.
10398 * systemd-timesyncd has been updated to automatically acquire
10399 NTP server information from systemd-networkd, which might
10400 have been discovered via DHCP.
10402 * systemd-resolved now includes a caching DNS stub resolver
10403 and a complete LLMNR name resolution implementation. A new
10404 NSS module "nss-resolve" has been added which can be used
10405 instead of glibc's own "nss-dns" to resolve hostnames via
10406 systemd-resolved. Hostnames, addresses and arbitrary RRs may
10407 be resolved via systemd-resolved D-Bus APIs. In contrast to
10408 the glibc internal resolver systemd-resolved is aware of
10409 multi-homed system, and keeps DNS server and caches separate
10410 and per-interface. Queries are sent simultaneously on all
10411 interfaces that have DNS servers configured, in order to
10412 properly handle VPNs and local LANs which might resolve
10413 separate sets of domain names. systemd-resolved may acquire
10414 DNS server information from systemd-networkd automatically,
10415 which in turn might have discovered them via DHCP. A tool
10416 "systemd-resolve-host" has been added that may be used to
10417 query the DNS logic in resolved. systemd-resolved implements
10418 IDNA and automatically uses IDNA or UTF-8 encoding depending
10419 on whether classic DNS or LLMNR is used as transport. In the
10420 next releases we intend to add a DNSSEC and mDNS/DNS-SD
10421 implementation to systemd-resolved.
10423 * A new NSS module nss-mymachines has been added, that
10424 automatically resolves the names of all local registered
10425 containers to their respective IP addresses.
10427 * A new client tool "networkctl" for systemd-networkd has been
10428 added. It currently is entirely passive and will query
10429 networking configuration from udev, rtnetlink and networkd,
10430 and present it to the user in a very friendly
10431 way. Eventually, we hope to extend it to become a full
10432 control utility for networkd.
10434 * .socket units gained a new DeferAcceptSec= setting that
10435 controls the kernels' TCP_DEFER_ACCEPT sockopt for
10436 TCP. Similarly, support for controlling TCP keep-alive
10437 settings has been added (KeepAliveTimeSec=,
10438 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
10439 turning off Nagle's algorithm on TCP has been added
10442 * logind learned a new session type "web", for use in projects
10443 like Cockpit which register web clients as PAM sessions.
10445 * timer units with at least one OnCalendar= setting will now
10446 be started only after time-sync.target has been
10447 reached. This way they will not elapse before the system
10448 clock has been corrected by a local NTP client or
10449 similar. This is particular useful on RTC-less embedded
10450 machines, that come up with an invalid system clock.
10452 * systemd-nspawn's --network-veth= switch should now result in
10453 stable MAC addresses for both the outer and the inner side
10456 * systemd-nspawn gained a new --volatile= switch for running
10457 container instances with /etc or /var unpopulated.
10459 * The kdbus client code has been updated to use the new Linux
10460 3.17 memfd subsystem instead of the old kdbus-specific one.
10462 * systemd-networkd's DHCP client and server now support
10463 FORCERENEW. There are also new configuration options to
10464 configure the vendor client identifier and broadcast mode
10467 * systemd will no longer inform the kernel about the current
10468 timezone, as this is necessarily incorrect and racy as the
10469 kernel has no understanding of DST and similar
10470 concepts. This hence means FAT timestamps will be always
10471 considered UTC, similar to what Android is already
10472 doing. Also, when the RTC is configured to the local time
10473 (rather than UTC) systemd will never synchronize back to it,
10474 as this might confuse Windows at a later boot.
10476 * systemd-analyze gained a new command "verify" for offline
10477 validation of unit files.
10479 * systemd-networkd gained support for a couple of additional
10480 settings for bonding networking setups. Also, the metric for
10481 statically configured routes may now be configured. For
10482 network interfaces where this is appropriate the peer IP
10483 address may now be configured.
10485 * systemd-networkd's DHCP client will no longer request
10486 broadcasting by default, as this tripped up some networks.
10487 For hardware where broadcast is required the feature should
10488 be switched back on using RequestBroadcast=yes.
10490 * systemd-networkd will now set up IPv4LL addresses (when
10491 enabled) even if DHCP is configured successfully.
10493 * udev will now default to respect network device names given
10494 by the kernel when the kernel indicates that these are
10495 predictable. This behavior can be tweaked by changing
10496 NamePolicy= in the relevant .link file.
10498 * A new library systemd-terminal has been added that
10499 implements full TTY stream parsing and rendering. This
10500 library is supposed to be used later on for implementing a
10501 full userspace VT subsystem, replacing the current kernel
10504 * A new tool systemd-journal-upload has been added to push
10505 journal data to a remote system running
10506 systemd-journal-remote.
10508 * journald will no longer forward all local data to another
10509 running syslog daemon. This change has been made because
10510 rsyslog (which appears to be the most commonly used syslog
10511 implementation these days) no longer makes use of this, and
10512 instead pulls the data out of the journal on its own. Since
10513 forwarding the messages to a non-existent syslog server is
10514 more expensive than we assumed we have now turned this
10515 off. If you run a syslog server that is not a recent rsyslog
10516 version, you have to turn this option on again
10517 (ForwardToSyslog= in journald.conf).
10519 * journald now optionally supports the LZ4 compressor for
10520 larger journal fields. This compressor should perform much
10521 better than XZ which was the previous default.
10523 * machinectl now shows the IP addresses of local containers,
10524 if it knows them, plus the interface name of the container.
10526 * A new tool "systemd-escape" has been added that makes it
10527 easy to escape strings to build unit names and similar.
10529 * sd_notify() messages may now include a new ERRNO= field
10530 which is parsed and collected by systemd and shown among the
10531 "systemctl status" output for a service.
10533 * A new component "systemd-firstboot" has been added that
10534 queries the most basic systemd information (timezone,
10535 hostname, root password) interactively on first
10536 boot. Alternatively it may also be used to provision these
10537 things offline on OS images installed into directories.
10539 * The default sysctl.d/ snippets will now set
10541 net.ipv4.conf.default.promote_secondaries=1
10543 This has the benefit of no flushing secondary IP addresses
10544 when primary addresses are removed.
10546 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
10547 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
10548 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
10549 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
10550 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
10551 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
10552 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
10553 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
10554 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
10555 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
10556 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
10557 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
10558 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
10559 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
10560 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
10562 — Berlin, 2014-08-19
10566 * A new tool systemd-sysusers has been added. This tool
10567 creates system users and groups in /etc/passwd and
10568 /etc/group, based on static declarative system user/group
10569 definitions in /usr/lib/sysusers.d/. This is useful to
10570 enable factory resets and volatile systems that boot up with
10571 an empty /etc directory, and thus need system users and
10572 groups created during early boot. systemd now also ships
10573 with two default sysusers.d/ files for the most basic
10574 users and groups systemd and the core operating system
10577 * A new tmpfiles snippet has been added that rebuilds the
10578 essential files in /etc on boot, should they be missing.
10580 * A directive for ensuring automatic clean-up of
10581 /var/cache/man/ has been removed from the default
10582 configuration. This line should now be shipped by the man
10583 implementation. The necessary change has been made to the
10584 man-db implementation. Note that you need to update your man
10585 implementation to one that ships this line, otherwise no
10586 automatic clean-up of /var/cache/man will take place.
10588 * A new condition ConditionNeedsUpdate= has been added that
10589 may conditionalize services to only run when /etc or /var
10590 are "older" than the vendor operating system resources in
10591 /usr. This is useful for reconstructing or updating /etc
10592 after an offline update of /usr or a factory reset, on the
10593 next reboot. Services that want to run once after such an
10594 update or reset should use this condition and order
10595 themselves before the new systemd-update-done.service, which
10596 will mark the two directories as fully updated. A number of
10597 service files have been added making use of this, to rebuild
10598 the udev hardware database, the journald message catalog and
10599 dynamic loader cache (ldconfig). The systemd-sysusers tool
10600 described above also makes use of this now. With this in
10601 place it is now possible to start up a minimal operating
10602 system with /etc empty cleanly. For more information on the
10603 concepts involved see this recent blog story:
10605 http://0pointer.de/blog/projects/stateless.html
10607 * A new system group "input" has been introduced, and all
10608 input device nodes get this group assigned. This is useful
10609 for system-level software to get access to input devices. It
10610 complements what is already done for "audio" and "video".
10612 * systemd-networkd learnt minimal DHCPv4 server support in
10613 addition to the existing DHCPv4 client support. It also
10614 learnt DHCPv6 client and IPv6 Router Solicitation client
10615 support. The DHCPv4 client gained support for static routes
10616 passed in from the server. Note that the [DHCPv4] section
10617 known in older systemd-networkd versions has been renamed to
10618 [DHCP] and is now also used by the DHCPv6 client. Existing
10619 .network files using settings of this section should be
10620 updated, though compatibility is maintained. Optionally, the
10621 client hostname may now be sent to the DHCP server.
10623 * networkd gained support for vxlan virtual networks as well
10624 as tun/tap and dummy devices.
10626 * networkd gained support for automatic allocation of address
10627 ranges for interfaces from a system-wide pool of
10628 addresses. This is useful for dynamically managing a large
10629 number of interfaces with a single network configuration
10630 file. In particular this is useful to easily assign
10631 appropriate IP addresses to the veth links of a large number
10632 of nspawn instances.
10634 * RPM macros for processing sysusers, sysctl and binfmt
10635 drop-in snippets at package installation time have been
10638 * The /etc/os-release file should now be placed in
10639 /usr/lib/os-release. The old location is automatically
10640 created as symlink. /usr/lib is the more appropriate
10641 location of this file, since it shall actually describe the
10642 vendor operating system shipped in /usr, and not the
10643 configuration stored in /etc.
10645 * .mount units gained a new boolean SloppyOptions= setting
10646 that maps to mount(8)'s -s option which enables permissive
10647 parsing of unknown mount options.
10649 * tmpfiles learnt a new "L+" directive which creates a symlink
10650 but (unlike "L") deletes a pre-existing file first, should
10651 it already exist and not already be the correct
10652 symlink. Similarly, "b+", "c+" and "p+" directives have been
10653 added as well, which create block and character devices, as
10654 well as fifos in the filesystem, possibly removing any
10655 pre-existing files of different types.
10657 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
10658 'argument' field (which so far specified the source to
10659 symlink/copy the files from) is now optional. If omitted the
10660 same file os copied from /usr/share/factory/ suffixed by the
10661 full destination path. This is useful for populating /etc
10662 with essential files, by copying them from vendor defaults
10663 shipped in /usr/share/factory/etc.
10665 * A new command "systemctl preset-all" has been added that
10666 applies the service preset settings to all installed unit
10667 files. A new switch --preset-mode= has been added that
10668 controls whether only enable or only disable operations
10671 * A new command "systemctl is-system-running" has been added
10672 that allows checking the overall state of the system, for
10673 example whether it is fully up and running.
10675 * When the system boots up with an empty /etc, the equivalent
10676 to "systemctl preset-all" is executed during early boot, to
10677 make sure all default services are enabled after a factory
10680 * systemd now contains a minimal preset file that enables the
10681 most basic services systemd ships by default.
10683 * Unit files' [Install] section gained a new DefaultInstance=
10684 field for defining the default instance to create if a
10685 template unit is enabled with no instance specified.
10687 * A new passive target cryptsetup-pre.target has been added
10688 that may be used by services that need to make they run and
10689 finish before the first LUKS cryptographic device is set up.
10691 * The /dev/loop-control and /dev/btrfs-control device nodes
10692 are now owned by the "disk" group by default, opening up
10693 access to this group.
10695 * systemd-coredump will now automatically generate a
10696 stack trace of all core dumps taking place on the system,
10697 based on elfutils' libdw library. This stack trace is logged
10700 * systemd-coredump may now optionally store coredumps directly
10701 on disk (in /var/lib/systemd/coredump, possibly compressed),
10702 instead of storing them unconditionally in the journal. This
10703 mode is the new default. A new configuration file
10704 /etc/systemd/coredump.conf has been added to configure this
10705 and other parameters of systemd-coredump.
10707 * coredumpctl gained a new "info" verb to show details about a
10708 specific coredump. A new switch "-1" has also been added
10709 that makes sure to only show information about the most
10710 recent entry instead of all entries. Also, as the tool is
10711 generally useful now the "systemd-" prefix of the binary
10712 name has been removed. Distributions that want to maintain
10713 compatibility with the old name should add a symlink from
10714 the old name to the new name.
10716 * journald's SplitMode= now defaults to "uid". This makes sure
10717 that unprivileged users can access their own coredumps with
10718 coredumpctl without restrictions.
10720 * New kernel command line options "systemd.wants=" (for
10721 pulling an additional unit during boot), "systemd.mask="
10722 (for masking a specific unit for the boot), and
10723 "systemd.debug-shell" (for enabling the debug shell on tty9)
10724 have been added. This is implemented in the new generator
10725 "systemd-debug-generator".
10727 * systemd-nspawn will now by default filter a couple of
10728 syscalls for containers, among them those required for
10729 kernel module loading, direct x86 IO port access, swap
10730 management, and kexec. Most importantly though
10731 open_by_handle_at() is now prohibited for containers,
10732 closing a hole similar to a recently discussed vulnerability
10733 in docker regarding access to files on file hierarchies the
10734 container should normally not have access to. Note that, for
10735 nspawn, we generally make no security claims anyway (and
10736 this is explicitly documented in the man page), so this is
10737 just a fix for one of the most obvious problems.
10739 * A new man page file-hierarchy(7) has been added that
10740 contains a minimized, modernized version of the file system
10741 layout systemd expects, similar in style to the FHS
10742 specification or hier(5). A new tool systemd-path(1) has
10743 been added to query many of these paths for the local
10746 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
10747 longer done. Since the directory now has a per-user size
10748 limit, and is cleaned on logout this appears unnecessary,
10749 in particular since this now brings the lifecycle of this
10750 directory closer in line with how IPC objects are handled.
10752 * systemd.pc now exports a number of additional directories,
10753 including $libdir (which is useful to identify the library
10754 path for the primary architecture of the system), and a
10755 couple of drop-in directories.
10757 * udev's predictable network interface names now use the dev_port
10758 sysfs attribute, introduced in linux 3.15 instead of dev_id to
10759 distinguish between ports of the same PCI function. dev_id should
10760 only be used for ports using the same HW address, hence the need
10763 * machined has been updated to export the OS version of a
10764 container (read from /etc/os-release and
10765 /usr/lib/os-release) on the bus. This is now shown in
10766 "machinectl status" for a machine.
10768 * A new service setting RestartForceExitStatus= has been
10769 added. If configured to a set of exit signals or process
10770 return values, the service will be restarted when the main
10771 daemon process exits with any of them, regardless of the
10774 * systemctl's -H switch for connecting to remote systemd
10775 machines has been extended so that it may be used to
10776 directly connect to a specific container on the
10777 host. "systemctl -H root@foobar:waldi" will now connect as
10778 user "root" to host "foobar", and then proceed directly to
10779 the container named "waldi". Note that currently you have to
10780 authenticate as user "root" for this to work, as entering
10781 containers is a privileged operation.
10783 Contributions from: Andreas Henriksson, Benjamin Steinwender,
10784 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
10785 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
10786 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
10787 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
10788 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
10789 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
10790 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
10791 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
10792 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
10793 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
10794 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
10796 — Berlin, 2014-07-03
10800 * As an experimental feature, udev now tries to lock the
10801 disk device node (flock(LOCK_SH|LOCK_NB)) while it
10802 executes events for the disk or any of its partitions.
10803 Applications like partitioning programs can lock the
10804 disk device node (flock(LOCK_EX)) and claim temporary
10805 device ownership that way; udev will entirely skip all event
10806 handling for this disk and its partitions. If the disk
10807 was opened for writing, the close will trigger a partition
10808 table rescan in udev's "watch" facility, and if needed
10809 synthesize "change" events for the disk and all its partitions.
10810 This is now unconditionally enabled, and if it turns out to
10811 cause major problems, we might turn it on only for specific
10812 devices, or might need to disable it entirely. Device Mapper
10813 devices are excluded from this logic.
10815 * We temporarily dropped the "-l" switch for fsck invocations,
10816 since they collide with the flock() logic above. util-linux
10817 upstream has been changed already to avoid this conflict,
10818 and we will re-add "-l" as soon as util-linux with this
10819 change has been released.
10821 * The dependency on libattr has been removed. Since a long
10822 time, the extended attribute calls have moved to glibc, and
10823 libattr is thus unnecessary.
10825 * Virtualization detection works without privileges now. This
10826 means the systemd-detect-virt binary no longer requires
10827 CAP_SYS_PTRACE file capabilities, and our daemons can run
10828 with fewer privileges.
10830 * systemd-networkd now runs under its own "systemd-network"
10831 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
10832 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
10833 loses the ability to write to files owned by root this way.
10835 * Similarly, systemd-resolved now runs under its own
10836 "systemd-resolve" user with no capabilities remaining.
10838 * Similarly, systemd-bus-proxyd now runs under its own
10839 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
10841 * systemd-networkd gained support for setting up "veth"
10842 virtual Ethernet devices for container connectivity, as well
10843 as GRE and VTI tunnels.
10845 * systemd-networkd will no longer automatically attempt to
10846 manually load kernel modules necessary for certain tunnel
10847 transports. Instead, it is assumed the kernel loads them
10848 automatically when required. This only works correctly on
10849 very new kernels. On older kernels, please consider adding
10850 the kernel modules to /etc/modules-load.d/ as a work-around.
10852 * The resolv.conf file systemd-resolved generates has been
10853 moved to /run/systemd/resolve/. If you have a symlink from
10854 /etc/resolv.conf, it might be necessary to correct it.
10856 * Two new service settings, ProtectHome= and ProtectSystem=,
10857 have been added. When enabled, they will make the user data
10858 (such as /home) inaccessible or read-only and the system
10859 (such as /usr) read-only, for specific services. This allows
10860 very light-weight per-service sandboxing to avoid
10861 modifications of user data or system files from
10862 services. These two new switches have been enabled for all
10863 of systemd's long-running services, where appropriate.
10865 * Socket units gained new SocketUser= and SocketGroup=
10866 settings to set the owner user and group of AF_UNIX sockets
10867 and FIFOs in the file system.
10869 * Socket units gained a new RemoveOnStop= setting. If enabled,
10870 all FIFOS and sockets in the file system will be removed
10871 when the specific socket unit is stopped.
10873 * Socket units gained a new Symlinks= setting. It takes a list
10874 of symlinks to create to file system sockets or FIFOs
10875 created by the specific Unix sockets. This is useful to
10876 manage symlinks to socket nodes with the same lifecycle as
10879 * The /dev/log socket and /dev/initctl FIFO have been moved to
10880 /run, and have been replaced by symlinks. This allows
10881 connecting to these facilities even if PrivateDevices=yes is
10882 used for a service (which makes /dev/log itself unavailable,
10883 but /run is left). This also has the benefit of ensuring
10884 that /dev only contains device nodes, directories and
10885 symlinks, and nothing else.
10887 * sd-daemon gained two new calls sd_pid_notify() and
10888 sd_pid_notifyf(). They are similar to sd_notify() and
10889 sd_notifyf(), but allow overriding of the source PID of
10890 notification messages if permissions permit this. This is
10891 useful to send notify messages on behalf of a different
10892 process (for example, the parent process). The
10893 systemd-notify tool has been updated to make use of this
10894 when sending messages (so that notification messages now
10895 originate from the shell script invoking systemd-notify and
10896 not the systemd-notify process itself. This should minimize
10897 a race where systemd fails to associate notification
10898 messages to services when the originating process already
10901 * A new "on-abnormal" setting for Restart= has been added. If
10902 set, it will result in automatic restarts on all "abnormal"
10903 reasons for a process to exit, which includes unclean
10904 signals, core dumps, timeouts and watchdog timeouts, but
10905 does not include clean and unclean exit codes or clean
10906 signals. Restart=on-abnormal is an alternative for
10907 Restart=on-failure for services that shall be able to
10908 terminate and avoid restarts on certain errors, by
10909 indicating so with an unclean exit code. Restart=on-failure
10910 or Restart=on-abnormal is now the recommended setting for
10911 all long-running services.
10913 * If the InaccessibleDirectories= service setting points to a
10914 mount point (or if there are any submounts contained within
10915 it), it is now attempted to completely unmount it, to make
10916 the file systems truly unavailable for the respective
10919 * The ReadOnlyDirectories= service setting and
10920 systemd-nspawn's --read-only parameter are now recursively
10921 applied to all submounts, too.
10923 * Mount units may now be created transiently via the bus APIs.
10925 * The support for SysV and LSB init scripts has been removed
10926 from the systemd daemon itself. Instead, it is now
10927 implemented as a generator that creates native systemd units
10928 from these scripts when needed. This enables us to remove a
10929 substantial amount of legacy code from PID 1, following the
10930 fact that many distributions only ship a very small number
10931 of LSB/SysV init scripts nowadays.
10933 * Privileged Xen (dom0) domains are not considered
10934 virtualization anymore by the virtualization detection
10935 logic. After all, they generally have unrestricted access to
10936 the hardware and usually are used to manage the unprivileged
10939 * systemd-tmpfiles gained a new "C" line type, for copying
10940 files or entire directories.
10942 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
10943 lines. So far, they have been non-globbing versions of the
10944 latter, and have thus been redundant. In future, it is
10945 recommended to only use "z". "m" has hence been removed
10946 from the documentation, even though it stays supported.
10948 * A tmpfiles snippet to recreate the most basic structure in
10949 /var has been added. This is enough to create the /var/run →
10950 /run symlink and create a couple of structural
10951 directories. This allows systems to boot up with an empty or
10952 volatile /var. Of course, while with this change, the core OS
10953 now is capable with dealing with a volatile /var, not all
10954 user services are ready for it. However, we hope that sooner
10955 or later, many service daemons will be changed upstream so
10956 that they are able to automatically create their necessary
10957 directories in /var at boot, should they be missing. This is
10958 the first step to allow state-less systems that only require
10959 the vendor image for /usr to boot.
10961 * systemd-nspawn has gained a new --tmpfs= switch to mount an
10962 empty tmpfs instance to a specific directory. This is
10963 particularly useful for making use of the automatic
10964 reconstruction of /var (see above), by passing --tmpfs=/var.
10966 * Access modes specified in tmpfiles snippets may now be
10967 prefixed with "~", which indicates that they shall be masked
10968 by whether the existing file or directory is currently
10969 writable, readable or executable at all. Also, if specified,
10970 the sgid/suid/sticky bits will be masked for all
10973 * A new passive target unit "network-pre.target" has been
10974 added which is useful for services that shall run before any
10975 network is configured, for example firewall scripts.
10977 * The "floppy" group that previously owned the /dev/fd*
10978 devices is no longer used. The "disk" group is now used
10979 instead. Distributions should probably deprecate usage of
10982 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
10983 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
10984 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
10985 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
10986 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
10987 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
10990 — Berlin, 2014-06-11
10994 * A new "systemd-timesyncd" daemon has been added for
10995 synchronizing the system clock across the network. It
10996 implements an SNTP client. In contrast to NTP
10997 implementations such as chrony or the NTP reference server,
10998 this only implements a client side, and does not bother with
10999 the full NTP complexity, focusing only on querying time from
11000 one remote server and synchronizing the local clock to
11001 it. Unless you intend to serve NTP to networked clients or
11002 want to connect to local hardware clocks, this simple NTP
11003 client should be more than appropriate for most
11004 installations. The daemon runs with minimal privileges, and
11005 has been hooked up with networkd to only operate when
11006 network connectivity is available. The daemon saves the
11007 current clock to disk every time a new NTP sync has been
11008 acquired, and uses this to possibly correct the system clock
11009 early at bootup, in order to accommodate for systems that
11010 lack an RTC such as the Raspberry Pi and embedded devices,
11011 and to make sure that time monotonically progresses on these
11012 systems, even if it is not always correct. To make use of
11013 this daemon, a new system user and group "systemd-timesync"
11014 needs to be created on installation of systemd.
11016 * The queue "seqnum" interface of libudev has been disabled, as
11017 it was generally incompatible with device namespacing as
11018 sequence numbers of devices go "missing" if the devices are
11019 part of a different namespace.
11021 * "systemctl list-timers" and "systemctl list-sockets" gained
11022 a --recursive switch for showing units of these types also
11023 for all local containers, similar in style to the already
11024 supported --recursive switch for "systemctl list-units".
11026 * A new RebootArgument= setting has been added for service
11027 units, which may be used to specify a kernel reboot argument
11028 to use when triggering reboots with StartLimitAction=.
11030 * A new FailureAction= setting has been added for service
11031 units which may be used to specify an operation to trigger
11032 when a service fails. This works similarly to
11033 StartLimitAction=, but unlike it, controls what is done
11034 immediately rather than only after several attempts to
11035 restart the service in question.
11037 * hostnamed got updated to also expose the kernel name,
11038 release, and version on the bus. This is useful for
11039 executing commands like hostnamectl with the -H switch.
11040 systemd-analyze makes use of this to properly display
11041 details when running non-locally.
11043 * The bootchart tool can now show cgroup information in the
11044 graphs it generates.
11046 * The CFS CPU quota cgroup attribute is now exposed for
11047 services. The new CPUQuota= switch has been added for this
11048 which takes a percentage value. Setting this will have the
11049 result that a service may never get more CPU time than the
11050 specified percentage, even if the machine is otherwise idle.
11052 * systemd-networkd learned IPIP and SIT tunnel support.
11054 * LSB init scripts exposing a dependency on $network will now
11055 get a dependency on network-online.target rather than simply
11056 network.target. This should bring LSB handling closer to
11057 what it was on SysV systems.
11059 * A new fsck.repair= kernel option has been added to control
11060 how fsck shall deal with unclean file systems at boot.
11062 * The (.ini) configuration file parser will now silently ignore
11063 sections whose names begin with "X-". This may be used to maintain
11064 application-specific extension sections in unit files.
11066 * machined gained a new API to query the IP addresses of
11067 registered containers. "machinectl status" has been updated
11068 to show these addresses in its output.
11070 * A new call sd_uid_get_display() has been added to the
11071 sd-login APIs for querying the "primary" session of a
11072 user. The "primary" session of the user is elected from the
11073 user's sessions and generally a graphical session is
11074 preferred over a text one.
11076 * A minimal systemd-resolved daemon has been added. It
11077 currently simply acts as a companion to systemd-networkd and
11078 manages resolv.conf based on per-interface DNS
11079 configuration, possibly supplied via DHCP. In the long run
11080 we hope to extend this into a local DNSSEC enabled DNS and
11083 * The systemd-networkd-wait-online tool is now enabled by
11084 default. It will delay network-online.target until a network
11085 connection has been configured. The tool primarily integrates
11086 with networkd, but will also make a best effort to make sense
11087 of network configuration performed in some other way.
11089 * Two new service options StartupCPUShares= and
11090 StartupBlockIOWeight= have been added that work similarly to
11091 CPUShares= and BlockIOWeight= however only apply during
11092 system startup. This is useful to prioritize certain services
11093 differently during bootup than during normal runtime.
11095 * hostnamed has been changed to prefer the statically
11096 configured hostname in /etc/hostname (unless set to
11097 'localhost' or empty) over any dynamic one supplied by
11098 dhcp. With this change, the rules for picking the hostname
11099 match more closely the rules of other configuration settings
11100 where the local administrator's configuration in /etc always
11101 overrides any other settings.
11103 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
11104 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
11105 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
11106 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
11107 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
11108 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
11109 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
11110 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
11111 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
11112 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
11113 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
11114 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
11115 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
11116 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
11117 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
11118 Lindskog, WaLyong Cho, Will Woods, Zbigniew
11121 — Beijing, 2014-05-28
11125 * When restoring the screen brightness at boot, stay away from
11126 the darkest setting or from the lowest 5% of the available
11127 range, depending on which is the larger value of both. This
11128 should effectively protect the user from rebooting into a
11129 black screen, should the brightness have been set to minimum
11132 * sd-login gained a new sd_machine_get_class() call to
11133 determine the class ("vm" or "container") of a machine
11134 registered with machined.
11136 * sd-login gained new calls
11137 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
11138 to query the identity of the peer of a local AF_UNIX
11139 connection. They operate similarly to their sd_pid_get_xyz()
11142 * PID 1 will now maintain a system-wide system state engine
11143 with the states "starting", "running", "degraded",
11144 "maintenance", "stopping". These states are bound to system
11145 startup, normal runtime, runtime with at least one failed
11146 service, rescue/emergency mode and system shutdown. This
11147 state is shown in the "systemctl status" output when no unit
11148 name is passed. It is useful to determine system state, in
11149 particularly when doing so for many systems or containers at
11152 * A new command "list-machines" has been added to "systemctl"
11153 that lists all local OS containers and shows their system
11154 state (see above), if systemd runs inside of them.
11156 * systemctl gained a new "-r" switch to recursively enumerate
11157 units on all local containers, when used with the
11158 "list-unit" command (which is the default one that is
11159 executed when no parameters are specified).
11161 * The GPT automatic partition discovery logic will now honour
11162 two GPT partition flags: one may be set on a partition to
11163 cause it to be mounted read-only, and the other may be set
11164 on a partition to ignore it during automatic discovery.
11166 * Two new GPT type UUIDs have been added for automatic root
11167 partition discovery, for 32-bit and 64-bit ARM. This is not
11168 particularly useful for discovering the root directory on
11169 these architectures during bare-metal boots (since UEFI is
11170 not common there), but still very useful to allow booting of
11171 ARM disk images in nspawn with the -i option.
11173 * MAC addresses of interfaces created with nspawn's
11174 --network-interface= switch will now be generated from the
11175 machine name, and thus be stable between multiple invocations
11178 * logind will now automatically remove all IPC objects owned
11179 by a user if she or he fully logs out. This makes sure that
11180 users who are logged out cannot continue to consume IPC
11181 resources. This covers SysV memory, semaphores and message
11182 queues as well as POSIX shared memory and message
11183 queues. Traditionally, SysV and POSIX IPC had no lifecycle
11184 limits. With this functionality, that is corrected. This may
11185 be turned off by using the RemoveIPC= switch of logind.conf.
11187 * The systemd-machine-id-setup and tmpfiles tools gained a
11188 --root= switch to operate on a specific root directory,
11191 * journald can now forward logged messages to the TTYs of all
11192 logged in users ("wall"). This is the default for all
11193 emergency messages now.
11195 * A new tool systemd-journal-remote has been added to stream
11196 journal log messages across the network.
11198 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
11199 controller trees are mounted into it. Note that the
11200 directories mounted beneath it are not read-only. This is a
11201 security measure and is particularly useful because glibc
11202 actually includes a search logic to pick any tmpfs it can
11203 find to implement shm_open() if /dev/shm is not available
11204 (which it might very well be in namespaced setups).
11206 * machinectl gained a new "poweroff" command to cleanly power
11207 down a local OS container.
11209 * The PrivateDevices= unit file setting will now also drop the
11210 CAP_MKNOD capability from the capability bound set, and
11211 imply DevicePolicy=closed.
11213 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
11214 comprehensively on all long-running systemd services where
11215 this is appropriate.
11217 * systemd-udevd will now run in a disassociated mount
11218 namespace. To mount directories from udev rules, make sure to
11219 pull in mount units via SYSTEMD_WANTS properties.
11221 * The kdbus support gained support for uploading policy into
11222 the kernel. sd-bus gained support for creating "monitoring"
11223 connections that can eavesdrop into all bus communication
11224 for debugging purposes.
11226 * Timestamps may now be specified in seconds since the UNIX
11227 epoch Jan 1st, 1970 by specifying "@" followed by the value
11230 * Native tcpwrap support in systemd has been removed. tcpwrap
11231 is old code, not really maintained anymore and has serious
11232 shortcomings, and better options such as firewalls
11233 exist. For setups that require tcpwrap usage, please
11234 consider invoking your socket-activated service via tcpd,
11235 like on traditional inetd.
11237 * A new system.conf configuration option
11238 DefaultTimerAccuracySec= has been added that controls the
11239 default AccuracySec= setting of .timer units.
11241 * Timer units gained a new WakeSystem= switch. If enabled,
11242 timers configured this way will cause the system to resume
11243 from system suspend (if the system supports that, which most
11246 * Timer units gained a new Persistent= switch. If enabled,
11247 timers configured this way will save to disk when they have
11248 been last triggered. This information is then used on next
11249 reboot to possible execute overdue timer events, that
11250 could not take place because the system was powered off.
11251 This enables simple anacron-like behaviour for timer units.
11253 * systemctl's "list-timers" will now also list the time a
11254 timer unit was last triggered in addition to the next time
11255 it will be triggered.
11257 * systemd-networkd will now assign predictable IPv4LL
11258 addresses to its local interfaces.
11260 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
11261 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
11262 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
11263 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
11264 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
11265 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
11266 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
11267 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
11270 — Berlin, 2014-03-25
11274 * A new unit file setting RestrictAddressFamilies= has been
11275 added to restrict which socket address families unit
11276 processes gain access to. This takes address family names
11277 like "AF_INET" or "AF_UNIX", and is useful to minimize the
11278 attack surface of services via exotic protocol stacks. This
11279 is built on seccomp system call filters.
11281 * Two new unit file settings RuntimeDirectory= and
11282 RuntimeDirectoryMode= have been added that may be used to
11283 manage a per-daemon runtime directories below /run. This is
11284 an alternative for setting up directory permissions with
11285 tmpfiles snippets, and has the advantage that the runtime
11286 directory's lifetime is bound to the daemon runtime and that
11287 the daemon starts up with an empty directory each time. This
11288 is particularly useful when writing services that drop
11289 privileges using the User= or Group= setting.
11291 * The DeviceAllow= unit setting now supports globbing for
11292 matching against device group names.
11294 * The systemd configuration file system.conf gained new
11295 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
11296 DefaultMemoryAccounting= to globally turn on/off accounting
11297 for specific resources (cgroups) for all units. These
11298 settings may still be overridden individually in each unit
11301 * systemd-gpt-auto-generator is now able to discover /srv and
11302 root partitions in addition to /home and swap partitions. It
11303 also supports LUKS-encrypted partitions now. With this in
11304 place, automatic discovery of partitions to mount following
11305 the Discoverable Partitions Specification
11306 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
11307 is now a lot more complete. This allows booting without
11308 /etc/fstab and without root= on the kernel command line on
11309 systems prepared appropriately.
11311 * systemd-nspawn gained a new --image= switch which allows
11312 booting up disk images and Linux installations on any block
11313 device that follow the Discoverable Partitions Specification
11314 (see above). This means that installations made with
11315 appropriately updated installers may now be started and
11316 deployed using container managers, completely
11317 unmodified. (We hope that libvirt-lxc will add support for
11318 this feature soon, too.)
11320 * systemd-nspawn gained a new --network-macvlan= setting to
11321 set up a private macvlan interface for the
11322 container. Similarly, systemd-networkd gained a new
11323 Kind=macvlan setting in .netdev files.
11325 * systemd-networkd now supports configuring local addresses
11328 * A new tool systemd-network-wait-online has been added to
11329 synchronously wait for network connectivity using
11332 * The sd-bus.h bus API gained a new sd_bus_track object for
11333 tracking the lifecycle of bus peers. Note that sd-bus.h is
11334 still not a public API though (unless you specify
11335 --enable-kdbus on the configure command line, which however
11336 voids your warranty and you get no API stability guarantee).
11338 * The $XDG_RUNTIME_DIR runtime directories for each user are
11339 now individual tmpfs instances, which has the benefit of
11340 introducing separate pools for each user, with individual
11341 size limits, and thus making sure that unprivileged clients
11342 can no longer negatively impact the system or other users by
11343 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
11344 RuntimeDirectorySize= has been introduced that allows
11345 controlling the default size limit for all users. It
11346 defaults to 10% of the available physical memory. This is no
11347 replacement for quotas on tmpfs though (which the kernel
11348 still does not support), as /dev/shm and /tmp are still
11349 shared resources used by both the system and unprivileged
11352 * logind will now automatically turn off automatic suspending
11353 on laptop lid close when more than one display is
11354 connected. This was previously expected to be implemented
11355 individually in desktop environments (such as GNOME),
11356 however has been added to logind now, in order to fix a
11357 boot-time race where a desktop environment might not have
11358 been started yet and thus not been able to take an inhibitor
11359 lock at the time where logind already suspends the system
11360 due to a closed lid.
11362 * logind will now wait at least 30s after each system
11363 suspend/resume cycle, and 3min after system boot before
11364 suspending the system due to a closed laptop lid. This
11365 should give USB docking stations and similar enough time to
11366 be probed and configured after system resume and boot in
11367 order to then act as suspend blocker.
11369 * systemd-run gained a new --property= setting which allows
11370 initialization of resource control properties (and others)
11371 for the created scope or service unit. Example: "systemd-run
11372 --property=BlockIOWeight=10 updatedb" may be used to run
11373 updatedb at a low block IO scheduling weight.
11375 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
11376 now also work in --scope mode.
11378 * When systemd is compiled with kdbus support, basic support
11379 for enforced policies is now in place. (Note that enabling
11380 kdbus still voids your warranty and no API compatibility
11381 promises are made.)
11383 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
11384 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
11385 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
11386 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
11387 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
11388 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
11389 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
11390 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
11391 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
11392 Zbigniew Jędrzejewski-Szmek
11394 — Berlin, 2014-03-12
11398 * systemd will now relabel /dev after loading the SMACK policy
11399 according to SMACK rules.
11401 * A new unit file option AppArmorProfile= has been added to
11402 set the AppArmor profile for the processes of a unit.
11404 * A new condition check ConditionArchitecture= has been added
11405 to conditionalize units based on the system architecture, as
11406 reported by uname()'s "machine" field.
11408 * systemd-networkd now supports matching on the system
11409 virtualization, architecture, kernel command line, hostname
11412 * logind is now a lot more aggressive when suspending the
11413 machine due to a closed laptop lid. Instead of acting only
11414 on the lid close action, it will continuously watch the lid
11415 status and act on it. This is useful for laptops where the
11416 power button is on the outside of the chassis so that it can
11417 be reached without opening the lid (such as the Lenovo
11418 Yoga). On those machines, logind will now immediately
11419 re-suspend the machine if the power button has been
11420 accidentally pressed while the laptop was suspended and in a
11421 backpack or similar.
11423 * logind will now watch SW_DOCK switches and inhibit reaction
11424 to the lid switch if it is pressed. This means that logind
11425 will not suspend the machine anymore if the lid is closed
11426 and the system is docked, if the laptop supports SW_DOCK
11427 notifications via the input layer. Note that ACPI docking
11428 stations do not generate this currently. Also note that this
11429 logic is usually not fully sufficient and Desktop
11430 Environments should take a lid switch inhibitor lock when an
11431 external display is connected, as systemd will not watch
11434 * nspawn will now make use of the devices cgroup controller by
11435 default, and only permit creation of and access to the usual
11436 API device nodes like /dev/null or /dev/random, as well as
11437 access to (but not creation of) the pty devices.
11439 * We will now ship a default .network file for
11440 systemd-networkd that automatically configures DHCP for
11441 network interfaces created by nspawn's --network-veth or
11442 --network-bridge= switches.
11444 * systemd will now understand the usual M, K, G, T suffixes
11445 according to SI conventions (i.e. to the base 1000) when
11446 referring to throughput and hardware metrics. It will stay
11447 with IEC conventions (i.e. to the base 1024) for software
11448 metrics, according to what is customary according to
11449 Wikipedia. We explicitly document which base applies for
11450 each configuration option.
11452 * The DeviceAllow= setting in unit files now supports a syntax to
11453 allow-list an entire group of devices node majors at once, based on
11454 the /proc/devices listing. For example, with the string "char-pts",
11455 it is now possible to allow-list all current and future pseudo-TTYs
11458 * sd-event learned a new "post" event source. Event sources of
11459 this type are triggered by the dispatching of any event
11460 source of a type that is not "post". This is useful for
11461 implementing clean-up and check event sources that are
11462 triggered by other work being done in the program.
11464 * systemd-networkd is no longer statically enabled, but uses
11465 the usual [Install] sections so that it can be
11466 enabled/disabled using systemctl. It still is enabled by
11469 * When creating a veth interface pair with systemd-nspawn, the
11470 host side will now be prefixed with "vb-" if
11471 --network-bridge= is used, and with "ve-" if --network-veth
11472 is used. This way, it is easy to distinguish these cases on
11473 the host, for example to apply different configuration to
11474 them with systemd-networkd.
11476 * The compatibility libraries for libsystemd-journal.so,
11477 libsystem-id128.so, libsystemd-login.so and
11478 libsystemd-daemon.so do not make use of IFUNC
11479 anymore. Instead, we now build libsystemd.so multiple times
11480 under these alternative names. This means that the footprint
11481 is drastically increased, but given that these are
11482 transitional compatibility libraries, this should not matter
11483 much. This change has been made necessary to support the ARM
11484 platform for these compatibility libraries, as the ARM
11485 toolchain is not really at the same level as the toolchain
11486 for other architectures like x86 and does not support
11487 IFUNC. Please make sure to use --enable-compat-libs only
11488 during a transitional period!
11490 * The .include syntax has been deprecated and is not documented
11491 anymore. Drop-in files in .d directories should be used instead.
11493 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
11494 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
11495 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
11496 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
11497 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
11498 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
11499 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
11500 Zbigniew Jędrzejewski-Szmek
11502 — Berlin, 2014-02-24
11506 * A new component "systemd-networkd" has been added that can
11507 be used to configure local network interfaces statically or
11508 via DHCP. It is capable of bringing up bridges, VLANs, and
11509 bonding. Currently, no hook-ups for interactive network
11510 configuration are provided. Use this for your initrd,
11511 container, embedded, or server setup if you need a simple,
11512 yet powerful, network configuration solution. This
11513 configuration subsystem is quite nifty, as it allows wildcard
11514 hotplug matching in interfaces. For example, with a single
11515 configuration snippet, you can configure that all Ethernet
11516 interfaces showing up are automatically added to a bridge,
11517 or similar. It supports link-sensing and more.
11519 * A new tool "systemd-socket-proxyd" has been added which can
11520 act as a bidirectional proxy for TCP sockets. This is
11521 useful for adding socket activation support to services that
11522 do not actually support socket activation, including virtual
11523 machines and the like.
11525 * Add a new tool to save/restore rfkill state on
11528 * Save/restore state of keyboard backlights in addition to
11529 display backlights on shutdown/boot.
11531 * udev learned a new SECLABEL{} construct to label device
11532 nodes with a specific security label when they appear. For
11533 now, only SECLABEL{selinux} is supported, but the syntax is
11534 prepared for additional security frameworks.
11536 * udev gained a new scheme to configure link-level attributes
11537 from files in /etc/systemd/network/*.link. These files can
11538 match against MAC address, device path, driver name and type,
11539 and will apply attributes like the naming policy, link speed,
11540 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
11541 address assignment policy (randomized, …).
11543 * The configuration of network interface naming rules for
11544 "permanent interface names" has changed: a new NamePolicy=
11545 setting in the [Link] section of .link files determines the
11546 priority of possible naming schemes (onboard, slot, MAC,
11547 path). The default value of this setting is determined by
11548 /usr/lib/net/links/99-default.link. Old
11549 80-net-name-slot.rules udev configuration file has been
11550 removed, so local configuration overriding this file should
11551 be adapted to override 99-default.link instead.
11553 * When the User= switch is used in a unit file, also
11554 initialize $SHELL= based on the user database entry.
11556 * systemd no longer depends on libdbus. All communication is
11557 now done with sd-bus, systemd's low-level bus library
11560 * kdbus support has been added to PID 1 itself. When kdbus is
11561 enabled, this causes PID 1 to set up the system bus and
11562 enable support for a new ".busname" unit type that
11563 encapsulates bus name activation on kdbus. It works a little
11564 bit like ".socket" units, except for bus names. A new
11565 generator has been added that converts classic dbus1 service
11566 activation files automatically into native systemd .busname
11567 and .service units.
11569 * sd-bus: add a light-weight vtable implementation that allows
11570 defining objects on the bus with a simple static const
11571 vtable array of its methods, signals and properties.
11573 * systemd will not generate or install static dbus
11574 introspection data anymore to /usr/share/dbus-1/interfaces,
11575 as the precise format of these files is unclear, and
11576 nothing makes use of it.
11578 * A proxy daemon is now provided to proxy clients connecting
11579 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
11580 compatibility with classic D-Bus.
11582 * A bus driver implementation has been added that supports the
11583 classic D-Bus bus driver calls on kdbus, also for
11584 compatibility purposes.
11586 * A new API "sd-event.h" has been added that implements a
11587 minimal event loop API built around epoll. It provides a
11588 couple of features that direct epoll usage is lacking:
11589 prioritization of events, scales to large numbers of timer
11590 events, per-event timer slack (accuracy), system-wide
11591 coalescing of timer events, exit handlers, watchdog
11592 supervision support using systemd's sd_notify() API, child
11595 * A new API "sd-rntl.h" has been added that provides an API
11596 around the route netlink interface of the kernel, similar in
11597 style to "sd-bus.h".
11599 * A new API "sd-dhcp-client.h" has been added that provides a
11600 small DHCPv4 client-side implementation. This is used by
11601 "systemd-networkd".
11603 * There is a new kernel command line option
11604 "systemd.restore_state=0|1". When set to "0", none of the
11605 systemd tools will restore saved runtime state to hardware
11606 devices. More specifically, the rfkill and backlight states
11609 * The FsckPassNo= compatibility option in mount/service units
11610 has been removed. The fstab generator will now add the
11611 necessary dependencies automatically, and does not require
11612 PID1's support for that anymore.
11614 * journalctl gained a new switch, --list-boots, that lists
11615 recent boots with their times and boot IDs.
11617 * The various tools like systemctl, loginctl, timedatectl,
11618 busctl, systemd-run, … have gained a new switch "-M" to
11619 connect to a specific, local OS container (as direct
11620 connection, without requiring SSH). This works on any
11621 container that is registered with machined, such as those
11622 created by libvirt-lxc or nspawn.
11624 * systemd-run and systemd-analyze also gained support for "-H"
11625 to connect to remote hosts via SSH. This is particularly
11626 useful for systemd-run because it enables queuing of jobs
11627 onto remote systems.
11629 * machinectl gained a new command "login" to open a getty
11630 login in any local container. This works with any container
11631 that is registered with machined (such as those created by
11632 libvirt-lxc or nspawn), and which runs systemd inside.
11634 * machinectl gained a new "reboot" command that may be used to
11635 trigger a reboot on a specific container that is registered
11636 with machined. This works on any container that runs an init
11637 system of some kind.
11639 * systemctl gained a new "list-timers" command to print a nice
11640 listing of installed timer units with the times they elapse
11643 * Alternative reboot() parameters may now be specified on the
11644 "systemctl reboot" command line and are passed to the
11645 reboot() system call.
11647 * systemctl gained a new --job-mode= switch to configure the
11648 mode to queue a job with. This is a more generic version of
11649 --fail, --irreversible, and --ignore-dependencies, which are
11650 still available but not advertised anymore.
11652 * /etc/systemd/system.conf gained new settings to configure
11653 various default timeouts of units, as well as the default
11654 start limit interval and burst. These may still be overridden
11657 * PID1 will now export on the bus profile data of the security
11658 policy upload process (such as the SELinux policy upload to
11661 * journald: when forwarding logs to the console, include
11662 timestamps (following the setting in
11663 /sys/module/printk/parameters/time).
11665 * OnCalendar= in timer units now understands the special
11666 strings "yearly" and "annually". (Both are equivalent)
11668 * The accuracy of timer units is now configurable with the new
11669 AccuracySec= setting. It defaults to 1min.
11671 * A new dependency type JoinsNamespaceOf= has been added that
11672 allows running two services within the same /tmp and network
11673 namespace, if PrivateNetwork= or PrivateTmp= are used.
11675 * A new command "cat" has been added to systemctl. It outputs
11676 the original unit file of a unit, and concatenates the
11677 contents of additional "drop-in" unit file snippets, so that
11678 the full configuration is shown.
11680 * systemctl now supports globbing on the various "list-xyz"
11681 commands, like "list-units" or "list-sockets", as well as on
11682 those commands which take multiple unit names.
11684 * journalctl's --unit= switch gained support for globbing.
11686 * All systemd daemons now make use of the watchdog logic so
11687 that systemd automatically notices when they hang.
11689 * If the $container_ttys environment variable is set,
11690 getty-generator will automatically spawn a getty for each
11691 listed tty. This is useful for container managers to request
11692 login gettys to be spawned on as many ttys as needed.
11694 * %h, %s, %U specifier support is not available anymore when
11695 used in unit files for PID 1. This is because NSS calls are
11696 not safe from PID 1. They stay available for --user
11697 instances of systemd, and as special case for the root user.
11699 * loginctl gained a new "--no-legend" switch to turn off output
11700 of the legend text.
11702 * The "sd-login.h" API gained three new calls:
11703 sd_session_is_remote(), sd_session_get_remote_user(),
11704 sd_session_get_remote_host() to query information about
11707 * The udev hardware database now also carries vendor/product
11708 information of SDIO devices.
11710 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
11711 determine whether watchdog notifications are requested by
11712 the system manager.
11714 * Socket-activated per-connection services now include a
11715 short description of the connection parameters in the
11718 * tmpfiles gained a new "--boot" option. When this is not used,
11719 only lines where the command character is not suffixed with
11720 "!" are executed. When this option is specified, those
11721 options are executed too. This partitions tmpfiles
11722 directives into those that can be safely executed at any
11723 time, and those which should be run only at boot (for
11724 example, a line that creates /run/nologin).
11726 * A new API "sd-resolve.h" has been added which provides a simple
11727 asynchronous wrapper around glibc NSS hostname resolution
11728 calls, such as getaddrinfo(). In contrast to glibc's
11729 getaddrinfo_a(), it does not use signals. In contrast to most
11730 other asynchronous name resolution libraries, this one does
11731 not reimplement DNS, but reuses NSS, so that alternate
11732 hostname resolution systems continue to work, such as mDNS,
11733 LDAP, etc. This API is based on libasyncns, but it has been
11734 cleaned up for inclusion in systemd.
11736 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
11737 "sd-daemon.h" are no longer found in individual libraries
11738 libsystemd-journal.so, libsystemd-login.so,
11739 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
11740 merged them into a single library, libsystemd.so, which
11741 provides all symbols. The reason for this is cyclic
11742 dependencies, as these libraries tend to use each other's
11743 symbols. So far, we have managed to workaround that by linking
11744 a copy of a good part of our code into each of these
11745 libraries again and again, which, however, makes certain
11746 things hard to do, like sharing static variables. Also, it
11747 substantially increases footprint. With this change, there
11748 is only one library for the basic APIs systemd
11749 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
11750 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
11751 library as well, however are subject to the --enable-kdbus
11752 switch (see below). Note that "sd-dhcp-client.h" is not part
11753 of this library (this is because it only consumes, never
11754 provides, services of/to other APIs). To make the transition
11755 easy from the separate libraries to the unified one, we
11756 provide the --enable-compat-libs compile-time switch which
11757 will generate stub libraries that are compatible with the
11758 old ones but redirect all calls to the new one.
11760 * All of the kdbus logic and the new APIs "sd-bus.h",
11761 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
11762 and "sd-utf8.h" are compile-time optional via the
11763 "--enable-kdbus" switch, and they are not compiled in by
11764 default. To make use of kdbus, you have to explicitly enable
11765 the switch. Note however, that neither the kernel nor the
11766 userspace API for all of this is considered stable yet. We
11767 want to maintain the freedom to still change the APIs for
11768 now. By specifying this build-time switch, you acknowledge
11769 that you are aware of the instability of the current
11772 * Also, note that while kdbus is pretty much complete,
11773 it lacks one thing: proper policy support. This means you
11774 can build a fully working system with all features; however,
11775 it will be highly insecure. Policy support will be added in
11776 one of the next releases, at the same time that we will
11777 declare the APIs stable.
11779 * When the kernel command line argument "kdbus" is specified,
11780 systemd will automatically load the kdbus.ko kernel module. At
11781 this stage of development, it is only useful for testing kdbus
11782 and should not be used in production. Note: if "--enable-kdbus"
11783 is specified, and the kdbus.ko kernel module is available, and
11784 "kdbus" is added to the kernel command line, the entire system
11785 runs with kdbus instead of dbus-daemon, with the above mentioned
11786 problem of missing the system policy enforcement. Also a future
11787 version of kdbus.ko or a newer systemd will not be compatible with
11788 each other, and will unlikely be able to boot the machine if only
11789 one of them is updated.
11791 * systemctl gained a new "import-environment" command which
11792 uploads the caller's environment (or parts thereof) into the
11793 service manager so that it is inherited by services started
11794 by the manager. This is useful to upload variables like
11795 $DISPLAY into the user service manager.
11797 * A new PrivateDevices= switch has been added to service units
11798 which allows running a service with a namespaced /dev
11799 directory that does not contain any device nodes for
11800 physical devices. More specifically, it only includes devices
11801 such as /dev/null, /dev/urandom, and /dev/zero which are API
11804 * logind has been extended to support behaviour like VT
11805 switching on seats that do not support a VT. This makes
11806 multi-session available on seats that are not the first seat
11807 (seat0), and on systems where kernel support for VTs has
11808 been disabled at compile-time.
11810 * If a process holds a delay lock for system sleep or shutdown
11811 and fails to release it in time, we will now log its
11812 identity. This makes it easier to identify processes that
11813 cause slow suspends or power-offs.
11815 * When parsing /etc/crypttab, support for a new key-slot=
11816 option as supported by Debian is added. It allows indicating
11817 which LUKS slot to use on disk, speeding up key loading.
11819 * The sd_journal_sendv() API call has been checked and
11820 officially declared to be async-signal-safe so that it may
11821 be invoked from signal handlers for logging purposes.
11823 * Boot-time status output is now enabled automatically after a
11824 short timeout if boot does not progress, in order to give
11825 the user an indication what she or he is waiting for.
11827 * The boot-time output has been improved to show how much time
11828 remains until jobs expire.
11830 * The KillMode= switch in service units gained a new possible
11831 value "mixed". If set, and the unit is shut down, then the
11832 initial SIGTERM signal is sent only to the main daemon
11833 process, while the following SIGKILL signal is sent to
11834 all remaining processes of the service.
11836 * When a scope unit is registered, a new property "Controller"
11837 may be set. If set to a valid bus name, systemd will send a
11838 RequestStop() signal to this name when it would like to shut
11839 down the scope. This may be used to hook manager logic into
11840 the shutdown logic of scope units. Also, scope units may now
11841 be put in a special "abandoned" state, in which case the
11842 manager process which created them takes no further
11843 responsibilities for it.
11845 * When reading unit files, systemd will now verify
11846 the access mode of these files, and warn about certain
11847 suspicious combinations. This has been added to make it
11848 easier to track down packaging bugs where unit files are
11849 marked executable or world-writable.
11851 * systemd-nspawn gained a new "--setenv=" switch to set
11852 container-wide environment variables. The similar option in
11853 systemd-activate was renamed from "--environment=" to
11854 "--setenv=" for consistency.
11856 * systemd-nspawn has been updated to create a new kdbus domain
11857 for each container that is invoked, thus allowing each
11858 container to have its own set of system and user buses,
11859 independent of the host.
11861 * systemd-nspawn gained a new --drop-capability= switch to run
11862 the container with less capabilities than the default. Both
11863 --drop-capability= and --capability= now take the special
11864 string "all" for dropping or keeping all capabilities.
11866 * systemd-nspawn gained new switches for executing containers
11867 with specific SELinux labels set.
11869 * systemd-nspawn gained a new --quiet switch to not generate
11870 any additional output but the container's own console
11873 * systemd-nspawn gained a new --share-system switch to run a
11874 container without PID namespacing enabled.
11876 * systemd-nspawn gained a new --register= switch to control
11877 whether the container is registered with systemd-machined or
11878 not. This is useful for containers that do not run full
11879 OS images, but only specific apps.
11881 * systemd-nspawn gained a new --keep-unit which may be used
11882 when invoked as the only program from a service unit, and
11883 results in registration of the unit service itself in
11884 systemd-machined, instead of a newly opened scope unit.
11886 * systemd-nspawn gained a new --network-interface= switch for
11887 moving arbitrary interfaces to the container. The new
11888 --network-veth switch creates a virtual Ethernet connection
11889 between host and container. The new --network-bridge=
11890 switch then allows assigning the host side of this virtual
11891 Ethernet connection to a bridge device.
11893 * systemd-nspawn gained a new --personality= switch for
11894 setting the kernel personality for the container. This is
11895 useful when running a 32-bit container on a 64-bit host. A
11896 similar option Personality= is now also available for service
11899 * logind will now also track a "Desktop" identifier for each
11900 session which encodes the desktop environment of it. This is
11901 useful for desktop environments that want to identify
11902 multiple running sessions of itself easily.
11904 * A new SELinuxContext= setting for service units has been
11905 added that allows setting a specific SELinux execution
11906 context for a service.
11908 * Most systemd client tools will now honour $SYSTEMD_LESS for
11909 settings of the "less" pager. By default, these tools will
11910 override $LESS to allow certain operations to work, such as
11911 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
11912 influence this logic.
11914 * systemd's "seccomp" hook-up has been changed to make use of
11915 the libseccomp library instead of using its own
11916 implementation. This has benefits for portability among
11919 * For usage together with SystemCallFilter=, a new
11920 SystemCallErrorNumber= setting has been introduced that
11921 allows configuration of a system error number to be returned
11922 on filtered system calls, instead of immediately killing the
11923 process. Also, SystemCallArchitectures= has been added to
11924 limit access to system calls of a particular architecture
11925 (in order to turn off support for unused secondary
11926 architectures). There is also a global
11927 SystemCallArchitectures= setting in system.conf now to turn
11928 off support for non-native system calls system-wide.
11930 * systemd requires a kernel with a working name_to_handle_at(),
11931 please see the kernel config requirements in the README file.
11933 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
11934 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
11935 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
11936 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
11937 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
11938 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
11939 Elia Pinto, Florian Weimer, George McCollister, Goffredo
11940 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
11941 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
11942 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
11943 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
11944 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
11945 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
11946 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
11947 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
11948 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
11949 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
11950 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
11951 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
11952 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
11953 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
11954 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
11955 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
11956 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
11958 — Berlin, 2014-02-20
11962 * logind has gained support for facilitating privileged input
11963 and drm device access for unprivileged clients. This work is
11964 useful to allow Wayland display servers (and similar
11965 programs, such as kmscon) to run under the user's ID and
11966 access input and drm devices which are normally
11967 protected. When this is used (and the kernel is new enough)
11968 logind will "mute" IO on the file descriptors passed to
11969 Wayland as long as it is in the background and "unmute" it
11970 if it returns into the foreground. This allows secure
11971 session switching without allowing background sessions to
11972 eavesdrop on input and display data. This also introduces
11973 session switching support if VT support is turned off in the
11974 kernel, and on seats that are not seat0.
11976 * A new kernel command line option luks.options= is understood
11977 now which allows specifying LUKS options for usage for LUKS
11978 encrypted partitions specified with luks.uuid=.
11980 * tmpfiles.d(5) snippets may now use specifier expansion in
11981 path names. More specifically %m, %b, %H, %v, are now
11982 replaced by the local machine id, boot id, hostname, and
11983 kernel version number.
11985 * A new tmpfiles.d(5) command "m" has been introduced which
11986 may be used to change the owner/group/access mode of a file
11987 or directory if it exists, but do nothing if it does not.
11989 * This release removes high-level support for the
11990 MemorySoftLimit= cgroup setting. The underlying kernel
11991 cgroup attribute memory.soft_limit= is currently badly
11992 designed and likely to be removed from the kernel API in its
11993 current form, hence we should not expose it for now.
11995 * The memory.use_hierarchy cgroup attribute is now enabled for
11996 all cgroups systemd creates in the memory cgroup
11997 hierarchy. This option is likely to be come the built-in
11998 default in the kernel anyway, and the non-hierarchical mode
11999 never made much sense in the intrinsically hierarchical
12002 * A new field _SYSTEMD_SLICE= is logged along with all journal
12003 messages containing the slice a message was generated
12004 from. This is useful to allow easy per-customer filtering of
12005 logs among other things.
12007 * systemd-journald will no longer adjust the group of journal
12008 files it creates to the "systemd-journal" group. Instead we
12009 rely on the journal directory to be owned by the
12010 "systemd-journal" group, and its setgid bit set, so that the
12011 kernel file system layer will automatically enforce that
12012 journal files inherit this group assignment. The reason for
12013 this change is that we cannot allow NSS look-ups from
12014 journald which would be necessary to resolve
12015 "systemd-journal" to a numeric GID, because this might
12016 create deadlocks if NSS involves synchronous queries to
12017 other daemons (such as nscd, or sssd) which in turn are
12018 logging clients of journald and might block on it, which
12019 would then dead lock. A tmpfiles.d(5) snippet included in
12020 systemd will make sure the setgid bit and group are
12021 properly set on the journal directory if it exists on every
12022 boot. However, we recommend adjusting it manually after
12023 upgrades too (or from RPM scriptlets), so that the change is
12024 not delayed until next reboot.
12026 * Backlight and random seed files in /var/lib/ have moved into
12027 the /var/lib/systemd/ directory, in order to centralize all
12028 systemd generated files in one directory.
12030 * Boot time performance measurements (as displayed by
12031 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
12032 performance information if that's available to determine how
12033 much time BIOS and boot loader initialization required. With
12034 a sufficiently new BIOS you hence no longer need to boot
12035 with Gummiboot to get access to such information.
12037 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
12038 Cristian Rodríguez, Dave Reisner, David Herrmann, David
12039 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
12040 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
12041 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
12042 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
12043 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
12045 — Berlin, 2013-10-02
12049 * The Restart= option for services now understands a new
12050 on-watchdog setting, which will restart the service
12051 automatically if the service stops sending out watchdog keep
12052 alive messages (as configured with WatchdogSec=).
12054 * The getty generator (which is responsible for bringing up a
12055 getty on configured serial consoles) will no longer only
12056 start a getty on the primary kernel console but on all
12057 others, too. This makes the order in which console= is
12058 specified on the kernel command line less important.
12060 * libsystemd-logind gained a new sd_session_get_vt() call to
12061 retrieve the VT number of a session.
12063 * If the option "tries=0" is set for an entry of /etc/crypttab
12064 its passphrase is queried indefinitely instead of any
12065 maximum number of tries.
12067 * If a service with a configure PID file terminates its PID
12068 file will now be removed automatically if it still exists
12069 afterwards. This should put an end to stale PID files.
12071 * systemd-run will now also take relative binary path names
12072 for execution and no longer insists on absolute paths.
12074 * InaccessibleDirectories= and ReadOnlyDirectories= now take
12075 paths that are optionally prefixed with "-" to indicate that
12076 it should not be considered a failure if they do not exist.
12078 * journalctl -o (and similar commands) now understands a new
12079 output mode "short-precise", it is similar to "short" but
12080 shows timestamps with usec accuracy.
12082 * The option "discard" (as known from Debian) is now
12083 synonymous to "allow-discards" in /etc/crypttab. In fact,
12084 "discard" is preferred now (since it is easier to remember
12087 * Some licensing clean-ups were made, so that more code is now
12088 LGPL-2.1 licensed than before.
12090 * A minimal tool to save/restore the display backlight
12091 brightness across reboots has been added. It will store the
12092 backlight setting as late as possible at shutdown, and
12093 restore it as early as possible during reboot.
12095 * A logic to automatically discover and enable home and swap
12096 partitions on GPT disks has been added. With this in place
12097 /etc/fstab becomes optional for many setups as systemd can
12098 discover certain partitions located on the root disk
12099 automatically. Home partitions are recognized under their
12100 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
12101 partitions are recognized under their GPT type ID
12102 0657fd6da4ab43c484e50933c84b4f4f.
12104 * systemd will no longer pass any environment from the kernel
12105 or initrd to system services. If you want to set an
12106 environment for all services, do so via the kernel command
12107 line systemd.setenv= assignment.
12109 * The systemd-sysctl tool no longer natively reads the file
12110 /etc/sysctl.conf. If desired, the file should be symlinked
12111 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
12112 legacy support by a symlink rather than built-in code, it
12113 also makes the otherwise hidden order of application of the
12114 different files visible. (Note that this partly reverts to a
12115 pre-198 application order of sysctl knobs!)
12117 * The "systemctl set-log-level" and "systemctl dump" commands
12118 have been moved to systemd-analyze.
12120 * systemd-run learned the new --remain-after-exit switch,
12121 which causes the scope unit not to be cleaned up
12122 automatically after the process terminated.
12124 * tmpfiles learned a new --exclude-prefix= switch to exclude
12125 certain paths from operation.
12127 * journald will now automatically flush all messages to disk
12128 as soon as a message at the log level CRIT, ALERT or EMERG
12131 Contributions from: Andrew Cook, Brandon Philips, Christian
12132 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
12133 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
12134 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
12135 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
12136 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
12137 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
12138 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
12139 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
12140 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
12141 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
12142 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
12143 William Giokas, Zbigniew Jędrzejewski-Szmek
12145 — Berlin, 2013-09-13
12149 * The documentation has been updated to cover the various new
12150 concepts introduced with 205.
12152 * Unit files now understand the new %v specifier which
12153 resolves to the kernel version string as returned by "uname
12156 * systemctl now supports filtering the unit list output by
12157 load state, active state and sub state, using the new
12158 --state= parameter.
12160 * "systemctl status" will now show the results of the
12161 condition checks (like ConditionPathExists= and similar) of
12162 the last start attempts of the unit. They are also logged to
12165 * "journalctl -b" may now be used to look for boot output of a
12166 specific boot. Try "journalctl -b -1" for the previous boot,
12167 but the syntax is substantially more powerful.
12169 * "journalctl --show-cursor" has been added which prints the
12170 cursor string the last shown log line. This may then be used
12171 with the new "journalctl --after-cursor=" switch to continue
12172 browsing logs from that point on.
12174 * "journalctl --force" may now be used to force regeneration
12177 * Creation of "dead" device nodes has been moved from udev
12178 into kmod and tmpfiles. Previously, udev would read the kmod
12179 databases to pre-generate dead device nodes based on meta
12180 information contained in kernel modules, so that these would
12181 be auto-loaded on access rather then at boot. As this
12182 does not really have much to do with the exposing actual
12183 kernel devices to userspace this has always been slightly
12184 alien in the udev codebase. Following the new scheme kmod
12185 will now generate a runtime snippet for tmpfiles from the
12186 module meta information and it now is tmpfiles' job to the
12187 create the nodes. This also allows overriding access and
12188 other parameters for the nodes using the usual tmpfiles
12189 facilities. As side effect this allows us to remove the
12190 CAP_SYS_MKNOD capability bit from udevd entirely.
12192 * logind's device ACLs may now be applied to these "dead"
12193 devices nodes too, thus finally allowing managed access to
12194 devices such as /dev/snd/sequencer without loading the
12195 backing module right-away.
12197 * A new RPM macro has been added that may be used to apply
12198 tmpfiles configuration during package installation.
12200 * systemd-detect-virt and ConditionVirtualization= now can
12201 detect User-Mode-Linux machines (UML).
12203 * journald will now implicitly log the effective capabilities
12204 set of processes in the message metadata.
12206 * systemd-cryptsetup has gained support for TrueCrypt volumes.
12208 * The initrd interface has been simplified (more specifically,
12209 support for passing performance data via environment
12210 variables and fsck results via files in /run has been
12211 removed). These features were non-essential, and are
12212 nowadays available in a much nicer way by having systemd in
12213 the initrd serialize its state and have the hosts systemd
12214 deserialize it again.
12216 * The udev "keymap" data files and tools to apply keyboard
12217 specific mappings of scan to key codes, and force-release
12218 scan code lists have been entirely replaced by a udev
12219 "keyboard" builtin and a hwdb data file.
12221 * systemd will now honour the kernel's "quiet" command line
12222 argument also during late shutdown, resulting in a
12223 completely silent shutdown when used.
12225 * There's now an option to control the SO_REUSEPORT socket
12226 option in .socket units.
12228 * Instance units will now automatically get a per-template
12229 subslice of system.slice unless something else is explicitly
12230 configured. For example, instances of sshd@.service will now
12231 implicitly be placed in system-sshd.slice rather than
12232 system.slice as before.
12234 * Test coverage support may now be enabled at build time.
12236 Contributions from: Dave Reisner, Frederic Crozat, Harald
12237 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
12238 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
12239 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
12240 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
12241 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
12242 Giokas, Zbigniew Jędrzejewski-Szmek
12244 — Berlin, 2013-07-23
12248 * Two new unit types have been introduced:
12250 Scope units are very similar to service units, however, are
12251 created out of pre-existing processes — instead of PID 1
12252 forking off the processes. By using scope units it is
12253 possible for system services and applications to group their
12254 own child processes (worker processes) in a powerful way
12255 which then maybe used to organize them, or kill them
12256 together, or apply resource limits on them.
12258 Slice units may be used to partition system resources in an
12259 hierarchical fashion and then assign other units to them. By
12260 default there are now three slices: system.slice (for all
12261 system services), user.slice (for all user sessions),
12262 machine.slice (for VMs and containers).
12264 Slices and scopes have been introduced primarily in
12265 context of the work to move cgroup handling to a
12266 single-writer scheme, where only PID 1
12267 creates/removes/manages cgroups.
12269 * There's a new concept of "transient" units. In contrast to
12270 normal units these units are created via an API at runtime,
12271 not from configuration from disk. More specifically this
12272 means it is now possible to run arbitrary programs as
12273 independent services, with all execution parameters passed
12274 in via bus APIs rather than read from disk. Transient units
12275 make systemd substantially more dynamic then it ever was,
12276 and useful as a general batch manager.
12278 * logind has been updated to make use of scope and slice units
12279 for managing user sessions. As a user logs in he will get
12280 his own private slice unit, to which all sessions are added
12281 as scope units. We also added support for automatically
12282 adding an instance of user@.service for the user into the
12283 slice. Effectively logind will no longer create cgroup
12284 hierarchies on its own now, it will defer entirely to PID 1
12285 for this by means of scope, service and slice units. Since
12286 user sessions this way become entities managed by PID 1
12287 the output of "systemctl" is now a lot more comprehensive.
12289 * A new mini-daemon "systemd-machined" has been added which
12290 may be used by virtualization managers to register local
12291 VMs/containers. nspawn has been updated accordingly, and
12292 libvirt will be updated shortly. machined will collect a bit
12293 of meta information about the VMs/containers, and assign
12294 them their own scope unit (see above). The collected
12295 meta-data is then made available via the "machinectl" tool,
12296 and exposed in "ps" and similar tools. machined/machinectl
12297 is compile-time optional.
12299 * As discussed earlier, the low-level cgroup configuration
12300 options ControlGroup=, ControlGroupModify=,
12301 ControlGroupPersistent=, ControlGroupAttribute= have been
12302 removed. Please use high-level attribute settings instead as
12303 well as slice units.
12305 * A new bus call SetUnitProperties() has been added to alter
12306 various runtime parameters of a unit. This is primarily
12307 useful to alter cgroup parameters dynamically in a nice way,
12308 but will be extended later on to make more properties
12309 modifiable at runtime. systemctl gained a new set-properties
12310 command that wraps this call.
12312 * A new tool "systemd-run" has been added which can be used to
12313 run arbitrary command lines as transient services or scopes,
12314 while configuring a number of settings via the command
12315 line. This tool is currently very basic, however already
12316 very useful. We plan to extend this tool to even allow
12317 queuing of execution jobs with time triggers from the
12318 command line, similar in fashion to "at".
12320 * nspawn will now inform the user explicitly that kernels with
12321 audit enabled break containers, and suggest the user to turn
12324 * Support for detecting the IMA and AppArmor security
12325 frameworks with ConditionSecurity= has been added.
12327 * journalctl gained a new "-k" switch for showing only kernel
12328 messages, mimicking dmesg output; in addition to "--user"
12329 and "--system" switches for showing only user's own logs
12332 * systemd-delta can now show information about drop-in
12333 snippets extending unit files.
12335 * libsystemd-bus has been substantially updated but is still
12336 not available as public API.
12338 * systemd will now look for the "debug" argument on the kernel
12339 command line and enable debug logging, similar to what
12340 "systemd.log_level=debug" already did before.
12342 * "systemctl set-default", "systemctl get-default" has been
12343 added to configure the default.target symlink, which
12344 controls what to boot into by default.
12346 * "systemctl set-log-level" has been added as a convenient
12347 way to raise and lower systemd logging threshold.
12349 * "systemd-analyze plot" will now show the time the various
12350 generators needed for execution, as well as information
12351 about the unit file loading.
12353 * libsystemd-journal gained a new sd_journal_open_files() call
12354 for opening specific journal files. journactl also gained a
12355 new switch to expose this new functionality. Previously we
12356 only supported opening all files from a directory, or all
12357 files from the system, as opening individual files only is
12358 racy due to journal file rotation.
12360 * systemd gained the new DefaultEnvironment= setting in
12361 /etc/systemd/system.conf to set environment variables for
12364 * If a privileged process logs a journal message with the
12365 OBJECT_PID= field set, then journald will automatically
12366 augment this with additional OBJECT_UID=, OBJECT_GID=,
12367 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
12368 system services want to log events about specific client
12369 processes. journactl/systemctl has been updated to make use
12370 of this information if all log messages regarding a specific
12373 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
12374 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
12375 Reisner, David Coppa, David King, David Strauss, Eelco
12376 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
12377 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
12378 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
12379 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
12380 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
12381 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
12382 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
12383 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
12384 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
12385 Łukasz Stelmach, 장동준
12389 * The Python bindings gained some minimal support for the APIs
12390 exposed by libsystemd-logind.
12392 * ConditionSecurity= gained support for detecting SMACK. Since
12393 this condition already supports SELinux and AppArmor we only
12394 miss IMA for this. Patches welcome!
12396 Contributions from: Karol Lewandowski, Lennart Poettering,
12397 Zbigniew Jędrzejewski-Szmek
12401 * systemd-nspawn will now create /etc/resolv.conf if
12402 necessary, before bind-mounting the host's file onto it.
12404 * systemd-nspawn will now store meta information about a
12405 container on the container's cgroup as extended attribute
12406 fields, including the root directory.
12408 * The cgroup hierarchy has been reworked in many ways. All
12409 objects any of the components systemd creates in the cgroup
12410 tree are now suffixed. More specifically, user sessions are
12411 now placed in cgroups suffixed with ".session", users in
12412 cgroups suffixed with ".user", and nspawn containers in
12413 cgroups suffixed with ".nspawn". Furthermore, all cgroup
12414 names are now escaped in a simple scheme to avoid collision
12415 of userspace object names with kernel filenames. This work
12416 is preparation for making these objects relocatable in the
12417 cgroup tree, in order to allow easy resource partitioning of
12418 these objects without causing naming conflicts.
12420 * systemctl list-dependencies gained the new switches
12421 --plain, --reverse, --after and --before.
12423 * systemd-inhibit now shows the process name of processes that
12424 have taken an inhibitor lock.
12426 * nss-myhostname will now also resolve "localhost"
12427 implicitly. This makes /etc/hosts an optional file and
12428 nicely handles that on IPv6 ::1 maps to both "localhost" and
12429 the local hostname.
12431 * libsystemd-logind.so gained a new call
12432 sd_get_machine_names() to enumerate running containers and
12433 VMs (currently only supported by very new libvirt and
12434 nspawn). sd_login_monitor can now be used to watch
12435 VMs/containers coming and going.
12437 * .include is not allowed recursively anymore, and only in
12438 unit files. Usually it is better to use drop-in snippets in
12439 .d/*.conf anyway, as introduced with systemd 198.
12441 * systemd-analyze gained a new "critical-chain" command that
12442 determines the slowest chain of units run during system
12443 boot-up. It is very useful for tracking down where
12444 optimizing boot time is the most beneficial.
12446 * systemd will no longer allow manipulating service paths in
12447 the name=systemd:/system cgroup tree using ControlGroup= in
12448 units. (But is still fine with it in all other dirs.)
12450 * There's a new systemd-nspawn@.service service file that may
12451 be used to easily run nspawn containers as system
12452 services. With the container's root directory in
12453 /var/lib/container/foobar it is now sufficient to run
12454 "systemctl start systemd-nspawn@foobar.service" to boot it.
12456 * systemd-cgls gained a new parameter "--machine" to list only
12457 the processes within a certain container.
12459 * ConditionSecurity= now can check for "apparmor". We still
12460 are lacking checks for SMACK and IMA for this condition
12461 check though. Patches welcome!
12463 * A new configuration file /etc/systemd/sleep.conf has been
12464 added that may be used to configure which kernel operation
12465 systemd is supposed to execute when "suspend", "hibernate"
12466 or "hybrid-sleep" is requested. This makes the new kernel
12467 "freeze" state accessible to the user.
12469 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
12470 the passed argument if applicable.
12472 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
12473 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
12474 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
12475 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
12476 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
12477 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
12478 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
12483 * The output of 'systemctl list-jobs' got some polishing. The
12484 '--type=' argument may now be passed more than once. A new
12485 command 'systemctl list-sockets' has been added which shows
12486 a list of kernel sockets systemd is listening on with the
12487 socket units they belong to, plus the units these socket
12490 * The experimental libsystemd-bus library got substantial
12491 updates to work in conjunction with the (also experimental)
12492 kdbus kernel project. It works well enough to exchange
12493 messages with some sophistication. Note that kdbus is not
12494 ready yet, and the library is mostly an elaborate test case
12495 for now, and not installable.
12497 * systemd gained a new unit 'systemd-static-nodes.service'
12498 that generates static device nodes earlier during boot, and
12499 can run in conjunction with udev.
12501 * libsystemd-login gained a new call sd_pid_get_user_unit()
12502 to retrieve the user systemd unit a process is running
12503 in. This is useful for systems where systemd is used as
12506 * systemd-nspawn now places all containers in the new /machine
12507 top-level cgroup directory in the name=systemd
12508 hierarchy. libvirt will soon do the same, so that we get a
12509 uniform separation of /system, /user and /machine for system
12510 services, user processes and containers/virtual
12511 machines. This new cgroup hierarchy is also useful to stick
12512 stable names to specific container instances, which can be
12513 recognized later this way (this name may be controlled
12514 via systemd-nspawn's new -M switch). libsystemd-login also
12515 gained a new call sd_pid_get_machine_name() to retrieve the
12516 name of the container/VM a specific process belongs to.
12518 * bootchart can now store its data in the journal.
12520 * libsystemd-journal gained a new call
12521 sd_journal_add_conjunction() for AND expressions to the
12522 matching logic. This can be used to express more complex
12523 logical expressions.
12525 * journactl can now take multiple --unit= and --user-unit=
12528 * The cryptsetup logic now understands the "luks.key=" kernel
12529 command line switch for specifying a file to read the
12530 decryption key from. Also, if a configured key file is not
12531 found the tool will now automatically fall back to prompting
12534 * Python systemd.journal module was updated to wrap recently
12535 added functions from libsystemd-journal. The interface was
12536 changed to bring the low level interface in s.j._Reader
12537 closer to the C API, and the high level interface in
12538 s.j.Reader was updated to wrap and convert all data about
12541 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
12542 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
12543 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
12544 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
12545 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
12546 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
12550 * journalctl --update-catalog now understands a new --root=
12551 option to operate on catalogs found in a different root
12554 * During shutdown after systemd has terminated all running
12555 services a final killing loop kills all remaining left-over
12556 processes. We will now print the name of these processes
12557 when we send SIGKILL to them, since this usually indicates a
12560 * If /etc/crypttab refers to password files stored on
12561 configured mount points automatic dependencies will now be
12562 generated to ensure the specific mount is established first
12563 before the key file is attempted to be read.
12565 * 'systemctl status' will now show information about the
12566 network sockets a socket unit is listening on.
12568 * 'systemctl status' will also shown information about any
12569 drop-in configuration file for units. (Drop-In configuration
12570 files in this context are files such as
12571 /etc/systemd/system/foobar.service.d/*.conf)
12573 * systemd-cgtop now optionally shows summed up CPU times of
12574 cgroups. Press '%' while running cgtop to switch between
12575 percentage and absolute mode. This is useful to determine
12576 which cgroups use up the most CPU time over the entire
12577 runtime of the system. systemd-cgtop has also been updated
12578 to be 'pipeable' for processing with further shell tools.
12580 * 'hostnamectl set-hostname' will now allow setting of FQDN
12583 * The formatting and parsing of time span values has been
12584 changed. The parser now understands fractional expressions
12585 such as "5.5h". The formatter will now output fractional
12586 expressions for all time spans under 1min, i.e. "5.123456s"
12587 rather than "5s 123ms 456us". For time spans under 1s
12588 millisecond values are shown, for those under 1ms
12589 microsecond values are shown. This should greatly improve
12590 all time-related output of systemd.
12592 * libsystemd-login and libsystemd-journal gained new
12593 functions for querying the poll() events mask and poll()
12594 timeout value for integration into arbitrary event
12597 * localectl gained the ability to list available X11 keymaps
12598 (models, layouts, variants, options).
12600 * 'systemd-analyze dot' gained the ability to filter for
12601 specific units via shell-style globs, to create smaller,
12602 more useful graphs. I.e. it is now possible to create simple
12603 graphs of all the dependencies between only target units, or
12604 of all units that Avahi has dependencies with.
12606 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
12607 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
12608 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
12609 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
12610 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
12611 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
12612 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
12616 * The boot-time readahead implementation for rotating media
12617 will now read the read-ahead data in multiple passes which
12618 consist of all read requests made in equidistant time
12619 intervals. This means instead of strictly reading read-ahead
12620 data in its physical order on disk we now try to find a
12621 middle ground between physical and access time order.
12623 * /etc/os-release files gained a new BUILD_ID= field for usage
12624 on operating systems that provide continuous builds of OS
12627 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
12628 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
12629 William Douglas, Zbigniew Jędrzejewski-Szmek
12633 * systemd-python gained an API exposing libsystemd-daemon.
12635 * The SMACK setup logic gained support for uploading CIPSO
12638 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
12639 ReadOnlyDirectories= and InaccessibleDirectories= has
12640 changed. The private /tmp and /var/tmp directories are now
12641 shared by all processes of a service (which means
12642 ExecStartPre= may now leave data in /tmp that ExecStart= of
12643 the same service can still access). When a service is
12644 stopped its temporary directories are immediately deleted
12645 (normal clean-up with tmpfiles is still done in addition to
12648 * By default, systemd will now set a couple of sysctl
12649 variables in the kernel: the safe sysrq options are turned
12650 on, IP route verification is turned on, and source routing
12651 disabled. The recently added hardlink and softlink
12652 protection of the kernel is turned on. These settings should
12653 be reasonably safe, and good defaults for all new systems.
12655 * The predictable network naming logic may now be turned off
12656 with a new kernel command line switch: net.ifnames=0.
12658 * A new libsystemd-bus module has been added that implements a
12659 pretty complete D-Bus client library. For details see:
12661 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
12663 * journald will now explicitly flush the journal files to disk
12664 at the latest 5min after each write. The file will then also
12665 be marked offline until the next write. This should increase
12666 reliability in case of a crash. The synchronization delay
12667 can be configured via SyncIntervalSec= in journald.conf.
12669 * There's a new remote-fs-setup.target unit that can be used
12670 to pull in specific services when at least one remote file
12671 system is to be mounted.
12673 * There are new targets timers.target and paths.target as
12674 canonical targets to pull user timer and path units in
12675 from. This complements sockets.target with a similar
12676 purpose for socket units.
12678 * libudev gained a new call udev_device_set_attribute_value()
12679 to set sysfs attributes of a device.
12681 * The udev daemon now sets the default number of worker
12682 processes executed in parallel based on the number of available
12683 CPUs instead of the amount of available RAM. This is supposed
12684 to provide a more reliable default and limit a too aggressive
12685 parallelism for setups with 1000s of devices connected.
12687 Contributions from: Auke Kok, Colin Walters, Cristian
12688 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
12689 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
12690 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
12691 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
12692 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
12693 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
12694 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
12695 Zbigniew Jędrzejewski-Szmek
12699 * Configuration of unit files may now be extended via drop-in
12700 files without having to edit/override the unit files
12701 themselves. More specifically, if the administrator wants to
12702 change one value for a service file foobar.service he can
12703 now do so by dropping in a configuration snippet into
12704 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
12705 will load all these snippets and apply them on top of the
12706 main unit configuration file, possibly extending or
12707 overriding its settings. Using these drop-in snippets is
12708 generally nicer than the two earlier options for changing
12709 unit files locally: copying the files from
12710 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
12711 them there; or creating a new file in /etc/systemd/system/
12712 that incorporates the original one via ".include". Drop-in
12713 snippets into these .d/ directories can be placed in any
12714 directory systemd looks for units in, and the usual
12715 overriding semantics between /usr/lib, /etc and /run apply
12718 * Most unit file settings which take lists of items can now be
12719 reset by assigning the empty string to them. For example,
12720 normally, settings such as Environment=FOO=BAR append a new
12721 environment variable assignment to the environment block,
12722 each time they are used. By assigning Environment= the empty
12723 string the environment block can be reset to empty. This is
12724 particularly useful with the .d/*.conf drop-in snippets
12725 mentioned above, since this adds the ability to reset list
12726 settings from vendor unit files via these drop-ins.
12728 * systemctl gained a new "list-dependencies" command for
12729 listing the dependencies of a unit recursively.
12731 * Inhibitors are now honored and listed by "systemctl
12732 suspend", "systemctl poweroff" (and similar) too, not only
12733 GNOME. These commands will also list active sessions by
12736 * Resource limits (as exposed by the various control group
12737 controllers) can now be controlled dynamically at runtime
12738 for all units. More specifically, you can now use a command
12739 like "systemctl set-cgroup-attr foobar.service cpu.shares
12740 2000" to alter the CPU shares a specific service gets. These
12741 settings are stored persistently on disk, and thus allow the
12742 administrator to easily adjust the resource usage of
12743 services with a few simple commands. This dynamic resource
12744 management logic is also available to other programs via the
12745 bus. Almost any kernel cgroup attribute and controller is
12748 * systemd-vconsole-setup will now copy all font settings to
12749 all allocated VTs, where it previously applied them only to
12752 * libsystemd-login gained the new sd_session_get_tty() API
12755 * This release drops support for a few legacy or
12756 distribution-specific LSB facility names when parsing init
12757 scripts: $x-display-manager, $mail-transfer-agent,
12758 $mail-transport-agent, $mail-transfer-agent, $smtp,
12759 $null. Also, the mail-transfer-agent.target unit backing
12760 this has been removed. Distributions which want to retain
12761 compatibility with this should carry the burden for
12762 supporting this themselves and patch support for these back
12763 in, if they really need to. Also, the facilities $syslog and
12764 $local_fs are now ignored, since systemd does not support
12765 early-boot LSB init scripts anymore, and these facilities
12766 are implied anyway for normal services. syslog.target has
12769 * There are new bus calls on PID1's Manager object for
12770 cancelling jobs, and removing snapshot units. Previously,
12771 both calls were only available on the Job and Snapshot
12772 objects themselves.
12774 * systemd-journal-gatewayd gained SSL support.
12776 * The various "environment" files, such as /etc/locale.conf
12777 now support continuation lines with a backslash ("\") as
12778 last character in the line, similarly in style (but different)
12779 to how this is supported in shells.
12781 * For normal user processes the _SYSTEMD_USER_UNIT= field is
12782 now implicitly appended to every log entry logged. systemctl
12783 has been updated to filter by this field when operating on a
12784 user systemd instance.
12786 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
12787 CAP_AUDIT_CONTROL capabilities to the capabilities set for
12788 the container. This makes it easier to boot unmodified
12789 Fedora systems in a container, which however still requires
12790 audit=0 to be passed on the kernel command line. Auditing in
12791 kernel and userspace is unfortunately still too broken in
12792 context of containers, hence we recommend compiling it out
12793 of the kernel or using audit=0. Hopefully this will be fixed
12794 one day for good in the kernel.
12796 * nspawn gained the new --bind= and --bind-ro= parameters to
12797 bind mount specific directories from the host into the
12800 * nspawn will now mount its own devpts file system instance
12801 into the container, in order not to leak pty devices from
12802 the host into the container.
12804 * systemd will now read the firmware boot time performance
12805 information from the EFI variables, if the used boot loader
12806 supports this, and takes it into account for boot performance
12807 analysis via "systemd-analyze". This is currently supported
12808 only in conjunction with Gummiboot, but could be supported
12809 by other boot loaders too. For details see:
12811 https://systemd.io/BOOT_LOADER_INTERFACE
12813 * A new generator has been added that automatically mounts the
12814 EFI System Partition (ESP) to /boot, if that directory
12815 exists, is empty, and no other file system has been
12816 configured to be mounted there.
12818 * logind will now send out PrepareForSleep(false) out
12819 unconditionally, after coming back from suspend. This may be
12820 used by applications as asynchronous notification for
12821 system resume events.
12823 * "systemctl unlock-sessions" has been added, that allows
12824 unlocking the screens of all user sessions at once, similar
12825 to how "systemctl lock-sessions" already locked all users
12826 sessions. This is backed by a new D-Bus call UnlockSessions().
12828 * "loginctl seat-status" will now show the master device of a
12829 seat. (i.e. the device of a seat that needs to be around for
12830 the seat to be considered available, usually the graphics
12833 * tmpfiles gained a new "X" line type, that allows
12834 configuration of files and directories (with wildcards) that
12835 shall be excluded from automatic cleanup ("aging").
12837 * udev default rules set the device node permissions now only
12838 at "add" events, and do not change them any longer with a
12839 later "change" event.
12841 * The log messages for lid events and power/sleep keypresses
12842 now carry a message ID.
12844 * We now have a substantially larger unit test suite, but this
12845 continues to be work in progress.
12847 * udevadm hwdb gained a new --root= parameter to change the
12848 root directory to operate relative to.
12850 * logind will now issue a background sync() request to the kernel
12851 early at shutdown, so that dirty buffers are flushed to disk early
12852 instead of at the last moment, in order to optimize shutdown
12855 * A new bootctl tool has been added that is an interface for
12856 certain boot loader operations. This is currently a preview
12857 and is likely to be extended into a small mechanism daemon
12858 like timedated, localed, hostnamed, and can be used by
12859 graphical UIs to enumerate available boot options, and
12860 request boot into firmware operations.
12862 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
12863 the rest of the package. It also has been updated to work
12864 correctly in initrds.
12866 * polkit previously has been runtime optional, and is now also
12867 compile time optional via a configure switch.
12869 * systemd-analyze has been reimplemented in C. Also "systemctl
12870 dot" has moved into systemd-analyze.
12872 * "systemctl status" with no further parameters will now print
12873 the status of all active or failed units.
12875 * Operations such as "systemctl start" can now be executed
12876 with a new mode "--irreversible" which may be used to queue
12877 operations that cannot accidentally be reversed by a later
12878 job queuing. This is by default used to make shutdown
12879 requests more robust.
12881 * The Python API of systemd now gained a new module for
12882 reading journal files.
12884 * A new tool kernel-install has been added that can install
12885 kernel images according to the Boot Loader Specification:
12887 https://systemd.io/BOOT_LOADER_SPECIFICATION
12889 * Boot time console output has been improved to provide
12890 animated boot time output for hanging jobs.
12892 * A new tool systemd-activate has been added which can be used
12893 to test socket activation with, directly from the command
12894 line. This should make it much easier to test and debug
12895 socket activation in daemons.
12897 * journalctl gained a new "--reverse" (or -r) option to show
12898 journal output in reverse order (i.e. newest line first).
12900 * journalctl gained a new "--pager-end" (or -e) option to jump
12901 to immediately jump to the end of the journal in the
12902 pager. This is only supported in conjunction with "less".
12904 * journalctl gained a new "--user-unit=" option, that works
12905 similarly to "--unit=" but filters for user units rather than
12908 * A number of unit files to ease adoption of systemd in
12909 initrds has been added. This moves some minimal logic from
12910 the various initrd implementations into systemd proper.
12912 * The journal files are now owned by a new group
12913 "systemd-journal", which exists specifically to allow access
12914 to the journal, and nothing else. Previously, we used the
12915 "adm" group for that, which however possibly covers more
12916 than just journal/log file access. This new group is now
12917 already used by systemd-journal-gatewayd to ensure this
12918 daemon gets access to the journal files and as little else
12919 as possible. Note that "make install" will also set FS ACLs
12920 up for /var/log/journal to give "adm" and "wheel" read
12921 access to it, in addition to "systemd-journal" which owns
12922 the journal files. We recommend that packaging scripts also
12923 add read access to "adm" + "wheel" to /var/log/journal, and
12924 all existing/future journal files. To normal users and
12925 administrators little changes, however packagers need to
12926 ensure to create the "systemd-journal" system group at
12927 package installation time.
12929 * The systemd-journal-gatewayd now runs as unprivileged user
12930 systemd-journal-gateway:systemd-journal-gateway. Packaging
12931 scripts need to create these system user/group at
12934 * timedated now exposes a new boolean property CanNTP that
12935 indicates whether a local NTP service is available or not.
12937 * systemd-detect-virt will now also detect xen PVs
12939 * The pstore file system is now mounted by default, if it is
12942 * In addition to the SELinux and IMA policies we will now also
12943 load SMACK policies at early boot.
12945 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
12946 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
12947 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
12948 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
12949 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
12950 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
12951 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
12952 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
12953 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
12954 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
12955 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
12956 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
12957 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
12958 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
12962 * Timer units now support calendar time events in addition to
12963 monotonic time events. That means you can now trigger a unit
12964 based on a calendar time specification such as "Thu,Fri
12965 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
12966 or fifth day of any month of the year 2013, given that it is
12967 a Thursday or a Friday. This brings timer event support
12968 considerably closer to cron's capabilities. For details on
12969 the supported calendar time specification language see
12972 * udev now supports a number of different naming policies for
12973 network interfaces for predictable names, and a combination
12974 of these policies is now the default. Please see this wiki
12975 document for details:
12977 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
12979 * Auke Kok's bootchart implementation has been added to the
12980 systemd tree. It is an optional component that can graph the
12981 boot in quite some detail. It is one of the best bootchart
12982 implementations around and minimal in its code and
12985 * nss-myhostname has been integrated into the systemd source
12986 tree. nss-myhostname guarantees that the local hostname
12987 always stays resolvable via NSS. It has been a weak
12988 requirement of systemd-hostnamed since a long time, and
12989 since its code is actually trivial we decided to just
12990 include it in systemd's source tree. It can be turned off
12991 with a configure switch.
12993 * The read-ahead logic is now capable of properly detecting
12994 whether a btrfs file system is on SSD or rotating media, in
12995 order to optimize the read-ahead scheme. Previously, it was
12996 only capable of detecting this on traditional file systems
12999 * In udev, additional device properties are now read from the
13000 IAB in addition to the OUI database. Also, Bluetooth company
13001 identities are attached to the devices as well.
13003 * In service files %U may be used as specifier that is
13004 replaced by the configured user name of the service.
13006 * nspawn may now be invoked without a controlling TTY. This
13007 makes it suitable for invocation as its own service. This
13008 may be used to set up a simple containerized server system
13009 using only core OS tools.
13011 * systemd and nspawn can now accept socket file descriptors
13012 when they are started for socket activation. This enables
13013 implementation of socket activated nspawn
13014 containers. i.e. think about autospawning an entire OS image
13015 when the first SSH or HTTP connection is received. We expect
13016 that similar functionality will also be added to libvirt-lxc
13019 * journalctl will now suppress ANSI color codes when
13020 presenting log data.
13022 * systemctl will no longer show control group information for
13023 a unit if the control group is empty anyway.
13025 * logind can now automatically suspend/hibernate/shutdown the
13028 * /etc/machine-info and hostnamed now also expose the chassis
13029 type of the system. This can be used to determine whether
13030 the local system is a laptop, desktop, handset or
13031 tablet. This information may either be configured by the
13032 user/vendor or is automatically determined from ACPI and DMI
13033 information if possible.
13035 * A number of polkit actions are now bound together with "imply"
13036 rules. This should simplify creating UIs because many actions
13037 will now authenticate similar ones as well.
13039 * Unit files learnt a new condition ConditionACPower= which
13040 may be used to conditionalize a unit depending on whether an
13041 AC power source is connected or not, of whether the system
13042 is running on battery power.
13044 * systemctl gained a new "is-failed" verb that may be used in
13045 shell scripts and suchlike to check whether a specific unit
13046 is in the "failed" state.
13048 * The EnvironmentFile= setting in unit files now supports file
13049 globbing, and can hence be used to easily read a number of
13050 environment files at once.
13052 * systemd will no longer detect and recognize specific
13053 distributions. All distribution-specific #ifdeffery has been
13054 removed, systemd is now fully generic and
13055 distribution-agnostic. Effectively, not too much is lost as
13056 a lot of the code is still accessible via explicit configure
13057 switches. However, support for some distribution specific
13058 legacy configuration file formats has been dropped. We
13059 recommend distributions to simply adopt the configuration
13060 files everybody else uses now and convert the old
13061 configuration from packaging scripts. Most distributions
13062 already did that. If that's not possible or desirable,
13063 distributions are welcome to forward port the specific
13064 pieces of code locally from the git history.
13066 * When logging a message about a unit systemd will now always
13067 log the unit name in the message meta data.
13069 * localectl will now also discover system locale data that is
13070 not stored in locale archives, but directly unpacked.
13072 * logind will no longer unconditionally use framebuffer
13073 devices as seat masters, i.e. as devices that are required
13074 to be existing before a seat is considered preset. Instead,
13075 it will now look for all devices that are tagged as
13076 "seat-master" in udev. By default, framebuffer devices will
13077 be marked as such, but depending on local systems, other
13078 devices might be marked as well. This may be used to
13079 integrate graphics cards using closed source drivers (such
13080 as NVidia ones) more nicely into logind. Note however, that
13081 we recommend using the open source NVidia drivers instead,
13082 and no udev rules for the closed-source drivers will be
13083 shipped from us upstream.
13085 Contributions from: Adam Williamson, Alessandro Crismani, Auke
13086 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
13087 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
13088 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
13089 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
13090 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
13091 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
13092 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
13093 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
13094 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
13095 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
13096 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
13101 * udev gained support for loading additional device properties
13102 from an indexed database that is keyed by vendor/product IDs
13103 and similar device identifiers. For the beginning this
13104 "hwdb" is populated with data from the well-known PCI and
13105 USB database, but also includes PNP, ACPI and OID data. In
13106 the longer run this indexed database shall grow into
13107 becoming the one central database for non-essential
13108 userspace device metadata. Previously, data from the PCI/USB
13109 database was only attached to select devices, since the
13110 lookup was a relatively expensive operation due to O(n) time
13111 complexity (with n being the number of entries in the
13112 database). Since this is now O(1), we decided to add in this
13113 data for all devices where this is available, by
13114 default. Note that the indexed database needs to be rebuilt
13115 when new data files are installed. To achieve this you need
13116 to update your packaging scripts to invoke "udevadm hwdb
13117 --update" after installation of hwdb data files. For
13118 RPM-based distributions we introduced the new
13119 %udev_hwdb_update macro for this purpose.
13121 * The Journal gained support for the "Message Catalog", an
13122 indexed database to link up additional information with
13123 journal entries. For further details please check:
13125 https://www.freedesktop.org/wiki/Software/systemd/catalog
13127 The indexed message catalog database also needs to be
13128 rebuilt after installation of message catalog files. Use
13129 "journalctl --update-catalog" for this. For RPM-based
13130 distributions we introduced the %journal_catalog_update
13131 macro for this purpose.
13133 * The Python Journal bindings gained support for the standard
13134 Python logging framework.
13136 * The Journal API gained new functions for checking whether
13137 the underlying file system of a journal file is capable of
13138 properly reporting file change notifications, or whether
13139 applications that want to reflect journal changes "live"
13140 need to recheck journal files continuously in appropriate
13143 * It is now possible to set the "age" field for tmpfiles
13144 entries to 0, indicating that files matching this entry
13145 shall always be removed when the directories are cleaned up.
13147 * coredumpctl gained a new "gdb" verb which invokes gdb
13148 right-away on the selected coredump.
13150 * There's now support for "hybrid sleep" on kernels that
13151 support this, in addition to "suspend" and "hibernate". Use
13152 "systemctl hybrid-sleep" to make use of this.
13154 * logind's HandleSuspendKey= setting (and related settings)
13155 now gained support for a new "lock" setting to simply
13156 request the screen lock on all local sessions, instead of
13157 actually executing a suspend or hibernation.
13159 * systemd will now mount the EFI variables file system by
13162 * Socket units now gained support for configuration of the
13163 SMACK security label.
13165 * timedatectl will now output the time of the last and next
13166 daylight saving change.
13168 * We dropped support for various legacy and distro-specific
13169 concepts, such as insserv, early-boot SysV services
13170 (i.e. those for non-standard runlevels such as 'b' or 'S')
13171 or ArchLinux /etc/rc.conf support. We recommend the
13172 distributions who still need support this to either continue
13173 to maintain the necessary patches downstream, or find a
13174 different solution. (Talk to us if you have questions!)
13176 * Various systemd components will now bypass polkit checks for
13177 root and otherwise handle properly if polkit is not found to
13178 be around. This should fix most issues for polkit-less
13179 systems. Quite frankly this should have been this way since
13180 day one. It is absolutely our intention to make systemd work
13181 fine on polkit-less systems, and we consider it a bug if
13182 something does not work as it should if polkit is not around.
13184 * For embedded systems it is now possible to build udev and
13185 systemd without blkid and/or kmod support.
13187 * "systemctl switch-root" is now capable of switching root
13188 more than once. I.e. in addition to transitions from the
13189 initrd to the host OS it is now possible to transition to
13190 further OS images from the host. This is useful to implement
13191 offline updating tools.
13193 * Various other additions have been made to the RPM macros
13194 shipped with systemd. Use %udev_rules_update() after
13195 installing new udev rules files. %_udevhwdbdir,
13196 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
13197 %_sysctldir are now available which resolve to the right
13198 directories for packages to place various data files in.
13200 * journalctl gained the new --full switch (in addition to
13201 --all, to disable ellipsation for long messages.
13203 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
13204 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
13205 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
13206 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
13207 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
13208 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
13209 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
13210 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
13211 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
13215 * journalctl gained new --since= and --until= switches to
13216 filter by time. It also now supports nice filtering for
13217 units via --unit=/-u.
13219 * Type=oneshot services may use ExecReload= and do the
13222 * The journal daemon now supports time-based rotation and
13223 vacuuming, in addition to the usual disk-space based
13226 * The journal will now index the available field values for
13227 each field name. This enables clients to show pretty drop
13228 downs of available match values when filtering. The bash
13229 completion of journalctl has been updated
13230 accordingly. journalctl gained a new switch -F to list all
13231 values a certain field takes in the journal database.
13233 * More service events are now written as structured messages
13234 to the journal, and made recognizable via message IDs.
13236 * The timedated, localed and hostnamed mini-services which
13237 previously only provided support for changing time, locale
13238 and hostname settings from graphical DEs such as GNOME now
13239 also have a minimal (but very useful) text-based client
13240 utility each. This is probably the nicest way to changing
13241 these settings from the command line now, especially since
13242 it lists available options and is fully integrated with bash
13245 * There's now a new tool "systemd-coredumpctl" to list and
13246 extract coredumps from the journal.
13248 * We now install a README each in /var/log/ and
13249 /etc/rc.d/init.d explaining where the system logs and init
13250 scripts went. This hopefully should help folks who go to
13251 that dirs and look into the otherwise now empty void and
13252 scratch their heads.
13254 * When user-services are invoked (by systemd --user) the
13255 $MANAGERPID env var is set to the PID of systemd.
13257 * SIGRTMIN+24 when sent to a --user instance will now result
13258 in immediate termination of systemd.
13260 * gatewayd received numerous feature additions such as a
13261 "follow" mode, for live syncing and filtering.
13263 * browse.html now allows filtering and showing detailed
13264 information on specific entries. Keyboard navigation and
13265 mouse screen support has been added.
13267 * gatewayd/journalctl now supports HTML5/JSON
13268 Server-Sent-Events as output.
13270 * The SysV init script compatibility logic will now
13271 heuristically determine whether a script supports the
13272 "reload" verb, and only then make this available as
13273 "systemctl reload".
13275 * "systemctl status --follow" has been removed, use "journalctl
13278 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
13279 have been removed since they are hardly useful to be
13282 * And I'd like to take the opportunity to specifically mention
13283 Zbigniew for his great contributions. Zbigniew, you rock!
13285 Contributions from: Andrew Eikum, Christian Hesse, Colin
13286 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
13287 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
13288 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
13289 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
13290 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
13291 Jędrzejewski-Szmek, Сковорода Никита Андреевич
13295 * If /etc/vconsole.conf is non-existent or empty we will no
13296 longer load any console font or key map at boot by
13297 default. Instead the kernel defaults will be left
13298 intact. This is definitely the right thing to do, as no
13299 configuration should mean no configuration, and hard-coding
13300 font names that are different on all archs is probably a bad
13301 idea. Also, the kernel default key map and font should be
13302 good enough for most cases anyway, and mostly identical to
13303 the userspace fonts/key maps we previously overloaded them
13304 with. If distributions want to continue to default to a
13305 non-kernel font or key map they should ship a default
13306 /etc/vconsole.conf with the appropriate contents.
13308 Contributions from: Colin Walters, Daniel J Walsh, Dave
13309 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
13310 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
13314 * journalctl gained a new --cursor= switch to show entries
13315 starting from the specified location in the journal.
13317 * We now enforce a size limit on journal entry fields exported
13318 with "-o json" in journalctl. Fields larger than 4K will be
13319 assigned null. This can be turned off with --all.
13321 * An (optional) journal gateway daemon is now available as
13322 "systemd-journal-gatewayd.service". This service provides
13323 access to the journal via HTTP and JSON. This functionality
13324 will be used to implement live log synchronization in both
13325 pull and push modes, but has various other users too, such
13326 as easy log access for debugging of embedded devices. Right
13327 now it is already useful to retrieve the journal via HTTP:
13329 # systemctl start systemd-journal-gatewayd.service
13330 # wget http://localhost:19531/entries
13332 This will download the journal contents in a
13333 /var/log/messages compatible format. The same as JSON:
13335 # curl -H"Accept: application/json" http://localhost:19531/entries
13337 This service is also accessible via a web browser where a
13338 single static HTML5 app is served that uses the JSON logic
13339 to enable the user to do some basic browsing of the
13340 journal. This will be extended later on. Here's an example
13341 screenshot of this app in its current state:
13343 http://0pointer.de/public/journal-gatewayd
13345 Contributions from: Kay Sievers, Lennart Poettering, Robert
13346 Milasan, Tom Gundersen
13350 * The bash completion logic is now available for journalctl
13353 * We do not mount the "cpuset" controller anymore together with
13354 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
13355 started if no parameters are assigned to it. "cpuset" hence
13356 broke code that assumed it could create "cpu" groups and
13359 * journalctl -f will now subscribe to terminal size changes,
13360 and line break accordingly.
13362 Contributions from: Dave Reisner, Kay Sievers, Lennart
13363 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
13367 * nspawn will now create a symlink /etc/localtime in the
13368 container environment, copying the host's timezone
13369 setting. Previously this has been done via a bind mount, but
13370 since symlinks cannot be bind mounted this has now been
13371 changed to create/update the appropriate symlink.
13373 * journalctl -n's line number argument is now optional, and
13374 will default to 10 if omitted.
13376 * journald will now log the maximum size the journal files may
13377 take up on disk. This is particularly useful if the default
13378 built-in logic of determining this parameter from the file
13379 system size is used. Use "systemctl status
13380 systemd-journald.service" to see this information.
13382 * The multi-seat X wrapper tool has been stripped down. As X
13383 is now capable of enumerating graphics devices via udev in a
13384 seat-aware way the wrapper is not strictly necessary
13385 anymore. A stripped down temporary stop-gap is still shipped
13386 until the upstream display managers have been updated to
13387 fully support the new X logic. Expect this wrapper to be
13388 removed entirely in one of the next releases.
13390 * HandleSleepKey= in logind.conf has been split up into
13391 HandleSuspendKey= and HandleHibernateKey=. The old setting
13392 is not available anymore. X11 and the kernel are
13393 distinguishing between these keys and we should too. This
13394 also means the inhibition lock for these keys has been split
13397 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
13398 Poettering, Lukas Nykryn, Václav Pavlín
13402 * Whenever a unit changes state we will now log this to the
13403 journal and show along the unit's own log output in
13404 "systemctl status".
13406 * ConditionPathIsMountPoint= can now properly detect bind
13407 mount points too. (Previously, a bind mount of one file
13408 system to another place in the same file system could not be
13409 detected as mount, since they shared struct stat's st_dev
13412 * We will now mount the cgroup controllers cpu, cpuacct,
13413 cpuset and the controllers net_cls, net_prio together by
13416 * nspawn containers will now have a virtualized boot
13417 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
13418 over with a randomized ID at container initialization). This
13419 has the effect of making "journalctl -b" do the right thing
13422 * The JSON output journal serialization has been updated not
13423 to generate "endless" list objects anymore, but rather one
13424 JSON object per line. This is more in line how most JSON
13425 parsers expect JSON objects. The new output mode
13426 "json-pretty" has been added to provide similar output, but
13427 neatly aligned for readability by humans.
13429 * We dropped all explicit sync() invocations in the shutdown
13430 code. The kernel does this implicitly anyway in the kernel
13431 reboot() syscall. halt(8)'s -n option is now a compatibility
13434 * We now support virtualized reboot() in containers, as
13435 supported by newer kernels. We will fall back to exit() if
13436 CAP_SYS_REBOOT is not available to the container. Also,
13437 nspawn makes use of this now and will actually reboot the
13438 container if the containerized OS asks for that.
13440 * journalctl will only show local log output by default
13441 now. Use --merge (-m) to show remote log output, too.
13443 * libsystemd-journal gained the new sd_journal_get_usage()
13444 call to determine the current disk usage of all journal
13445 files. This is exposed in the new "journalctl --disk-usage"
13448 * journald gained a new configuration setting SplitMode= in
13449 journald.conf which may be used to control how user journals
13450 are split off. See journald.conf(5) for details.
13452 * A new condition type ConditionFileNotEmpty= has been added.
13454 * tmpfiles' "w" lines now support file globbing, to write
13455 multiple files at once.
13457 * We added Python bindings for the journal submission
13458 APIs. More Python APIs for a number of selected APIs will
13459 likely follow. Note that we intend to add native bindings
13460 only for the Python language, as we consider it common
13461 enough to deserve bindings shipped within systemd. There are
13462 various projects outside of systemd that provide bindings
13463 for languages such as PHP or Lua.
13465 * Many conditions will now resolve specifiers such as %i. In
13466 addition, PathChanged= and related directives of .path units
13467 now support specifiers as well.
13469 * There's now a new RPM macro definition for the system preset
13472 * journald will now warn if it ca not forward a message to the
13473 syslog daemon because its socket is full.
13475 * timedated will no longer write or process /etc/timezone,
13476 except on Debian. As we do not support late mounted /usr
13477 anymore /etc/localtime always being a symlink is now safe,
13478 and hence the information in /etc/timezone is not necessary
13481 * logind will now always reserve one VT for a text getty (VT6
13482 by default). Previously if more than 6 X sessions where
13483 started they took up all the VTs with auto-spawned gettys,
13484 so that no text gettys were available anymore.
13486 * udev will now automatically inform the btrfs kernel logic
13487 about btrfs RAID components showing up. This should make
13488 simple hotplug based btrfs RAID assembly work.
13490 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
13491 (but not for its children which will stay at the kernel
13492 default). This should allow setups with a lot more listening
13495 * systemd will now always pass the configured timezone to the
13496 kernel at boot. timedated will do the same when the timezone
13499 * logind's inhibition logic has been updated. By default,
13500 logind will now handle the lid switch, the power and sleep
13501 keys all the time, even in graphical sessions. If DEs want
13502 to handle these events on their own they should take the new
13503 handle-power-key, handle-sleep-key and handle-lid-switch
13504 inhibitors during their runtime. A simple way to achieve
13505 that is to invoke the DE wrapped in an invocation of:
13507 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
13509 * Access to unit operations is now checked via SELinux taking
13510 the unit file label and client process label into account.
13512 * systemd will now notify the administrator in the journal
13513 when he over-mounts a non-empty directory.
13515 * There are new specifiers that are resolved in unit files,
13516 for the hostname (%H), the machine ID (%m) and the boot ID
13519 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
13520 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
13521 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
13522 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
13523 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
13524 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
13525 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
13529 * Support for reading structured kernel messages from
13530 /dev/kmsg has now been added and is enabled by default.
13532 * Support for reading kernel messages from /proc/kmsg has now
13533 been removed. If you want kernel messages in the journal
13534 make sure to run a recent kernel (>= 3.5) that supports
13535 reading structured messages from /dev/kmsg (see
13536 above). /proc/kmsg is now exclusive property of classic
13537 syslog daemons again.
13539 * The libudev API gained the new
13540 udev_device_new_from_device_id() call.
13542 * The logic for file system namespace (ReadOnlyDirectory=,
13543 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
13544 require pivot_root() anymore. This means fewer temporary
13545 directories are created below /tmp for this feature.
13547 * nspawn containers will now see and receive all submounts
13548 made on the host OS below the root file system of the
13551 * Forward Secure Sealing is now supported for Journal files,
13552 which provide cryptographical sealing of journal files so
13553 that attackers cannot alter log history anymore without this
13554 being detectable. Lennart will soon post a blog story about
13555 this explaining it in more detail.
13557 * There are two new service settings RestartPreventExitStatus=
13558 and SuccessExitStatus= which allow configuration of exit
13559 status (exit code or signal) which will be excepted from the
13560 restart logic, resp. consider successful.
13562 * journalctl gained the new --verify switch that can be used
13563 to check the integrity of the structure of journal files and
13564 (if Forward Secure Sealing is enabled) the contents of
13567 * nspawn containers will now be run with /dev/stdin, /dev/fd/
13568 and similar symlinks pre-created. This makes running shells
13569 as container init process a lot more fun.
13571 * The fstab support can now handle PARTUUID= and PARTLABEL=
13574 * A new ConditionHost= condition has been added to match
13575 against the hostname (with globs) and machine ID. This is
13576 useful for clusters where a single OS image is used to
13577 provision a large number of hosts which shall run slightly
13578 different sets of services.
13580 * Services which hit the restart limit will now be placed in a
13583 Contributions from: Bertram Poettering, Dave Reisner, Huang
13584 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
13585 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
13589 * When running in --user mode systemd will now become a
13590 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
13591 tree a lot more organized.
13593 * A new PartOf= unit dependency type has been introduced that
13594 may be used to group services in a natural way.
13596 * "systemctl enable" may now be used to enable instances of
13599 * journalctl now prints error log levels in red, and
13600 warning/notice log levels in bright white. It also supports
13601 filtering by log level now.
13603 * cgtop gained a new -n switch (similar to top), to configure
13604 the maximum number of iterations to run for. It also gained
13605 -b, to run in batch mode (accepting no input).
13607 * The suffix ".service" may now be omitted on most systemctl
13608 command lines involving service unit names.
13610 * There's a new bus call in logind to lock all sessions, as
13611 well as a loginctl verb for it "lock-sessions".
13613 * libsystemd-logind.so gained a new call sd_journal_perror()
13614 that works similar to libc perror() but logs to the journal
13615 and encodes structured information about the error number.
13617 * /etc/crypttab entries now understand the new keyfile-size=
13620 * shutdown(8) now can send a (configurable) wall message when
13621 a shutdown is cancelled.
13623 * The mount propagation mode for the root file system will now
13624 default to "shared", which is useful to make containers work
13625 nicely out-of-the-box so that they receive new mounts from
13626 the host. This can be undone locally by running "mount
13627 --make-rprivate /" if needed.
13629 * The prefdm.service file has been removed. Distributions
13630 should maintain this unit downstream if they intend to keep
13631 it around. However, we recommend writing normal unit files
13632 for display managers instead.
13634 * Since systemd is a crucial part of the OS we will now
13635 default to a number of compiler switches that improve
13636 security (hardening) such as read-only relocations, stack
13637 protection, and suchlike.
13639 * The TimeoutSec= setting for services is now split into
13640 TimeoutStartSec= and TimeoutStopSec= to allow configuration
13641 of individual time outs for the start and the stop phase of
13644 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
13645 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
13646 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
13647 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
13648 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
13649 Gundersen, Zbigniew Jędrzejewski-Szmek
13653 * The journal and id128 C APIs are now fully documented as man
13656 * Extra safety checks have been added when transitioning from
13657 the initial RAM disk to the main system to avoid accidental
13660 * /etc/crypttab entries now understand the new keyfile-offset=
13663 * systemctl -t can now be used to filter by unit load state.
13665 * The journal C API gained the new sd_journal_wait() call to
13666 make writing synchronous journal clients easier.
13668 * journalctl gained the new -D switch to show journals from a
13669 specific directory.
13671 * journalctl now displays a special marker between log
13672 messages of two different boots.
13674 * The journal is now explicitly flushed to /var via a service
13675 systemd-journal-flush.service, rather than implicitly simply
13676 by seeing /var/log/journal to be writable.
13678 * journalctl (and the journal C APIs) can now match for much
13679 more complex expressions, with alternatives and
13682 * When transitioning from the initial RAM disk to the main
13683 system we will now kill all processes in a killing spree to
13684 ensure no processes stay around by accident.
13686 * Three new specifiers may be used in unit files: %u, %h, %s
13687 resolve to the user name, user home directory resp. user
13688 shell. This is useful for running systemd user instances.
13690 * We now automatically rotate journal files if their data
13691 object hash table gets a fill level > 75%. We also size the
13692 hash table based on the configured maximum file size. This
13693 together should lower hash collisions drastically and thus
13694 speed things up a bit.
13696 * journalctl gained the new "--header" switch to introspect
13697 header data of journal files.
13699 * A new setting SystemCallFilters= has been added to services which may
13700 be used to apply deny lists or allow lists to system calls. This is
13701 based on SECCOMP Mode 2 of Linux 3.5.
13703 * nspawn gained a new --link-journal= switch (and quicker: -j)
13704 to link the container journal with the host. This makes it
13705 very easy to centralize log viewing on the host for all
13706 guests while still keeping the journal files separated.
13708 * Many bugfixes and optimizations
13710 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
13711 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
13712 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
13717 * Several tools now understand kernel command line arguments,
13718 which are only read when run in an initial RAM disk. They
13719 usually follow closely their normal counterparts, but are
13722 * There's a new tool to analyze the readahead files that are
13723 automatically generated at boot. Use:
13725 /usr/lib/systemd/systemd-readahead analyze /.readahead
13727 * We now provide an early debug shell on tty9 if this enabled. Use:
13729 systemctl enable debug-shell.service
13731 * All plymouth related units have been moved into the Plymouth
13732 package. Please make sure to upgrade your Plymouth version
13735 * systemd-tmpfiles now supports getting passed the basename of
13736 a configuration file only, in which case it will look for it
13737 in all appropriate directories automatically.
13739 * udevadm info now takes a /dev or /sys path as argument, and
13740 does the right thing. Example:
13742 udevadm info /dev/sda
13743 udevadm info /sys/class/block/sda
13745 * systemctl now prints a warning if a unit is stopped but a
13746 unit that might trigger it continues to run. Example: a
13747 service is stopped but the socket that activates it is left
13750 * "systemctl status" will now mention if the log output was
13751 shortened due to rotation since a service has been started.
13753 * The journal API now exposes functions to determine the
13754 "cutoff" times due to rotation.
13756 * journald now understands SIGUSR1 and SIGUSR2 for triggering
13757 immediately flushing of runtime logs to /var if possible,
13758 resp. for triggering immediate rotation of the journal
13761 * It is now considered an error if a service is attempted to
13762 be stopped that is not loaded.
13764 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
13766 * systemd-analyze now supports Python 3
13768 * tmpfiles now supports cleaning up directories via aging
13769 where the first level dirs are always kept around but
13770 directories beneath it automatically aged. This is enabled
13771 by prefixing the age field with '~'.
13773 * Seat objects now expose CanGraphical, CanTTY properties
13774 which is required to deal with very fast bootups where the
13775 display manager might be running before the graphics drivers
13776 completed initialization.
13778 * Seat objects now expose a State property.
13780 * We now include RPM macros for service enabling/disabling
13781 based on the preset logic. We recommend RPM based
13782 distributions to make use of these macros if possible. This
13783 makes it simpler to reuse RPM spec files across
13786 * We now make sure that the collected systemd unit name is
13787 always valid when services log to the journal via
13790 * There's a new man page kernel-command-line(7) detailing all
13791 command line options we understand.
13793 * The fstab generator may now be disabled at boot by passing
13794 fstab=0 on the kernel command line.
13796 * A new kernel command line option modules-load= is now understood
13797 to load a specific kernel module statically, early at boot.
13799 * Unit names specified on the systemctl command line are now
13800 automatically escaped as needed. Also, if file system or
13801 device paths are specified they are automatically turned
13802 into the appropriate mount or device unit names. Example:
13804 systemctl status /home
13805 systemctl status /dev/sda
13807 * The SysVConsole= configuration option has been removed from
13808 system.conf parsing.
13810 * The SysV search path is no longer exported on the D-Bus
13813 * The Names= option has been removed from unit file parsing.
13815 * There's a new man page bootup(7) detailing the boot process.
13817 * Every unit and every generator we ship with systemd now
13818 comes with full documentation. The self-explanatory boot is
13821 * A couple of services gained "systemd-" prefixes in their
13822 name if they wrap systemd code, rather than only external
13823 code. Among them fsck@.service which is now
13824 systemd-fsck@.service.
13826 * The HaveWatchdog property has been removed from the D-Bus
13829 * systemd.confirm_spawn= on the kernel command line should now
13832 * There's a new man page crypttab(5) which details all options
13833 we actually understand.
13835 * systemd-nspawn gained a new --capability= switch to pass
13836 additional capabilities to the container.
13838 * timedated will now read known NTP implementation unit names
13839 from /usr/lib/systemd/ntp-units.d/*.list,
13840 systemd-timedated-ntp.target has been removed.
13842 * journalctl gained a new switch "-b" that lists log data of
13843 the current boot only.
13845 * The notify socket is in the abstract namespace again, in
13846 order to support daemons which chroot() at start-up.
13848 * There is a new Storage= configuration option for journald
13849 which allows configuration of where log data should go. This
13850 also provides a way to disable journal logging entirely, so
13851 that data collected is only forwarded to the console, the
13852 kernel log buffer or another syslog implementation.
13854 * Many bugfixes and optimizations
13856 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
13857 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
13858 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
13859 Shawn Landden, Tom Gundersen
13863 * "systemctl help <unit>" now shows the man page if one is
13866 * Several new man pages have been added.
13868 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
13869 MaxLevelConsole= can now be specified in
13870 journald.conf. These options allow reducing the amount of
13871 data stored on disk or forwarded by the log level.
13873 * TimerSlackNSec= can now be specified in system.conf for
13874 PID1. This allows system-wide power savings.
13876 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
13877 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
13882 * logind is now capable of (optionally) handling power and
13883 sleep keys as well as the lid switch.
13885 * journalctl now understands the syntax "journalctl
13886 /usr/bin/avahi-daemon" to get all log output of a specific
13889 * CapabilityBoundingSet= in system.conf now also influences
13890 the capability bound set of usermode helpers of the kernel.
13892 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
13893 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
13894 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
13895 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
13899 * Note that we skipped 139 releases here in order to set the
13900 new version to something that is greater than both udev's
13901 and systemd's most recent version number.
13903 * udev: all udev sources are merged into the systemd source tree now.
13904 All future udev development will happen in the systemd tree. It
13905 is still fully supported to use the udev daemon and tools without
13906 systemd running, like in initramfs or other init systems. Building
13907 udev though, will require the *build* of the systemd tree, but
13908 udev can be properly *run* without systemd.
13910 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
13911 should be used to create dead device nodes as workarounds for broken
13914 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
13915 no longer supported. udev_monitor_new_from_netlink() needs to be
13916 used to subscribe to events.
13918 * udev: when udevd is started by systemd, processes which are left
13919 behind by forking them off of udev rules, are unconditionally cleaned
13920 up and killed now after the event handling has finished. Services or
13921 daemons must be started as systemd services. Services can be
13922 pulled-in by udev to get started, but they can no longer be directly
13923 forked by udev rules.
13925 * udev: the daemon binary is called systemd-udevd now and installed
13926 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
13927 to adapt to that, create symlink, or rename the binary after building
13930 * libudev no longer provides these symbols:
13931 udev_monitor_from_socket()
13932 udev_queue_get_failed_list_entry()
13933 udev_get_{dev,sys,run}_path()
13934 The versions number was bumped and symbol versioning introduced.
13936 * systemd-loginctl and systemd-journalctl have been renamed
13937 to loginctl and journalctl to match systemctl.
13939 * The config files: /etc/systemd/systemd-logind.conf and
13940 /etc/systemd/systemd-journald.conf have been renamed to
13941 logind.conf and journald.conf. Package updates should rename
13942 the files to the new names on upgrade.
13944 * For almost all files the license is now LGPL2.1+, changed
13945 from the previous GPL2.0+. Exceptions are some minor stuff
13946 of udev (which will be changed to LGPL2.1 eventually, too),
13947 and the MIT licensed sd-daemon.[ch] library that is suitable
13948 to be used as drop-in files.
13950 * systemd and logind now handle system sleep states, in
13951 particular suspending and hibernating.
13953 * logind now implements a sleep/shutdown/idle inhibiting logic
13954 suitable for a variety of uses. Soonishly Lennart will blog
13955 about this in more detail.
13957 * var-run.mount and var-lock.mount are no longer provided
13958 (which previously bind mounted these directories to their new
13959 places). Distributions which have not converted these
13960 directories to symlinks should consider stealing these files
13961 from git history and add them downstream.
13963 * We introduced the Documentation= field for units and added
13964 this to all our shipped units. This is useful to make it
13965 easier to explore the boot and the purpose of the various
13968 * All smaller setup units (such as
13969 systemd-vconsole-setup.service) now detect properly if they
13970 are run in a container and are skipped when
13971 appropriate. This guarantees an entirely noise-free boot in
13972 Linux container environments such as systemd-nspawn.
13974 * A framework for implementing offline system updates is now
13975 integrated, for details see:
13976 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
13978 * A new service type Type=idle is available now which helps us
13979 avoiding ugly interleaving of getty output and boot status
13982 * There's now a system-wide CapabilityBoundingSet= option to
13983 globally reduce the set of capabilities for the
13984 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
13985 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
13986 even CAP_NET_ADMIN system-wide for secure systems.
13988 * There are now system-wide DefaultLimitXXX= options to
13989 globally change the defaults of the various resource limits
13990 for all units started by PID 1.
13992 * Harald Hoyer's systemd test suite has been integrated into
13993 systemd which allows easy testing of systemd builds in qemu
13994 and nspawn. (This is really awesome! Ask us for details!)
13996 * The fstab parser is now implemented as generator, not inside
13999 * systemctl will now warn you if .mount units generated from
14000 /etc/fstab are out of date due to changes in fstab that
14001 have not been read by systemd yet.
14003 * systemd is now suitable for usage in initrds. Dracut has
14004 already been updated to make use of this. With this in place
14005 initrds get a slight bit faster but primarily are much
14006 easier to introspect and debug since "systemctl status" in
14007 the host system can be used to introspect initrd services,
14008 and the journal from the initrd is kept around too.
14010 * systemd-delta has been added, a tool to explore differences
14011 between user/admin configuration and vendor defaults.
14013 * PrivateTmp= now affects both /tmp and /var/tmp.
14015 * Boot time status messages are now much prettier and feature
14016 proper english language. Booting up systemd has never been
14019 * Read-ahead pack files now include the inode number of all
14020 files to pre-cache. When the inode changes the pre-caching
14021 is not attempted. This should be nicer to deal with updated
14022 packages which might result in changes of read-ahead
14025 * We now temporaritly lower the kernel's read_ahead_kb variable
14026 when collecting read-ahead data to ensure the kernel's
14027 built-in read-ahead does not add noise to our measurements
14028 of necessary blocks to pre-cache.
14030 * There's now RequiresMountsFor= to add automatic dependencies
14031 for all mounts necessary for a specific file system path.
14033 * MountAuto= and SwapAuto= have been removed from
14034 system.conf. Mounting file systems at boot has to take place
14037 * nspawn now learned a new switch --uuid= to set the machine
14038 ID on the command line.
14040 * nspawn now learned the -b switch to automatically search
14041 for an init system.
14043 * vt102 is now the default TERM for serial TTYs, upgraded from
14046 * systemd-logind now works on VT-less systems.
14048 * The build tree has been reorganized. The individual
14049 components now have directories of their own.
14051 * A new condition type ConditionPathIsReadWrite= is now available.
14053 * nspawn learned the new -C switch to create cgroups for the
14054 container in other hierarchies.
14056 * We now have support for hardware watchdogs, configurable in
14059 * The scheduled shutdown logic now has a public API.
14061 * We now mount /tmp as tmpfs by default, but this can be
14062 masked and /etc/fstab can override it.
14064 * Since udisks does not make use of /media anymore we are not
14065 mounting a tmpfs on it anymore.
14067 * journalctl gained a new --local switch to only interleave
14068 locally generated journal files.
14070 * We can now load the IMA policy at boot automatically.
14072 * The GTK tools have been split off into a systemd-ui.
14074 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
14075 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
14076 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
14077 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
14078 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
14079 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
14080 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
14081 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
14082 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
14087 * This is mostly a bugfix release
14089 * Support optional initialization of the machine ID from the
14090 KVM or container configured UUID.
14092 * Support immediate reboots with "systemctl reboot -ff"
14094 * Show /etc/os-release data in systemd-analyze output
14096 * Many bugfixes for the journal, including endianness fixes and
14097 ensuring that disk space enforcement works
14099 * sd-login.h is C++ compatible again
14101 * Extend the /etc/os-release format on request of the Debian
14104 * We now refuse non-UTF8 strings used in various configuration
14105 and unit files. This is done to ensure we do not pass invalid
14106 data over D-Bus or expose it elsewhere.
14108 * Register Mimo USB Screens as suitable for automatic seat
14111 * Read SELinux client context from journal clients in a race
14114 * Reorder configuration file lookup order. /etc now always
14115 overrides /run in order to allow the administrator to always
14116 and unconditionally override vendor-supplied or
14117 automatically generated data.
14119 * The various user visible bits of the journal now have man
14120 pages. We still lack man pages for the journal API calls
14123 * We now ship all man pages in HTML format again in the
14126 Contributions from: Dave Reisner, Dirk Eibach, Frederic
14127 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
14128 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
14133 * This is mostly a bugfix release
14135 * systems lacking /etc/os-release are no longer supported.
14137 * Various functionality updates to libsystemd-login.so
14139 * Track class of PAM logins to distinguish greeters from
14140 normal user logins.
14142 Contributions from: Kay Sievers, Lennart Poettering, Michael
14147 * This is an important bugfix release for v41.
14149 * Building man pages is now optional which should be useful
14150 for those building systemd from git but unwilling to install
14153 * Watchdog support for supervising services is now usable. In
14154 a future release support for hardware watchdogs
14155 (i.e. /dev/watchdog) will be added building on this.
14157 * Service start rate limiting is now configurable and can be
14158 turned off per service. When a start rate limit is hit a
14159 reboot can automatically be triggered.
14161 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
14163 Contributions from: Benjamin Franzke, Bill Nottingham,
14164 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
14165 Schmidt, Michał Górny, Piotr Drąg
14169 * The systemd binary is installed /usr/lib/systemd/systemd now;
14170 An existing /sbin/init symlink needs to be adapted with the
14173 * The code that loads kernel modules has been ported to invoke
14174 libkmod directly, instead of modprobe. This means we do not
14175 support systems with module-init-tools anymore.
14177 * Watchdog support is now already useful, but still not
14180 * A new kernel command line option systemd.setenv= is
14181 understood to set system wide environment variables
14182 dynamically at boot.
14184 * We now limit the set of capabilities of systemd-journald.
14186 * We now set SIGPIPE to ignore by default, since it only is
14187 useful in shell pipelines, and has little use in general
14188 code. This can be disabled with IgnoreSIPIPE=no in unit
14191 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
14192 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
14197 * This is mostly a bugfix release
14199 * We now expose the reason why a service failed in the
14200 "Result" D-Bus property.
14202 * Rudimentary service watchdog support (will be completed over
14203 the next few releases.)
14205 * When systemd forks off in order execute some service we will
14206 now immediately changes its argv[0] to reflect which process
14207 it will execute. This is useful to minimize the time window
14208 with a generic argv[0], which makes bootcharts more useful
14210 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
14211 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
14212 Mike Kazantsev, Ray Strode
14216 * This is mostly a test release, but incorporates many
14219 * New systemd-cgtop tool to show control groups by their
14222 * Linking against libacl for ACLs is optional again. If
14223 disabled, support tracking device access for active logins
14224 goes becomes unavailable, and so does access to the user
14225 journals by the respective users.
14227 * If a group "adm" exists, journal files are automatically
14228 owned by them, thus allow members of this group full access
14229 to the system journal as well as all user journals.
14231 * The journal now stores the SELinux context of the logging
14232 client for all entries.
14234 * Add C++ inclusion guards to all public headers
14236 * New output mode "cat" in the journal to print only text
14237 messages, without any meta data like date or time.
14239 * Include tiny X server wrapper as a temporary stop-gap to
14240 teach XOrg udev display enumeration. This is used by display
14241 managers such as gdm, and will go away as soon as XOrg
14242 learned native udev hotplugging for display devices.
14244 * Add new systemd-cat tool for executing arbitrary programs
14245 with STDERR/STDOUT connected to the journal. Can also act as
14246 BSD logger replacement, and does so by default.
14248 * Optionally store all locally generated coredumps in the
14249 journal along with meta data.
14251 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
14252 writing short strings to files (for usage for /sys), and for
14253 creating symlinks, character and block device nodes.
14255 * New unit file option ControlGroupPersistent= to make cgroups
14256 persistent, following the mechanisms outlined in
14257 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
14259 * Support multiple local RTCs in a sane way
14261 * No longer monopolize IO when replaying readahead data on
14262 rotating disks, since we might starve non-file-system IO to
14263 death, since fanotify() will not see accesses done by blkid,
14266 * Do not show kernel threads in systemd-cgls anymore, unless
14267 requested with new -k switch.
14269 Contributions from: Dan Horák, Kay Sievers, Lennart
14270 Poettering, Michal Schmidt
14274 * This is mostly a test release, but incorporates many
14277 * The git repository moved to:
14278 git://anongit.freedesktop.org/systemd/systemd
14279 ssh://git.freedesktop.org/git/systemd/systemd
14281 * First release with the journal
14282 http://0pointer.de/blog/projects/the-journal.html
14284 * The journal replaces both systemd-kmsg-syslogd and
14285 systemd-stdout-bridge.
14287 * New sd_pid_get_unit() API call in libsystemd-logind
14289 * Many systemadm clean-ups
14291 * Introduce remote-fs-pre.target which is ordered before all
14292 remote mounts and may be used to start services before all
14295 * Added Mageia support
14297 * Add bash completion for systemd-loginctl
14299 * Actively monitor PID file creation for daemons which exit in
14300 the parent process before having finished writing the PID
14301 file in the daemon process. Daemons which do this need to be
14302 fixed (i.e. PID file creation must have finished before the
14303 parent exits), but we now react a bit more gracefully to them.
14305 * Add colourful boot output, mimicking the well-known output
14306 of existing distributions.
14308 * New option PassCredentials= for socket units, for
14309 compatibility with a recent kernel ABI breakage.
14311 * /etc/rc.local is now hooked in via a generator binary, and
14312 thus will no longer act as synchronization point during
14315 * systemctl list-unit-files now supports --root=.
14317 * systemd-tmpfiles now understands two new commands: z, Z for
14318 relabelling files according to the SELinux database. This is
14319 useful to apply SELinux labels to specific files in /sys,
14320 among other things.
14322 * Output of SysV services is now forwarded to both the console
14323 and the journal by default, not only just the console.
14325 * New man pages for all APIs from libsystemd-login.
14327 * The build tree got reorganized and the build system is a
14328 lot more modular allowing embedded setups to specifically
14329 select the components of systemd they are interested in.
14331 * Support for Linux systems lacking the kernel VT subsystem is
14334 * configure's --with-rootdir= got renamed to
14335 --with-rootprefix= to follow the naming used by udev and
14338 * Unless specified otherwise we will now install to /usr instead
14339 of /usr/local by default.
14341 * Processes with '@' in argv[0][0] are now excluded from the
14342 final shut-down killing spree, following the logic explained
14344 https://systemd.io/ROOT_STORAGE_DAEMONS/
14346 * All processes remaining in a service cgroup when we enter
14347 the START or START_PRE states are now killed with
14348 SIGKILL. That means it is no longer possible to spawn
14349 background processes from ExecStart= lines (which was never
14350 supported anyway, and bad style).
14352 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
14353 reloading of units together.
14355 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
14356 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
14357 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
14358 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
14359 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek